Robert Carver discusses systematic trading principles, diversification, and AI in finance
Summary
Robert Carver explains his transition from discretionary to systematic trading and introduces his book 'The Art and Science of Trading'. He argues that while entry signals can be subjective, position sizing and exits must be scientific, and warns that unleveraged trading limits risk optimization. Carver advocates for instrument diversification over strategy diversification and suggests using a 0-20 forecast scale to quantify confidence. He also discusses the role of AI in idea generation versus implementation and promotes open-source code for transparency.
Key takeaways
- 1 Carver states he has been a systematic trader for 20 years, starting at AHL in 2006, and has traded his own money for the last 13 years.
Unidentified speaker (speaker_3) · 2:05
- 2 Good quants come to the data with a hypothesis that they then get the data to confirm for them.
Unidentified speaker (speaker_3) · 11:19
- 3 Carver proposes a 'forecast' scale of 0 to 20 (0=don't buy, 10=average opinion, 20=super bullish) to quantify entry confidence, which aids in managing multiple instruments.
Unidentified speaker (speaker_2) · 13:55
Episode sections
- 0:00 – 3:35
Introduction and Background
Robert Carver introduces himself as an independent systematic trader and visiting lecturer, detailing his transition from discretionary trading to systematic trading at AHL and then trading his own money for 13 years. He explains that systematic trading suits his personality better than the high-pressure bank trading floor environment.
- 3:35 – 7:42
Motivation for 'The Art and Science of Trading'
Carver explains his new book targets a broader audience, aiming to be accessible with minimal jargon. He emphasizes trading Delta One (unleveraged) assets for beginners and argues that while the 'opening method' (entry) can be artistic or subjective, position sizing and exit rules must be scientific and rigid.
- 7:42 – 10:35
Challenges of Unleveraged Trading
Carver argues that trading without leverage is harder than with it because it restricts the ability to stay on the efficient frontier. He illustrates this with a comparison between Bitcoin and 2-year US bonds, showing that unleveraged investors cannot optimally balance risk across assets with vastly different volatilities.
- 10:35 – 12:25
Creativity in Quantitative Trading
Carver discusses the role of 'art' in quant trading, stating that creativity is essential for hypothesis generation and imagining failure modes. He distinguishes good quants, who test hypotheses, from bad quants, who data mine without a prior idea.
- 12:25 – 16:21
The Opening Method and Forecasting
Carver defines the 'opening method' as the entry signal, which can be subjective or systematic. He advocates for using a 'forecast' (a scale of 0-20) rather than binary buy/sell signals to quantify confidence, which helps in managing multiple instruments and trading speeds.
- 16:21 – 22:02
Forecasting Horizon and Trading Costs
The forecasting horizon determines trading speed and transaction costs. Carver warns that mismatching the horizon with the exit method leads to excessive costs or premature exits. He highlights overfitting as a major risk for both systematic and discretionary traders who assume past patterns will repeat.
- 22:02 – 24:12
Volatility Sizing and Risk Targets
Carver explains volatility sizing, where position size is adjusted based on asset volatility relative to a target risk (e.g., 16%). He notes that without leverage, traders may undershoot their risk target on low-volatility assets or be forced to exclude them entirely.
- 24:12 – 26:19
Realistic Expectations and YouTube Influencers
Carver advises traders to set realistic performance expectations by comparing themselves to top hedge funds like Renaissance Technologies' Medallion Fund. He criticizes retail traders for having inflated expectations based on social media influencers.
- 26:19 – 27:37
Adjusting Position Size for Volatility
Speaker_3 discusses adjusting trade size when volatility changes during a trade, noting it improves performance but adds complexity, time constraints, and trading costs. He advocates for simplicity, stating that additional complexity must be justified by practical benefits over costs.
- 27:37 – 29:16
Trailing Stop Loss Construction
Speaker_3 explains using a trailing stop loss that ratchets up with price increases to a high watermark. The stop width is determined by the instrument's volatility and the forecasting horizon, requiring wider stops for more volatile assets like crypto compared to safe bonds.
- 29:16 – 30:44
Forecast-Based Exits
An alternative exit method is closing a trade when the forecast turns zero or negative. Speaker_3 warns this is dangerous for discretionary traders without proven methods, as subjective whims can lead to poor decisions, and recommends preconditions for abandoning stop losses.
- 30:44 – 31:56
Instrument vs. Strategy Diversification
Speaker_3 identifies three axes of diversification: instruments, opening methods, and time horizons. He argues instrument diversification offers more benefit than strategy diversification because there are more instruments across asset classes and countries than there are distinct trading methods for a single instrument.
- 31:56 – 33:47
Pool vs. Portfolio Concepts
The speaker defines a 'pool' as the universe of available assets (e.g., S&P 500) and a 'portfolio' as the subset held at any time. Forecasts are used to select the best assets from the pool to enter the portfolio, especially when capital or position limits exist.
- 33:47 – 54:51
Diversifying Opening Methods
To maximize diversification benefits, one should combine uncorrelated opening methods, such as trend-following and mean-reversion, rather than similar trend methods. Speaker_3 suggests avoiding strategies with correlation above 0.85-0.90 as they add little value.
Assets discussed
| Asset | Kind | Stance | Context |
|---|---|---|---|
| BitcoinBTC | crypto | neutral | Well, let's say you want to trade, say, both um a a two-year two year US bond future and Bitcoin in the same portfolio. |
| NVIDIANVDA | equity | neutral | So that, you know, the counterpart in instrument diversification is, well, if you already trade, say, NVIDIA, there's probably not a huge amount of of point in in trading, say, I don't know, Anthropic as well or Microsoft, because they're, you know, they're very close |
| MicrosoftMSFT | equity | neutral | So that, you know, the counterpart in instrument diversification is, well, if you already trade, say, NVIDIA, there's probably not a huge amount of of point in in trading, say, I don't know, Anthropic as well or Microsoft, because they're, you know, they're very close |
| S&P 500 | index | neutral | Some number like that. I mean, yeah, in the grand scheme of things, that maybe you you're particularly good at this, but that kind of sounds quite low relative to how much the SP five hundred moves a year. So you know whether that's point eight percent or even one percent or one and a half percent, does that really matter or not in terms of being so obsessed with trading costs? |
“Stance” is how the speakers talked about the asset, not a recommendation.
Claims worth verifying
Carver's first trade was in 1999 at the height of the tech bubble.
Unidentified speaker (speaker_3) ·1:21
How to verify: Biographical sources on Robert Carver or his books. For recent events, verify 2026-08-20 through 2026-11-20. Use another period only when the speaker explicitly names it.
Carver worked at AHL from 2006 until 2013.
Unidentified speaker (UNKNOWN) ·2:05
How to verify: Robert Carver's LinkedIn profile or AHL press releases from that period. For recent events, verify 2026-08-20 through 2026-11-20. Use another period only when the speaker explicitly names it.
Carver has been trading his own money for the last 13 years.
Unidentified speaker (speaker_3) ·2:25
How to verify: Robert Carver's website or professional biography. For recent events, verify 2026-08-20 through 2026-11-20. Use another period only when the speaker explicitly names it.
Bitcoin has an annualized volatility of 64% (in the speaker's example).
Unidentified speaker (speaker_3) ·23:01
How to verify: Historical volatility data for Bitcoin from a financial data provider (e.g., Bloomberg, Yahoo Finance) for the relevant period. For recent events, verify 2026-08-20 through 2026-11-20. Use another period only when the speaker explicitly names it.
The S&P 500 (SP) has an annualized standard deviation of returns of approximately 16%.
Unidentified speaker (speaker_3) ·22:22
How to verify: Historical volatility calculations for the S&P 500 index. For recent events, verify 2026-08-20 through 2026-11-20. Use another period only when the speaker explicitly names it.
Renaissance Technologies' Medallion fund is the best hedge fund in the world.
Unidentified speaker (speaker_3) ·25:11
How to verify: Publicly available performance records of hedge funds (e.g., Hedge Fund Research, Bloomberg) to compare returns and Sharpe ratios. For recent events, verify 2026-08-20 through 2026-11-20. Use another period only when the speaker explicitly names it.
Carver is a visiting lecturer at Queen Mary University of London.
Unidentified speaker (speaker_1) ·0:00
How to verify: Queen Mary University of London faculty directory or press releases. For recent events, verify 2026-08-20 through 2026-11-20. Use another period only when the speaker explicitly names it.
Stylus Capital trades digital assets systematically.
Unidentified speaker (speaker_1) ·0:00
How to verify: Stylus Capital website or public filings/statements regarding their trading strategy. For recent events, verify 2026-08-20 through 2026-11-20. Use another period only when the speaker explicitly names it.
Primary sourceCarver worked at AHL creating its fundamental global macro strategy.
Unidentified speaker (speaker_1) ·0:00
How to verify: AHL press releases or Robert Carver's professional biography. For recent events, verify 2026-08-20 through 2026-11-20. Use another period only when the speaker explicitly names it.
Carver managed AHL's multi-billion-dollar fixed income portfolio.
Unidentified speaker (speaker_1) ·0:00
How to verify: AHL press releases or Robert Carver's professional biography. For recent events, verify 2026-08-20 through 2026-11-20. Use another period only when the speaker explicitly names it.
Speaker_3 states that 85% of UK retail traders lose money.
Unidentified speaker (UNKNOWN) ·40:25
How to verify: Check UK financial conduct authority (FCA) or Financial Services Compensation Scheme (FSCS) reports on retail trading losses and CFD/spread bet statistics. For recent events, verify 2026-08-20 through 2026-11-20. Use another period only when the speaker explicitly names it.
Speaker_3 claims his annual trading costs are approximately 0.88%.
Unidentified speaker (UNKNOWN) ·39:59
How to verify: Verify against Robert Carver's published performance records or his book 'Systematic Trading' where he details cost assumptions. For recent events, verify 2026-08-20 through 2026-11-20. Use another period only when the speaker explicitly names it.
Speaker_3 states that futures trading requires at least $100,000 to be accessible to the average person.
Unidentified speaker (speaker_2) ·41:28
How to verify: Check futures exchange margin requirements for major contracts (e.g., CME, ICE) and typical broker leverage requirements. For recent events, verify 2026-08-20 through 2026-11-20. Use another period only when the speaker explicitly names it.
Speaker_3 suggests a heuristic that trading costs should not exceed one-third of expected pre-cost returns.
Unidentified speaker (UNKNOWN) ·43:41
How to verify: Verify in Robert Carver's book 'Systematic Trading' or 'The Art and Science of Trading' under sections discussing transaction costs. For recent events, verify 2026-08-20 through 2026-11-20. Use another period only when the speaker explicitly names it.
Speaker_3 recommends optimizing for geometric return (CAGR) but using Sharpe ratio as the key input metric.
Unidentified speaker (speaker_2) ·44:26
How to verify: Verify in Robert Carver's book 'The Art and Science of Trading' under performance evaluation chapters. For recent events, verify 2026-08-20 through 2026-11-20. Use another period only when the speaker explicitly names it.
Primary sourceSpeaker_3 advises using AI for idea generation and research but stopping its use once code is generated for backtesting.
Unidentified speaker (UNKNOWN) ·49:11
How to verify: Verify in Robert Carver's blog post on AI in quant research or his book chapters on systematic implementation. For recent events, verify 2026-08-20 through 2026-11-20. Use another period only when the speaker explicitly names it.
Speaker_3 mentions that CFDs and spread bets involve implicit leverage and are legal in the UK but not elsewhere.
Unidentified speaker (UNKNOWN) ·40:48
How to verify: Check UK Financial Conduct Authority (FCA) regulations on CFDs and spread betting versus US regulations (e.g., CFTC) which restrict these products. For recent events, verify 2026-08-20 through 2026-11-20. Use another period only when the speaker explicitly names it.
Speaker_3 states that holding periods of a month are typical for his futures trading.
Unidentified speaker (UNKNOWN) ·41:47
How to verify: Verify against Robert Carver's published performance data or books detailing his average holding periods. For recent events, verify 2026-08-20 through 2026-11-20. Use another period only when the speaker explicitly names it.
Speaker_3 claims that Warren Buffett's average holding period is 'forever'.
Unidentified speaker (UNKNOWN) ·41:51
How to verify: Check Berkshire Hathaway annual shareholder letters or SEC filings for average holding periods of equity positions. For recent events, verify 2026-08-20 through 2026-11-20. Use another period only when the speaker explicitly names it.
Primary sourceAI is good at the start of the process to generate ideas but too important to leave to trust with AI during testing and implementation.
Unidentified speaker (speaker_2) ·51:40
How to verify: Primary source: Transcript text at [51:40]. No external verification possible as this is a subjective opinion. For recent events, verify 2026-08-20 through 2026-11-20. Use another period only when the speaker explicitly names it.
Pi System Trade is the open-source version of the speaker's trading strategy, now hosted on an organizational GitHub page and maintained by others.
Unidentified speaker (speaker_2) ·52:19
How to verify: Primary source: Transcript text at [52:19]. Verify by searching GitHub for 'Pi System Trade' or 'Robert Carver' to confirm existence and maintenance status. For recent events, verify 2026-08-20 through 2026-11-20. Use another period only when the speaker explicitly names it.
Open source is one of the greatest inventions in the last 50 years.
Unidentified speaker (speaker_2) ·53:20
How to verify: Primary source: Transcript text at [53:20]. No external verification possible as this is a subjective opinion. For recent events, verify 2026-08-20 through 2026-11-20. Use another period only when the speaker explicitly names it.
The book 'The Art and Science of Trading' will be published on December 1st.
Unidentified speaker (speaker_2) ·53:56
How to verify: Primary source: Transcript text at [53:56]. Verify by checking publisher announcements or retailer listings for the release date of Robert Carver's book titled 'The Art and Science of Trading'. For recent events, verify 2026-08-20 through 2026-11-20. Use another period only when the speaker explicitly names it.
The author's webpage is systematicmoney.org.
Unidentified speaker (speaker_2) ·54:16
How to verify: Primary source: Transcript text at [54:16]. Verify by navigating to systematicmoney.org to confirm it hosts information on Robert Carver's books. For recent events, verify 2026-08-20 through 2026-11-20. Use another period only when the speaker explicitly names it.
Do your own research
- Verify Robert Carver's trading history and AHL tenure through his LinkedIn profile or professional biography.
- Check the publication date and publisher of 'The Art and Science of Trading' on Amazon or the publisher's website.
- Review the 'Pi System Trade' code on GitHub to assess its maintenance status and community activity.
- Compare the performance of unleveraged vs. leveraged portfolios using historical volatility data for assets like Bitcoin and US Treasuries.
- Investigate the definition and historical performance of the 'Medallion Fund' to validate claims about realistic hedge fund returns.
- Search for academic or industry critiques of 'Rob's Razor' or similar simplicity heuristics in quantitative finance.
- Verify the current status of the systematicmoney.org website and its links to book purchases.
- Look for independent reviews of Carver's book to assess its accessibility and accuracy for beginners.
Full transcriptexpand
0:00Unidentified speaker (speaker_1): Now, on to this episode's guest, Robert Carver. Robert is an independent systematic futures trader, investor, writer, and research consultant. He is currently a visiting lecturer at Queen Mary University of London and is a research advisor to Stylus Capital Hedge Fund, which trades digital assets systematically. Robert is also the author of several books, including Systematic Trading and most recently The Art and Science of Trading. Until 2013, Robert worked for AHL, a large systematic hedge fund and part of the Mann Group. He was responsible for the creation of AHL's fundamental global macro strategy and then managed the fund's multi billion dollar fixed income portfolio. Now onto our conversation. Greetings and welcome, Rob. It's fantastic to have you back on the podcast show. I think the last time I had you on was back in the dark days of COVID. So it's been a while. So welcome back.
0:47Unidentified speaker (speaker_2): Really glad to be back. Thanks for inviting me on again. Really appreciate the opportunity. to speak to you guys again.
0:52Unidentified speaker (speaker_1): Now, before we get into your new book called The Art and Science of Trading, just for the benefit of our audience, I know I asked you last time, but that was a while ago. Can you give us kind of a brief sort of background to yourself? In particular, why did you end up going into systematic trading rather than discretionary?
1:10Unidentified speaker (speaker_2): Yeah, so I actually started off as a discretionary trader. Firstly, as a retail discretionary trader. So my first trade was in 1999, which is nearly 30 years ago now. It's depressing. And uh that was at the height
1:21Unidentified speaker (speaker_3): of the what was then called the tech bubble, although I guess we should call it the tech bubble one, perhaps, given what appears to be going on now with AI and NVIDIA and all that kind of stuff. And then I got my first job in sort of finance was working as a what I'd call a quantitative but not systematic trader. So I was trading um exotic. interest rate auctions, which obviously requires a lot of maths, but you know, the actual trading decision whether to buy or sell, I was making purely with, you know, well, it depends on the kind of trade it was as like as investment banks, as a combination of customer flow, hedging, a bit of prop trading. So, you know, very much um just what I would call discretionary trading still. But actually rewinding slightly before that, I'd I done an internship with a systematic uh hedge fund called AHL. And then I when I left uh the banking world
2:05Unidentified speaker (UNKNOWN): did
2:05Unidentified speaker (speaker_3): something after a couple of years and then and then I got a job a full-time job with them so that that was from 2006 onwards really so that's now 20 years used to not quite as depressing but still pretty depressing. For 20 years now I would say I've been mostly a a systematic trader firstly working in industry and then for the last 13 years um purely trading my own money. So yeah a combination of uh happy accidents, but also because I, you know, get I wouldn't have necessarily become a systematic trader, you know, with if I hadn't ended up working out in that industry, which isn't something I sought out to do. It just kind of happened by accident really. But also it suits me much more from a kind of personality perspective. So those two years I spent on a bank trading floor and um, you know, it's no disrespect to the bank. but but it was not for me. It was I did not find it a very a very um pleasant environment to work in. It didn't really play to my strengths in terms of my personalities. So I'm I'm a kind of guy that likes to sit and think quietly about things, which isn't really possible when you know you have lots of people shouting at you and expecting you to make prices and you're constantly able to watch the markets. market. So systematic trading suits me uh as a person much
3:14more as well. No,
3:15Unidentified speaker (speaker_1): that's great. Great sort of background there. And yet you're author of a number of books. And uh the book where I first discovered you was a book called Systematic Trading, which I've you know referred to often in my own work. And also I actually give it to all the new analysts on my desk. It's part of the reading list that they get. So it's it's an excellent book. And uh you've actually come out with a new book called The Art and Science of Trading. How to Combine Intuition and Rules When Trading Globally in Equities, ETFs, and Crypto. It's kind of geared to more for general audience rather than kind of hardcore institutional audience. And so what was the motivation for writing a book like this for a broader audience?
3:55Unidentified speaker (speaker_3): Yeah, I mean, so I've I've also written my books backwards. So systematic trading is I describe as the most advanced book I've written. so far, although I plan to write a more advanced one. This book is very much the entry point into the, you know, the the Robert Carver world, if you like, which is now five books. And as an entry point, yes, it needs to be accessible to a wider audience. That was one thing I tried to do. I tried to write a book that people with a basic understanding of trading and a bit of maths could pick up and read. And so you know minimizing the equations to an absolute minimum really. minimizing the terminology and the jargon and just sort of stripping back and making it as simple as it could be without being dangerous. So this is my version of Occam's razor. Everyone's heard of Occam's razor, that something should be as simple as possible, but no simpler.
4:40Unidentified speaker (UNKNOWN): I
4:40Unidentified speaker (speaker_3): have Rob's Razor, which is as simple as possible without becoming dangerous. So that was the The kind of audience I was aiming for. And then in terms of what an audience like that should be trading, well, all my previous books. pretty much not the second one, but most of the others have been about trading with leverage. So whether that be futures, um, you know, CFDs, spread bets, you know, trading on margin. And I'm not sure that when you start trading, you should be using leverage, to be honest. I think you should be sticking to what in the industry we'd call the Delta One instruments. That's just buying the actual asset without any out any leverage. And it turns out it's actually quite hard to adapt my the framework I use for things like volatility sizing of positions. It's quite hard to adapt that to a world without leverage. It's actually more bizarrely in a way, slightly more complicated to trade without leverage than it is with leverage if you're doing things properly. So I had to sort of rewrite the kind of framework I use to account for that.
5:34Unidentified speaker (UNKNOWN): You
5:34Unidentified speaker (speaker_3): know, that sort of um was an interesting challenge. But yeah, I wanted to. allow people who were less experienced to be able to pick this book up and say, right, let me trade. And the other thing, the other aspect of it is you is in the title, really, the art and science of trading. Well, I think I have spent a long time trying to convince everyone there should be pure systematic traders. And as hard as I try, there will always be a an all, you know, a group of people who don't want to be systematic traders. And that's fine. I've grown to accept that fact. But that doesn't mean they can't read my books. So uh the other thing I tried to do in the book is to say, well, which bits of trading absolutely have to be science, have to be systematic, absolutely have to be involved following rules, and which could you use something else for? And that something else could be, you know, some kind of qualitative analysis, it could be quantitative analysis, is not systematic, it could be you know looking at candlesticks handlesticks and stuff like that, which are not I'm you know, personally I don't think works, but if it works for you, that's fine. It could be using an AI robot to tell you what shares to buy.
6:34Unidentified speaker (UNKNOWN): It
6:34Unidentified speaker (speaker_3): doesn't really matter. The point is that sort of surrounding that core of something should be a set, a framework in the modern parlance, a harness, if you like, of rules that you absolutely have to follow, irrespective of how that key trading training decision is being made. So that was the sort. of multiple sort of sorts of people I was writing the book for, partly because, you know, hey, let's face it, any author wants to sell lots of copies, um, partly because of the money, but but um mainly because you and them want to be really famous and have people coming up to you in the street and saying, hey, you're Rob Carver, which does occasionally happen, but normally at financial conferences like the one we were both at last week, it doesn't tend to happen just randomly on the street.
7:15Unidentified speaker (UNKNOWN): But
7:15Unidentified speaker (speaker_3): in all seriousness, I do want to get the sort of message of how this is how you trade properly to as many people as possible. And so yeah, it's all about making you kind of accessible introductory book for people who want to trade Delta One asset classes, who potentially don't want to be fully systematic necessarily, but you know, to sort of say, at least as a starting point, these are the rules you must follow. And then beyond that, well, it's up to you.
7:37Unidentified speaker (speaker_1): Understood. I just wanted to pick up on that point uh you mentioned about leverage. Why would
7:42Unidentified speaker (UNKNOWN): it
7:42Unidentified speaker (speaker_0): be harder in some ways to come up with a framework around Delta One? Because I mean I need to think through this, but like my first instinct would be isn't just a scalar issue. You know, if if you're lever three times, you just scale everything down to
7:56Unidentified speaker (UNKNOWN): ratio it down.
7:57Unidentified speaker (speaker_2): Well, let's say you want to trade, say, both um a a two-year two year US bond future and Bitcoin in the same portfolio.
8:03Unidentified speaker (speaker_0): Yeah. And
8:04Unidentified speaker (speaker_2): you're working at say a, I don't know, a twenty five percent risk target.
8:08Unidentified speaker (speaker_0): Yeah.
8:08Unidentified speaker (speaker_2): On the Bitcoin, it's fine because your position up your let's say the vol is twenty seventy five percent a year is an arbitrary number. So your correct position side is zero point three three one third. Yeah. So that just means you put if you've got a hundred thousand pound account, out you and put thirty 33,000 three thousand pounds into Bitcoin.
8:23Unidentified speaker (speaker_0): Yeah.
8:24Unidentified speaker (speaker_2): But to trade the two year bond at uh um 25 vol, you'd need to leverage it by approximately 30 times. So your ideal position size will be three million pounds on a hundred thousand, which is fine with leverage.
8:36Unidentified speaker (speaker_3): But without leverage, you the biggest position you can possibly have is a hundred thousand. And that means that you would be undershooting your risk. So that means if you're trading both Bitcoin and the two year bonds, then you'll have much more risk in Bitcoin than in the two year bond. Unless you scale down the average risk of your portfolio to the level of the two year bond, which was going to mean that you're probably going to be,
8:59Unidentified speaker (UNKNOWN): you know,
8:59Unidentified speaker (speaker_3): well below the level of volume you'd optimally have, according to some of the things.
9:03Unidentified speaker (UNKNOWN): Okay,
9:03Unidentified speaker (speaker_0): understood. So
9:04Unidentified speaker (UNKNOWN): in
9:04Unidentified speaker (speaker_0): some ways you're losing some flexibility to
9:09Unidentified speaker (UNKNOWN): the old kind of um
9:10Unidentified speaker (speaker_3): financial economics one on one idea of an efficient frontier.
9:13Unidentified speaker (speaker_0): Yeah. You
9:14Unidentified speaker (UNKNOWN): know,
9:14Unidentified speaker (speaker_3): the whole point of an efficient frontier is if you're an unleveraged investor, you can't go beyond that frontier. And that means that you're limited to accessing only points on that efficient frontier. If you're a leveraged investor, you do, you
9:24Unidentified speaker (UNKNOWN): know,
9:25Unidentified speaker (speaker_3): they do the classic thing of finding the maximum sharp ratio point. And then you can, you know, which in my stylized example, that maximum sharp ratio point would probably be at a very at the far left of that efficient frontoy, like like quite low down. It would have a very low vol and a uh you know uh obviously a commensurally low return you want a higher return you want to go up you're gonna need to um up even up to the close to the Kelly optimal or even partway there you need to apply leverage you can't do that in practice so all you can do is introduce more Bitcoin into your portfolio to get risk and return. Well that's going to reduce the sharp ratio and and mean that your You're portfolio
9:57Unidentified speaker (speaker_2): probably
9:57Unidentified speaker (speaker_0): is a suboptimal. Yeah understood
9:59Unidentified speaker (UNKNOWN): so that
9:59Unidentified speaker (speaker_2): that's the difference. So it's about um you know from a kind of very abstract theoretical way. And believe me, in the book, I don't discuss this at all because it's well above the audience I'm aiming for. But I know that the audience your podcast is probably a bit more sophisticated. I am it's about staying on that efficient frontier, moving along that in terms of your optimization rather than moving up and down a line that you can access with leverage.
10:19Unidentified speaker (speaker_0): Understood, I guess very helpful. And you know, you talk about the art and the science, and I can understand people who use discretion needing. some science, some kind of rigor to their process. But what about the other way round? I mean, do quants need some art as well?
10:35Unidentified speaker (speaker_2): I mean, I guess most quants were like me and um their worst subjects of school were art and music. So and in fact in my school you had to do like one kind of creative G GCSE. C S E. So the G GCSE's C S E examples Samsung you do it at sixteen 16 for international viewers. And um my so I I chose drama
10:52Unidentified speaker (speaker_3): because um, you know,
10:54Unidentified speaker (UNKNOWN): I
10:54Unidentified speaker (speaker_3): am quite a dramatic person and I know I can uh ham it up and I need to, but yeah, no art music worked for me.
11:00Unidentified speaker (UNKNOWN): Although
11:01Unidentified speaker (speaker_3): people always said, well, isn't it true that if you're good at maths, you're good at music, and vice versa? Certainly doesn't work for me. Anyway.
11:07Unidentified speaker (UNKNOWN): Um,
11:08Unidentified speaker (speaker_3): yeah, I I mean
11:09Unidentified speaker (UNKNOWN): I
11:09Unidentified speaker (speaker_3): think creativity is definitely very important in Laquant. a quant. I mean, obviously in the sort of idea generation phase. because you know, bad quants data mine, they let the data tell them
11:18Unidentified speaker (UNKNOWN): what
11:19Unidentified speaker (speaker_3): trade they should be doing, what training strategy they should be using. Good quants come to the data with a hypothesis that they then get the data to confirm for them.
11:27Unidentified speaker (UNKNOWN): Um
11:27Unidentified speaker (speaker_3): so definitely in idea generation. I think you you need to be creative and artistic for sure. But I think it's also about I think good quants have a good imagination, which allows them to attack things in a critical way. So a good quant will think imagine all the terrible things that could go wrong and go to their strategy and like, well, let me just kind of make sure that, you know, whereas a bad quant who's got no imagination will just assume everything's fine. So
11:53Unidentified speaker (UNKNOWN): so
11:54Unidentified speaker (speaker_3): yeah, I think there's definitely good people who are good in this um arena, shall I say, are people who are good at lots of things and very kind of pure data. down the line scientists will would definitely be as bad as people who don't know any maths at all, I'd say.
12:07Unidentified speaker (speaker_0): No understood. And uh in your new book, uh you know, you have a kind of a very organized way of looking at how to construct trades. Um you start off in an earlier part of the book, you talk about the opening method and you have kind of a series of considerations for that. Now, can you explain what you mean by the opening method and the things you have to consider?
12:25Unidentified speaker (speaker_3): Yeah. So I guess what I do is I say there are three basic things that go into trading strategy. Okay. The first thing is that which is what I call the opening method is essentially when should I buy something and where okay. So when is so we're dealing with uh an unleveraged framework here. So we're only ever buying and then closing positions.
12:41Unidentified speaker (speaker_2): We're never selling short. So the opening method says, Well, when should you buy? Okay. The second thing is how big should your position be? Position sizing. And the third thing is when should you close your position? So the art and science part of the book says, well, basically that first thing, the opening method, when you should buy, that's the bit that can be artistic, if you like. You don't, if you come to me and say, well, Rob, I've got this rule where I look at the window, and there are if there are three clouds in the window, I buy, and if there aren't, I don't. That's the way you want to trade.
13:08Unidentified speaker (speaker_3): Fine. That's your opening method. It can be anything you like.
13:12Unidentified speaker (UNKNOWN): If
13:12Unidentified speaker (speaker_3): you come to me and say, well, I'll use a, you know, um, a moving average crossover with these parameters, then I'd personally be more comfortable with that approach. But it's an equally valid thing to do. So in the book, there's a lot of freedom about what the opening method could be. Whereas the sizing and the closing, that's where they have to be scientific and where I believe you have to follow rules. And I set down quite strict rules about what those rules rule should be. So the I the the main thing about the
13:35Unidentified speaker (UNKNOWN): way
13:36Unidentified speaker (speaker_2): I've sort of described opening methods in the book and the the thing that will be unusual for many people, because most people used to think about buying it as a binary decision. So, you know, should we now buy this? Is today a good day to buy X? Okay. Or should we be buying Y instead? But I'm a big fan of um what I of a quantification of that decision into a forecast, what I call a forecast, which is just a scaled number between zero, which means I don't think I should buy this a flat. 10, which means I sort of have an average size opinion about this, and 20, which is I'm super bullish on this. This is great. And there are a number of reasons why that um discipline is helpful, especially when it comes to trading with um multiple opening methods. When it comes to trading with multiple instruments, for example, you really need to have this idea of a scale of a quantification of how big you're, you know, how confident you are that something's going to go up or down.
14:27Unidentified speaker (UNKNOWN): I mean,
14:27Unidentified speaker (speaker_0): when I hear the word forecast, I start to feel a bit nervous around are you kind of introducing some subjectivity into the process? So, I mean, how do you derive the forecasts?
14:38Unidentified speaker (speaker_2): Well, at the end of the day, if your trading method is not scientific, then it must contain some subjectivity.
14:43Unidentified speaker (speaker_0): Yeah.
14:44Unidentified speaker (speaker_2): And that's fine. If that if you are a discretionary trader, then you know, by definition, you are making a subjective decision. Because if it was an objective decision, you'd be a systematic trader because you'd be able to write down exactly what the process is you follow and exactly what you look for when
14:57Unidentified speaker (speaker_3): you're looking for a particular stock. But if any part of that analysis process involves you basically going, yeah, okay, normally my spreadsheet says I should buy it this NAV value produced by this analysis of the cash flows. But on this particular occasion, I'm very nervous about the fact that hypothetical example, the CEO of this company's just been fired for having three affairs, two with subordinates and one with a client. So I'm not going to buy this particular stock. So that's an element of subjectivity. Other people might be okay with that. That's not a subjectivity which any discretionary or part discretionary trading decision is going to have in it. And that also means that any quantification of your feeling of confidence is also going to have an element of subjectivity in it. Because you might say that that this hypothetical large chip company whose name begins it then is currently a sort of a five out of 20 on your scale. And I might sell it's a 15 out of 20, because I still think the financials are strong and I'm discounting the CEO affair thing. And that's completely fine. Now of course if we were pure systematic traders and our opening method was purely systematic, well there shouldn't be any subjectivity in there at all. And the forecast is going to be produced by a mathematical former formula that will produce exactly the same number given a seven certain set of data every single time.
16:08Unidentified speaker (speaker_0): Okay, understood. Okay, so for with the opening method, you kind of have this forecast then. What are the other considerations? So one is to have this forecasting so you have some kind of level of confidence, I guess, that helps you understand sizing. Anything else that we need to consider?
16:21Unidentified speaker (speaker_3): For the opening method. Yeah. So a very important thing to consider is um forecasting horizon. So how long ahead is this forecast being made for? And the reason that's important is because it ties into trading speed. Because if you're forecasting about a month ahead, then that's going to translate into you changing your positions around every month, although maybe slightly slower than that. And there are you know there are ways of checking that. And the trading speed is really important because um you if you're trading too quickly. you'll end up spending too much money on commissions and costs. So costs is a big, big part of the science of trading. And I devote two chapters to talking about costs because that's how important it is. But the other thing is that you need to know how fast you expect to change your forecast because you then need to sort of line that up alongside your method that you're using for closing positions.
17:13Unidentified speaker (UNKNOWN): So
17:14Unidentified speaker (speaker_3): if you've got a very slow forecasting horizon, you don't want to have a method for closing positions that's going to close positions when the price drops very slightly, because that that'll happen very quickly probably. You want something that's going to allow the position to fall further and take more of a loss because that's fine because you're, you know, you're making a judgment the price is going to go up over the next month. What happens the next few minutes or is expected to happen the next few minutes isn't really relevant to you. Converse Conversely, if your trading horizon is very short and you expect to be turning over your positions very quickly because you're forecasting, say, just a couple of days ahead, then you you want to have um, you know, a stop loss or rather method for closing positions that reflects that. You know, you don't want to be um leaving those positions open for months because you're only predicting one two days ahead. So why should it be open for months? That doesn't make any sense. The other thing about that's important about any. kind of opening method, and this it's very obvious how this applies in a systematic world, because we've got the the concept of backtesting where we test training strategies using past data. It's less obvious in the discretionary world, but it's still they're just different sides of the same problem. And that's the problem of overfitting. Uh and overfitting is basically saying, I think the future's going to be very much like the past, too much like the past. Um now no one who's designing any kind of systematic trading strategy and and and any kind of discretionary trader who's looking for sort of
18:33patterns or something they've seen in the past to repeat. Of course you think the future's going to be like the past. You know, it's otherwise, you know, what are we all doing with ourselves essentially? But the important thing about overfitting is that when you is when you take a leap and say, I think the future's going to be exactly like the past or very close to like the past. And therefore when that doesn't happen, you end up with this with a strategy that's not robust.
18:55Unidentified speaker (UNKNOWN): And
18:55Unidentified speaker (speaker_3): as I said, as a systematic trader, it's very easy for me to do that because I can write programs that can test and optimize to their heart's content. But a discretionary trader who does something like look at a chart and say, well, you know, every time this has happened on a Tuesday and needs these to be we've seen, these three patterns on the chart on Tuesday, the last year. every time that's happened, the price has gone up. Okay, in that case, I'm gonna see it again, I'm gonna buy. Well, that that is a effectively a fitting decision. And just because you haven't done that optimization, you know, in a with a piece of program, programming code, doesn't mean you haven't made the same mistake of assuming that that pattern will repeat itself. If anything, it's even worse because you're, you know, you're you're applying subjectivity and you're not applying any kind of you can't apply any sort of outer sample or any of the stuff that you try and do when you're doing a proper back test. So that's the other thing that's pretty dangerous about about uh opening methods. And the other thing that really worries me about a lot of new traders is is they'll they'll come to um trading with an opening method that someone on YouTube has told them is guaranteed to make the money. And um you know I I think uh that's obviously very dangerous and I obviously strongly advise against doing that.
20:03Unidentified speaker (speaker_0): And in terms of um some of the things you've mentioned, on the more professional side, do you find people fall down on some of these things in terms of the opening method?
20:13Unidentified speaker (UNKNOWN): In
20:13Unidentified speaker (speaker_2): terms of overfitting.
20:14Unidentified speaker (speaker_0): In terms of not having a forecast or getting uh the horizons wrong or
20:19Unidentified speaker (UNKNOWN): um
20:20Unidentified speaker (speaker_2): yeah,
20:20Unidentified speaker (speaker_3): so the sort of binary open open-close close decision is, you know, a lot so there
20:25Unidentified speaker (UNKNOWN): are
20:25Unidentified speaker (speaker_3): funds
20:26Unidentified speaker (UNKNOWN): that do that in
20:26Unidentified speaker (speaker_2): my particular
20:27Unidentified speaker (speaker_3): corner of the trading industry, which is the CTA industry. There are funds that will have binary opening and closing rules, but they will tend to trade a lot of rules at the same
20:35Unidentified speaker (speaker_2): time.
20:35Unidentified speaker (UNKNOWN): And
20:36Unidentified speaker (speaker_3): when you kind of stack them on top of each other, then they don't look bind as binary anymore. So
20:40Unidentified speaker (UNKNOWN): I
20:40Unidentified speaker (speaker_3): I'll have rules that make forecasts, and as a result, when they see a trend appearing, they'll gradually buy bind into to that
20:45Unidentified speaker (UNKNOWN): trend. And then
20:46Unidentified speaker (speaker_3): when the trend turns, they'll gradually sell out. You'll see very similar behavior if you have lots of binary rules that that go, oh, the trend started by, trends ended close
20:56Unidentified speaker (UNKNOWN): in
20:56Unidentified speaker (speaker_3): a binary way. But if they're looking at different lengths of time, so you can you can sort of reproduce the the forecast based behavior with the binary based behavior. But that it's a very common it's a more complicated way of doing it. And again, the idea of forecasting is particularly helpful when you don't have leverage a lot of um you know professional funds can use leverage, so it's less of an issue for them.
21:17Unidentified speaker (UNKNOWN): But
21:18Unidentified speaker (speaker_3): yeah, I've certainly seen people trading too quickly in in the professional world as you know, it's not something that's limited to retail traders. You know, so um I've seen very sophisticated people
21:28Unidentified speaker (UNKNOWN): do
21:28Unidentified speaker (speaker_3): analysis of trading costs and then um sort of look at those, the graph of those trading costs and then extrapolate it saying, well, if we double our trading, we'll still only pay this in costs, that's going to be fine.
21:39Unidentified speaker (UNKNOWN): Well,
21:40Unidentified speaker (speaker_3): extrapolation is always a dangerous game. So because you're, you know, you're especially when you talk about trading costs, because you're going to have a greater impact on the market than you thought you were. So yeah, no professionals don't necessarily make these mistakes as often as amateurs, or they make slightly different variations of the mistake, but everyone I think everyone is is at risk of making these mistakes for sure.
21:58Unidentified speaker (UNKNOWN): And
21:58Unidentified speaker (speaker_0): then in terms of sizing the trade, how should one think about that?
22:02Unidentified speaker (UNKNOWN): So
22:02Unidentified speaker (speaker_3): the basic for sizing is as we talked about actually, you know, quite quickly earlier, was the the idea of volatility sizing. So the more volatile the other idea is you come up with some risk target. So you might say, well I want my risk to be 16% a year, annualized standard deviation returns, which is roughly what you get from the SP, although not recently. And then you say, well I'm going to gonna buy, I need to then measure the risk of the thing I'm buying. And if that has the same volatility as the SP, as the 16% that I want, then my position size is 100%. I just put my entire investment into that. Although I do caution in the book about putting all of your money into one thing, that's fine if it's the SP and isn't it, you know, an index ETF, that's perfectly safe. But I would advise against putting 100% of your money even into like a large cop US cap US stock because you know it turns out they can go down. So if that thing then has got more risk, like we described earlier, something like Bitcoin. So if you know it will make the math simple. So if we assume that that Bitcoin's got um analyzed and deviation of say 64%, which is four times your risk target, then your position would be a quarter of that 25%. So if you have, as I said, if you have 100,000 pounds, you put 25,000 pounds into position and the rest would you you would keep as cash. And yeah, the problem arises if you want some to trade something that's much safer than that, because yeah, the theoretical position is
23:21higher than your actual position size, which means you need to use leverage. Uh in this book is banned. So there are leverage ETFs. I'll talk about the pluses and minuses of using those. Those are potentially an option, although they have their downsides as well. But as a rule, I think. you know, the the simplest thing to do is just not to trade things like short volatility bonds that have a you know very low risk level. It's, you know, they can just stay outside of your portfolio. That's the sort of easy way of dealing with the mass of that if you like.
23:51Unidentified speaker (UNKNOWN): Now
23:52Unidentified speaker (speaker_3): the kind of one big unanswered question is what should that risk target be? Where does this 16% come from? And that comes mainly from having a realistic expectation of how well you're likely to do when you're in trading. your trading. And most the although that's a very simple thing to do in terms of the maths, where that expectation comes from, particularly if you're not a systematic trader. So as a systematic trader, you might look at a back test, you might deflate that, but based on the fact that you think there's probably some overfitting in there, something that that's uh risky, and then you end up with a number you can use. As a discretionary trader, you could look at your trading record if you've gone. got a trading record. And I'll talk in the last couple of chapters how to analyze a trading record properly to determine if your performance is systematically significantly good or not, whether just luck. But if you haven't got either of those things, then you know what should your expectations of performance be? And most retail traders have inf expectations of performance that are highly inflated because you know they've watched too many YouTube videos or in. Instagram videos or whatever. And one of the fun things I do in the book is compare a sort of a YouTube influencer who claims they can trade with a certain level of return and then say, well, okay, let's compare them to the best hedge fund in the world, you know, Renaissance technology's Medallion fund. And actually, this YouTube trader is a is an infinitely better trader than the Renaissance technology, which seems
25:11unlikely. So I suggest one should be critical of. the claims made by these people. So having kind of made people think about what realistic expectations look like, I then suggest what is probably quite a relatively low risk target that be surprised many people would be surprised to see. But um in practice, unless you're trading crypto or small cap stocks, it's probably not really a limiting factor because you are, you know, you're never going to get to a very high risk target without leverage anyway. So the lack of leverage is kind of a comfort blanket, really, at this point. level of trading, which I think is good.
25:40Unidentified speaker (speaker_0): One issue when you are sort of ball targeting is that if volatility spikes for whatever reason, then you have to scale down your position size. Is that how it's how
25:50Unidentified speaker (UNKNOWN): do you
25:50Unidentified speaker (speaker_0): think about that?
25:51Unidentified speaker (speaker_2): Yeah. So in this board car, I do what I've done in um a couple of my previous books, actually leverage trading and advanced futures Which is trading strategies, which is I I sort of start with a very simple strategy and then I gradually introduce complexities to it.
26:02Unidentified speaker (UNKNOWN): So
26:03Unidentified speaker (speaker_2): the initial very simple strategy, once the trade is open, there is no further position sizing. Okay, so volatility spikes, as long as the that doesn't affect the, you know, if you're using a stop loss to close the trade, which is what I said again, what I suggest for beginners.
26:17Unidentified speaker (UNKNOWN): If
26:18Unidentified speaker (speaker_3): that doesn't hit the stop loss in
26:19Unidentified speaker (UNKNOWN): that
26:19Unidentified speaker (speaker_2): volt
26:20Unidentified speaker (speaker_3): spike,
26:21Unidentified speaker (speaker_2): so
26:21Unidentified speaker (speaker_3): maybe there's the price spikes
26:22Unidentified speaker (speaker_2): up,
26:23Unidentified speaker (UNKNOWN): so
26:23Unidentified speaker (speaker_2): no problem there. We like that. Or maybe it does spike
26:26Unidentified speaker (speaker_3): down, but not enough to hit the stop loss. The trade would stay open in the same size. That's not a problem.
26:31Unidentified speaker (speaker_2): And then in
26:31Unidentified speaker (UNKNOWN): later
26:31Unidentified speaker (speaker_3): chapters, I do introduce and say, well, actually, you can adjust your trade size if volatility changes during the size of the trade. And that will actually lead to an improvement in performance. However, it it also it increases complexity,
26:43Unidentified speaker (UNKNOWN): it
26:44Unidentified speaker (speaker_3): increases time, the time constraint, because
26:47Unidentified speaker (UNKNOWN): you
26:47Unidentified speaker (speaker_3): know, I trade fully automatically. So you know, it doesn't cost me any time to do anything really. But obviously, if you're trading manually, then having to worry about changing position size every day is another element of time you've got to consider. But also importantly, introduces trading additional trading costs. So I talk about how you could reduce the effect of those and how you can sort of judge whether actually it's worth adding this,
27:09Unidentified speaker (UNKNOWN): you know.
27:10Unidentified speaker (speaker_3): So my approach is as simple
27:12Unidentified speaker (speaker_2): as possible, as I've already said, any additional complexity. has
27:15Unidentified speaker (speaker_3): to be justified. It shouldn't be something you just do because, yes, theoretically you're actually
27:19Unidentified speaker (speaker_2): right. Position
27:20Unidentified speaker (speaker_3): changing position size when involved spikes is something you should
27:23Unidentified speaker (speaker_2): do.
27:24Unidentified speaker (speaker_3): But in practical terms, when you think about costs and time, it may not necessarily make sense for everybody.
27:29Unidentified speaker (speaker_0): Understood. And you mentioned stops there. And I guess more generally, how do you know when to exit a trade? Or I guess how to construct the stop as well.
27:37Unidentified speaker (speaker_2): Yeah, so I advocate two ways. to exit
27:39Unidentified speaker (UNKNOWN): a
27:39Unidentified speaker (speaker_2): trade. So the the simplest way, which
27:41Unidentified speaker (speaker_3): is
27:41Unidentified speaker (speaker_2): what as I said,
27:42Unidentified speaker (speaker_3): you should use initially, and the first part of the book is used, is a trailing stop loss.
27:47Unidentified speaker (UNKNOWN): So,
27:47Unidentified speaker (speaker_3): you know, there's the price
27:48Unidentified speaker (UNKNOWN): you
27:49Unidentified speaker (speaker_2): buy,
27:49Unidentified speaker (speaker_3): the price hopefully goes
27:50Unidentified speaker (UNKNOWN): up. You'd
27:51Unidentified speaker (speaker_3): start with a stop, say,
27:53Unidentified speaker (UNKNOWN): I don't know,
27:53Unidentified speaker (speaker_3): a random figure, $10 below the entry price. If the price goes up $20, then the stop would also go up because it goes ratchet up, ratchets up as the price goes up.
28:02Unidentified speaker (UNKNOWN): to
28:02Unidentified speaker (speaker_3): what you might we call a high watermark level.
28:05Unidentified speaker (UNKNOWN): If
28:05Unidentified speaker (speaker_3): the price slips down from that level, the stock stays the same, the stop doesn't move. So if the price subsequently does drop $10, then you'll close the trade and you'll hopefully have taken some profit out of that trade along the way. So the key issue then, of course, is how wide that stop should be. And it's a function of two things. Firstly, the again, how volatile the instrument is. So something that's got the risk of a, you know, a safe bond ETF, the stock. stop's going to be look quite tight because they don't move around very much.
28:31Unidentified speaker (UNKNOWN): You know,
28:31Unidentified speaker (speaker_3): a cryptocurrency will have a much wider stop in price terms because they're much more volatile.
28:36Unidentified speaker (UNKNOWN): And
28:36Unidentified speaker (speaker_3): then the other thing that goes into that, and this is the thing that most people miss actually, is the again the forecasting horizon. So essentially you can calibrate the size of your stop by taking the volatility of the instrument and multiplier multiply by a number. And the bigger that number is, the longer you're likely to hold a trade for. And therefore, that means that, you know, if you've got a two-day holding horizon, that'll be quite a tight stop, quite a small number you're multiplying the volatility by. If you expect to hold for a number of months, it'll be much wider. So that's the kind of the construction of the trailing stock. And then in terms of the other way up to close trades, and this is the way I actually I use it in my own trade trading system, but again, it's more work, it's potentially more costs.
29:19Unidentified speaker (UNKNOWN): So
29:19Unidentified speaker (speaker_3): both of those factors should be taken into consideration when deciding whether to do this or not, is and is to actually use the forecast itself to tell you when to close the trade. So a positive forecast means you want to buy something. Zero or negative means you don't. Therefore, you if the forecast was positive and goes zero or negative, then at that point you close the trade with that. Now that this is clearly an issue if you are a discretionary trader, a subjective trader, because your forecast is a subjective number you've made up. And if you just don't like something and you can't say why,
29:52Unidentified speaker (UNKNOWN): I
29:52Unidentified speaker (speaker_3): think it's potentially quite dangerous if you on a whim can then close your trade. So I in the book I talk about you know the sort of preconditions before you should abandon a stop loss. And uh one of those is if that you are a discretionary trader, then you need to have sort of proven yourself. And I explain how to do that in the last couple of chapters. So again, it's about you know, the the method message of the book is you know, here's a simple way of doing things. You make it more complicated if you will, but these are the downsides and this is what you need to think about if you do that.
30:21Unidentified speaker (speaker_0): Okay. And is that what you mean by forecasting stops as well? Yes,
30:24Unidentified speaker (UNKNOWN): a forecast based stops. stop Yeah, because that's what you talk about.
30:27Unidentified speaker (speaker_0): in your book. Um and then in terms of diversification, you kind of touched on it a number of times in different ways. I mean, what's your kind of thinking around that all in terms of presumably you have a a choice of instruments that you a range of instruments, you could have your you have your target vol. What are the considerations here?
30:44Unidentified speaker (UNKNOWN): So
30:44Unidentified speaker (speaker_2): there's two well three you know, th we always talk about three main axes of diversification. So the two main ones are, as you say, instrument or trading strategy, opening opening method in my parlance. So you can diversify by either having trading more instruments or you can diversify by having more opening methods and then combining them together to make to make a something that should have better forecasting properties. And both are good, uh, but actually you get more bang for your buck if you do instrument diversification. It gives you a much more diversification because there's
31:16Unidentified speaker (speaker_3): only so many ways to trade a given instrument. Whereas there are lots of instruments out there, particularly if you're uh well, I'm in the future space, but also in the ETF space as well. You can trade, you know, different asset classes, different countries very easily. Whereas if you're trading just stocks or worse, just crypto, then the there's you know, within a single, say just US stocks, there's obviously less diversification available there. Now, the downside of diversification by instruments is that again, you know, in a world where you can use leverage, it's a bit simpler because you can just keep buying things until you run out of margin in the broad sense. But um in a world when you can't use leverage, it's a bit more complicated. And um so in the book I introduced the concept of a a pool of instruments and a portfolio of instruments. So a pool is um is a a unival universe, or we also call a universe in quant space, is a universe of a set of stocks that you can choose or other assets that you can choose from.
32:08Unidentified speaker (UNKNOWN): So
32:09Unidentified speaker (speaker_3): that could be the whole SP 500, it could be a bunch of ETS you pre-selected or a bunch of crypto coins you happen to like. And the portfolio is the number of those that you hold at any one time.
32:19Unidentified speaker (UNKNOWN): So
32:19Unidentified speaker (speaker_3): the most basic strategy starts up with a with the where the pool is one, the portfolio is one. So you can you trade one thing and you're either holding it or you're not holding it. So you're either in cash, you're not in cash. And then I say, well, you could trade one thing, but drawn from a larger pool. So maybe there are, you know, 10 US stocks you really like, but you're only ever going to trade one of those at a time. And then that's when the forecasting rule becomes useful because you might be bullish on all of those stocks. their You think they're all going to go up. Which one do you buy? Well, you buy the one with the highest forecast. So that's where the discipline. of quantifying your expectations in the form of a forecast becomes useful. And then moving to a portfolio, a portfolio would be where maybe you had a hundred stocks you could buy in your pool and you choose to own, say, 20 or 30 of those at any given time. And again, the forecast becomes useful because you know, if you you're gonna you end up closing one of those positions, well, who comes from the pool into the portfolio to report? replace it? Well, again, that'll be the stock with the highest positive forecast. So that's where the the forecast becomes useful. Because the the forecast is telling you, well, I I think this thing will go up more than this other thing. So that's what we should own, right? If we're limited on what we can buy. And in terms of um sort of quantifying how useful these things are or how much money they will make you, the big the portfolio is where the benefit really comes. So big portfolio is holding lots of things.
33:39at the same time, is where you really do get a big diversification benefit, particularly as I said, if you can trade across lots of assets and countries, you're not limited to a single one.
33:47Unidentified speaker (speaker_0): Okay, understood. And in terms of the diversification of opening methods, would that also incorporate diversification of time horizons as well?
33:55Unidentified speaker (speaker_3): Yeah, exactly that. Yeah. So I'd like to sort of say, well,
33:58Unidentified speaker (UNKNOWN): if
33:59Unidentified speaker (speaker_3): you're going to do any diversification, you should try and get, you know, the most bang for your buck again to use that rather trite expression. So that, you know, the counterpart in instrument diversification is, well, if you already trade, say, NVIDIA, there's probably not a huge amount of of point in in trading, say, I don't know, Anthropic as well or Microsoft, because they're, you know, they're very close gonna clearly be close related. They're essentially customers of each other.
34:22Unidentified speaker (UNKNOWN): The
34:23Unidentified speaker (speaker_3): counterpart to that in um in opening method space is well you know, to get the most bang for your butt. Let's say you've got um like a trend-based method. Well, adding a uh a mean reversion, but more of a mean reversion style method to that will give you more sort of thing like say a carry strategy, if anyone's familiar with those, will give you much more diversification. Slightly less diversification would be if you've got a trend method and you add a slightly different trend method. So maybe you've got a moving average method and you add a breakdown method. And then the the least good diversification is probably adding things that trade at different time horizons, but are basically the same thing. So they're likely to be
35:01Unidentified speaker (UNKNOWN): to
35:01Unidentified speaker (speaker_3): be more correlated. And you also run into the the issue that very um things that the certain stars are trading will tend to work back at different frequencies. So trend falling following is sort of a medium speed kind of thing. It works quite well if you're for trade trade holding holding method. methods of a around a month and a few months outside of that mean reversion tends to come more into play. So um you don't necessarily want to trade every possible permutation of a trend method because they're not all necessarily going to be going to be profitable. And but the yeah there's no point if you're trading two things that have a correlation of point nine nine nine well you know as someone who's got a fully automated trading system I even I wouldn't do that even though it's not really costing me anything apart from a few CPU cycles, but and some database storage. But as someone who's got to manually make those forecasts, well, that's really silly. So you want to you wanna kind of have things that have a correlation of maybe 0.8 or 0.85 at the most. And anything that's more correlated than that, it's not really adding much to your portfolio. So it's probably worth They're leaving out. But that's not to say that, you know, diversification through time isn't important. Of course it is. Um, it's just that it's not as important as the other ones.
36:05Unidentified speaker (speaker_0): Understood. And in your book, you also talk about continuous trading. I mean, what do you mean by that? Is that like always on type trading? Or what do
36:13Unidentified speaker (UNKNOWN): you mean by
36:13Unidentified speaker (speaker_0): continuous trading?
36:14Unidentified speaker (speaker_2): Yeah, so
36:15Unidentified speaker (speaker_3): continuous trading is where you are always you don't really have so people have a concept of a trade. So I did a trade today, I bought some stuff,
36:24Unidentified speaker (UNKNOWN): and
36:24Unidentified speaker (speaker_3): then in a few weeks that I'll were that, be like, oh, I closed my trade. I sold my things. It's very much a binary way of things thinking about. And this is the very very simple system in the book, as I've already said. The opening method says buy, I buy a certain amount, I wait till the stop loss hits and then I close it. And I don't make any adjustments to that in between. Continuous trading is about saying what is the best thing I should, shouldn't, what's the best, what's the current the best portfolio position I should have on right now, today, given the information. I've got? And that would be, as we've already alluded to, the the volatility of the instruments you're trading, the current level of your forecasts, the price outside the scope of this book, but things like correlation, for example, would be potentially in there as well. Based on that, that I can come up with a The set of positions that I want to hold my optimal set of positions. And then I compare that to the positions I actually have on. And the difference between those two is the trade I should do. So it's very much about thinking about optimal positions rather than about optimal trades. And if you said to me, What trades have you done recently? Well, one answer to that is, oh, well, I bought and sold this yesterday, but because those weren't me opening or closing an entire new position, probably some of them would have been, they're not really what you might think of as a trade, you know, because I'm not going to hold. I didn't sell something
37:40yesterday. I'm going to buy back in three weeks' time because something's happened. It's just that yesterday my optimal position was this. So I sold a few weeks' time, the optimal position will have changed. So I'll buy it, buy it back. It's just a different way of thinking about things. And that's that's sort of the last substantive chapter in the book, really, because that that is, especially if you're uh a discretionary trade or if you're not uh fully automated like I am, then operating that way. is quite intensive. So it's not for everybody. But yeah, theoretically it's a better way of doing things because you're continuously taking all the current information you have about volatility and forecasts in into ha in hand when you're thinking about what your position should be, rather than your position being something happened a few weeks ago and you you know, it's no longer true today. So you're closing it.
38:23Unidentified speaker (speaker_0): Okay. And in terms of performance analysis, like when you kind of review your performance, are you any good or not? How does one do that without just getting caught up with watching your PL every day?
38:35Unidentified speaker (speaker_2): That's another story, isn't it? Watching your PL every day for sure. Yeah. So I think um the mistake that most people make when doing any kind of performance analysis, particularly retail traders, is they misunderestimate the relative amount of sort of noise and signal. So your daily returns that you're, you know, you're so focused on are going to be very much mostly noise. And it's not until you've got a decent history of returns that you can actually do some analysis
38:59Unidentified speaker (speaker_3): and say, well, actually, there's a fairly good percentage that I am genuinely a profitable trader. And it's not just every day I have to be lucky at throwing this particular dice essentially. So the key is to keep good records. So for me, it's easy because every trade I do goes into a database. For a manual trader, that means probably building a spreadsheet. And I describe in the book in quite tedious detail exactly what that spreadsheet should look like, even what the column name should be. And it may seem a bit like overkill, but actually to do a proper analysis of your trading. So for example, to disaggregate the effect of how good you are as a trader from your trading costs is really important because you need to make sure that the assumptions you've made. about trading costs were correct. And if not, you need to revise those assumptions and take appropriate action. So that involves capturing a bit more information than just the price you bought or sold at. You've got to add some extra stuff in there. I
39:50Unidentified speaker (UNKNOWN): mean,
39:50Unidentified speaker (speaker_0): just on trading costs, actually, we didn't talk much about trading costs, but like I think in your book you talk about your, I think your trading costs are 0.88% eight to eight percent a year or or
39:59Unidentified speaker (UNKNOWN): sounds about
39:59Unidentified speaker (speaker_2): right.
39:59Unidentified speaker (speaker_0): Some number like that. I mean, yeah, in the grand scheme of things, that maybe you you're particularly good at this, but that kind of sounds quite low relative to how much the SP five hundred moves a year. So you know whether that's point eight percent or even one percent or one and a half percent, does that really matter or not in terms of being so obsessed with trading costs?
40:16Unidentified speaker (speaker_2): Well most retail traders okay, so there's a statistic in here in the UK that eighty 85% five percent of traders roughly retail traders roughly speaking lose money.
40:25Unidentified speaker (UNKNOWN): Yeah.
40:26Unidentified speaker (speaker_2): Now I don't think that's because they've got rubbish opening methods. Okay. On average, let's say, let's be pessimistic and say that on average, their opening method is flipping a coin. Okay. And actually, it's likely it's better than that because most of them are trading the stock market. Stocks go up on in over the long term, and therefore, even a coin flip should have a better than a 50% success rate.
40:48Unidentified speaker (UNKNOWN): So
40:48Unidentified speaker (speaker_2): why isn't it 50% of traders losing money or 45% or whatever you think it should be? Well, it's because of a lot of them is because they're using too much leverage. So if they're trading, if you look at people trading like the CFDs and spread pets, which are legal in the UK but not elsewhere, that involves implicit leverage or options for that matter. That involves leverage and that makes it more likely you're going to blow up, lose money that way. And the other reason is because they trade too much. They're incurring too much in trading costs. So yes, it'd be it really does matter. Now my trading costs are very low, partly well, three there's three reasons why. One is in trading futures, which are naturally quite a cheap thing to trade, but they're not, you know, well A, they're a leveraged instrument, so they're not out of scope for this book, but B also because of the amount of money you need to trade futures, they're not really accessible Successful to the average person. I say you need at least $100,000 to even think about trading futures, for example. The second reason is my holding period is on average is say a month, which is, you know, would see, you know, I'm not day trading here. You
41:47Unidentified speaker (UNKNOWN): sound like Warren
41:48Unidentified speaker (speaker_0): Buffett right now.
41:49Unidentified speaker (UNKNOWN): Oh,
41:49Unidentified speaker (speaker_2): well, no, his average pay holding period
41:51Unidentified speaker (UNKNOWN): is
41:51Unidentified speaker (speaker_2): forever, right? That's what he's
41:52Unidentified speaker (UNKNOWN): saying. I know, I'm
41:52Unidentified speaker (speaker_0): just uh I'm just
41:53Unidentified speaker (UNKNOWN): comparing it to
41:53Unidentified speaker (speaker_0): retail guys
41:55Unidentified speaker (UNKNOWN): around.
41:56Unidentified speaker (speaker_2): I sound like Warren Buffett. I'd probably look like him as well to a you know millennial or Gen Z. The other reason is you say are you good at this? Well, I'm good at it in the sense that I analyze trading costs very carefully and then use them in a very obsessive obsessive. I use them in my optimization, my calibration, my trading strategy. So I deliberately avoid, for example, trading things that are too expensive. I deliberately trade more slowly things that are too expensive. So yeah, I do quite a lot of work. And the reason I do that is well, let's say if I didn't do any of this work, say my trading costs would be four percent a year. So or let's make the numbers round three point so they make them three say there'll be three point eight eight percent a year. So that's three percent a year higher in trading costs. Well if you said to um somebody I can guarantee you you'll earn three percent a year extra every year. Guarantee I guarantee it for you. Like you this definitely gonna happen. Who wouldn't take that bet, right?
42:47Unidentified speaker (speaker_0): Absolutely. Yeah. So
42:48Unidentified speaker (speaker_2): I'd be giving up that three 3% percent of certain like you know, non negative return, if you want to put it like that, in exchange for perhaps earning more through some potential higher alpha, which is much more uncertain. So that's one of the things I say in the book. And I've said a lot before costs are certain, pre cost returns are not. Therefore, given the, you know, given the choice between two marginal options, one of which has slightly look looks slightly better. after cost, but the costs are much higher, go for the other one because those much higher costs definitely going to happen. The extra turn may not happen.
43:23Unidentified speaker (UNKNOWN): But
43:23Unidentified speaker (speaker_2): yeah, point, I mean, it's unrealistic for someone trading stocks with a holding period of a few days to pay those kinds of costs. They'll be paying more. But the point that, you know, I have a heuristic in the book, a rule of thumb, which is to say if you're spending more than a third of your expected pre-cost return on costs, that's too much. That's way too much.
43:40Unidentified speaker (speaker_0): Yeah.
43:41Unidentified speaker (UNKNOWN): For me that fraction. So I
43:42Unidentified speaker (speaker_0): understand the importance. Uh
43:43Unidentified speaker (UNKNOWN): I'm
43:44Unidentified speaker (speaker_0): reading a bit so provocative.
43:46Unidentified speaker (UNKNOWN): For me, that
43:46Unidentified speaker (speaker_2): fraction is one in 20. Yeah. Um, so I'm well below that level. And if I was at one in three, I would be worried. But so one in three is literally like you must not go more than this. I I call it the speed limit in my first book. Um, it's literally like do not go beyond this level because this is the absolute level, and ideally you should be well below that.
44:05Unidentified speaker (speaker_0): Yeah. And so in terms of performance analysis, then uh you know, keeping very good records, including costs and sort of breaking down all the sorts of source of returns is really important. But in terms of evaluation, I mean, should one look at sharp ratios, your drawdowns, should you have the benchmark that you refer, you know, back to? What's kind of the good hygiene around this, so to speak?
44:26Unidentified speaker (speaker_2): Yeah. So in the book I say you shouldn't actually optimize for sharp ratio. You should optimize for something called geometric return or C A G R. But the key input to that is what your sharp ratio is or what you think it is. And the advantage of sharp ratio is obviously it doesn't matter what you're trading, whether you, you know, what the risk on it is. And obviously for leverage traders, sharp ratio is particularly important because sharp ratio is invariant to leverage. But sharp ratio is the metric I think you should use when evaluating performance. I talk in the book about how to calculate sharp ratio from, you know, different sources. So you can get it from your history of daily returns, or you can get it from your history of trades that you've done, or you can get it also by more complicated, but in some ways the best way, which is to look at your forecasting history and see how well your forecasts have turned out as well. And then given that you can get a sharp ratio. take costs out of that. So extract costs. So you're looking at your pre-cost sharp ratio, analyze your costs separately because that's a separate job. Look at your pre-cost sharp ratio and then do your analysis on that. You know, we could spend hours talking about the disadvantages of sharp ratios, but you know, I think there I wouldn't use, you know, there are disadvantages to all performance statistics, none are perfect. And sharp chart ratio for me is the is very intuitive. and relatively simple to calculate. So I always tend to come back to chart ratio.
45:44Unidentified speaker (speaker_0): Okay, yep, that makes sense. Now I did want to kind of go a bit away from the book as well. I mean, obviously AI has been a huge uh theme, you know, both in terms of uh bubble talk and markets, but also in terms of techniques people can now use. I mean, this is not just for retail or general audience, but maybe also for professional investors. How do you think people should be using AI? I mean, everyone's talking about agentic AI and deploying swarms of agents to do the work for you. And there's so many possibilities now. And so
46:12Unidentified speaker (UNKNOWN): how how
46:12Unidentified speaker (speaker_0): do you think about this?
46:14Unidentified speaker (speaker_2): You really are reading in my mind today because um I'm actually literally just before this podcast, I'm actually writing a blog post on the best way of using AI in A quant research.
46:22Unidentified speaker (UNKNOWN): Okay,
46:22Unidentified speaker (speaker_0): Well we've got
46:23Unidentified speaker (UNKNOWN): a we've got
46:23Unidentified speaker (speaker_0): a little preview here. Nice.
46:24Unidentified speaker (speaker_2): Yeah exactly. So the short answer is go by the time this podcast is out, the blog code post will definitely be out. So go and look at my blog and you can read it in full. But the short answer is that I think there's an awful lot of bad use of AI going on out there. There are things that that really worry me and scare me. I guess it's like any technology. If you know what you're doing, it can be really useful. If you don't, it can To be really dangerous. And I mean dangerous in terms of here, I I'm mean in terms of financial costs, cost, basically, because you know. we talked about overfitting. Well, AI is the best overfitting tool ever created, as far as I'm concerned. So I would say if you think about the typical workflow of someone creating a and running a quantitative training strategy, which is you know why I do every day, I would say that the way I would use AI is things it's good for, idea generation, research, whether that be summarizing academic research papers, finding data source. sources, you know, I want to trade this particular idea. Well, I need this data source. Where's a place I can get that? Okay, where's a place I can get that for free? Ideally, at least to begin with, while I'm just looking at it. But using them in those with a very critical view. So for example, take the you know, analyze this academic research for me. I'd still go and look at the original papers. So the value of AI for me would be actually telling me about papers. I didn't necessarily know about. So I'd I'd then go and look at those because they may, I mean, it could be a pure
47:44elucidation, you know, worst case scenario, or it could have been summarizing the paper wrong actually, the paper's not really relevant. But to me, it's a step up from, you know, like a Google Scholar search where you'd have, you know, 500 papers, best case scenarios, you can sort them by citation, which is an imperfect measure of paper quality or relevance. at the best of times.
48:04Unidentified speaker (UNKNOWN): I think actually on
48:05Unidentified speaker (speaker_0): that, I actually have a specific prompt I use to summarize papers where it's I've got certain sort of prompts to summarize it well. But and then on top of that, if the paper is citing a trading strategy, I don't have a checklist that it has to pass. Um I ask like there's 15 points. So one would be did it report T cost or transaction cost, did it show out-of-samped. out of sample testing? Did it do this? Did it do that? And so on. So it's checklist. And then it gives you
48:30Unidentified speaker (UNKNOWN): a
48:30Unidentified speaker (speaker_0): score of like it sort of ticked 10 out of the 15 things. And then that also gives me a sense of how serious that paper was around uh doing the back test. So
48:38Unidentified speaker (UNKNOWN): I find
48:39Unidentified speaker (speaker_0): that to be quite helpful.
48:40Unidentified speaker (speaker_2): Yeah. That's almost like another filter, isn't it? So
48:42Unidentified speaker (speaker_0): yeah, it's an additional filter to
48:44Unidentified speaker (UNKNOWN): kind of just get through the noise. Yeah. The
48:45Unidentified speaker (speaker_2): first prompt might give you fifty 50 papers, but actually yeah, 25 twenty five of them are rubbish and you shouldn't even bother. reading.
48:51Unidentified speaker (UNKNOWN): Yeah.
48:52Unidentified speaker (speaker_2): That's nice. Yeah. So now you're at the point where you have an idea of, you know. So at that point, you can either yourself or get the I to write you some I use Python to write a Python function that will basically implement that trading strategy, given some data. But very importantly, at this point, you stop using AI.
49:11Unidentified speaker (UNKNOWN): You
49:12Unidentified speaker (speaker_2): put your AI away. Don't look at it, you're not going to use it again. You take what you do then is you take that strategy, you parameterize it, which means you include parameters in it so it can be fitted. So very simple, like let's assume and I'm, you know, the AI assists to you, well, you know, you can use, say, you say, or you start saying, well, what ideas can I use? Oh, well, trend following works. Oh, trend following that's interesting. Summarize the reason research for trend of following falling on me, all these papers come back. You see a few of them mention a thing called a moving average crossover. Okay. So you say, Well, write me a Python function to implement moving average crossover. And then it writes that for you. And then you you sort of strip out the fixed parameters it's put in because it might most likely, if it's a consumer grade AI, it has given you the golden cross, which is a 50, 200 day moving average. So you take out the 50 and 100 day parameters, replace them with you know A and B, and then that's it. You're done with AI. AI AI's is given you, and obviously. for someone like me I wouldn't need AI to do any of that because I already know all these things. But if I was starting from scratch or if I was working in an area that's novel to me or if I wasn't a confident Python programmer, then definitely AI can accelerate that part of the of the process for me. I would then put take that Python code and put it into a a proper robust fitting backtesting Testing framework that had have not been written by AI. Although having said that I you know, for example, even with that, I would say there are two other good things that I think AI will be
50:31useful for. One is is to suggest changes to your code, not necessarily to make the changes, but I think it's quite good at potentially finding bugs and things like that or making style improvements. Although, having said that, based on my experience, if your code is really badly like messed up spaghetti code to begin with, then you know the AI help is not. gonna be helpful because it's you know you want to have something that's giving you a very targeted suggestion on the small part of your code. So if your code's properly modularized, then that the AI will do a good job of that. Otherwise it won't necessarily work. And it's also good for things like writing tests because, you know, or short pieces of code that aren't uh you can could easily verify the correctness of yourself without, you know, so less than fifty lines probably, I'm afraid I'm not a fan of like fully agentic AI code generation where you've got some AI code written some other code and checking it thinks it's okay and then just pushing that all to Git without any humans doing anything more than scanning it. I'm sorry, that's going too far for me. So yeah, this modularized back testing framework would be um human written, even if you've used AI to kind of help with the edges as I of said. Because you know, the last thing you want is your um I th I believe it's just too dangerous, no
51:40matter how good your prompts are, how good your agents are, and I know Bell, you've got a lot more experience in this area than me. So you'll you'll no doubt think that I'm being far too old fashioned. But um you know I I think um AI is really good at the start of the process, generate the idea, get the idea to be tested. But once we move into the testing and implementation phase, that's just too important to leave the let to trust with AI with. That has to be done properly.
52:04Unidentified speaker (speaker_0): I think that's a very fair point that you're making there. And then in terms of you doing work uh for the public good, so to speak, you also do publish a lot of your code into GitHub, which is available to anyone, as far as I understand. Is that correct?
52:19Unidentified speaker (speaker_2): Yeah, so I've got one big project which is called Pi System Trade, which is basically the the open source version of my trading strategy. And that um actually so I I'm no longer the sort of maintainer of that effectively. I've sort of found, I've sort of pushed it fully into the public domain. So it's now actually hosted on a uh on a an organizational GitHub page um and other people now taking over the maintenance of that. But it's still my name on the door, so to speak. So that's the main thing thing that I've sort of pushed out there. But yeah, I'm always I'm always writing on bits of code for doing things like optimization and stuff. And uh, you know, you go back to your earlier point about um for um you know an academic paper, you know, this is my checklist. Well, for me, you know, I feel like if I'm saying publishing a, you know, writing about a training strategy or something like that, then I why is why wouldn't I put my code out there so people can well firstly they can check if they want to see if I've, you know, just made this stuff up. The more But more importantly, you know, if you want to implement it yourself, then it's a lot easier to do that if you've got a starting point rather than just
53:20trying to take someone else's language and and convert it into Python. So yeah, I'm a big fan of I think open source is one of the greatest inventions in the last 50 years. I mean, without open source, we wouldn't I
53:28Unidentified speaker (UNKNOWN): mean
53:28Unidentified speaker (speaker_2): I've got so I've got um an Android phone. So, you know, that's running effectively running an open source operating system under the inside the evil Google wrapper that I put around it. I'm running Linux. There's no doubt Linux are um being used somewhere in the uh in on the internet between you and I at this moment. So without open source we wouldn't have had any of that. So yeah I'm a big fan of open source. I practice what I preach for sure.
53:49Unidentified speaker (UNKNOWN): And
53:49Unidentified speaker (speaker_0): that's great. And just to round off, when is your book going to be published, The Arts and Science of Trading and how can people get hold of it?
53:56Unidentified speaker (speaker_2): So it's going to be published on December the first. So if you can buy it through a route that doesn't involve going to the largest monopoly bookseller in the world, that would be great. Partly because I believe people should put you know support local bookshops and local booksellers. But also, I get slightly more money that way. But uh, if you can't find it a good bookshop, then Amazon definitely stocks it. But if you go to my webpage, which is um systematicmoney.org, all one word, then you'll find a link to more information about all. of my books and also kind of links to places where you can order or pre-order the book from.
54:31Unidentified speaker (speaker_0): That's great. No, so with that, uh Rob, thanks a lot as usual. It's fantastic uh speaking to you. Again, it's great to get kind of uh a reminder of good discipline. You know, I think this applies to everyone, you know, not just uh, you know, less experienced people. And you know, keep up the good work.
54:46Unidentified speaker (speaker_2): Thanks very much. Thanks again for inviting me on, Bilal, good to speak.