Photo courtesy of Harriman House. As you can see, the book makes an excellent books-stand for itself |
* Official publication date is 29th October. Actually I finished the book in late August, and I've had print copies in my hand since mid September (and I know some other people have received their copies already), but for some reason I have never understood my books are always released at the end of October in odd numbered years (Systematic Trading 2015, Smart Portfolios 2017, Leveraged Trading 2019). Whether I stick to this schedule for book four is an open question.
You can order the book from the publishers, Harriman House, here (other, massive monopolistic online bookshops are available, but I get a better royalty if you buy direct from the publishers). There is plenty of information on my website, here.
This post is aimed at regular readers of my stuff who might be wondering if it's worth buying this third book (short answer: yes, of course! Even shorter and more honest answer: maybe!).
Where did the idea for this book come from?
Two sources of ideas really. The first was a constant stream of people telling me that they loved "Systematic Trading", but with two huge caveats. First caveat was the portfolio sizes in the book, which started at $100,000 and just kept on going up into seven and eight figures. The second caveat was that it was all just too hard and required far too much pre-existing technical knowledge. Classic amazon review "This won't not be the first book on systematic trading you buy, or even the tenth...". Thanks: I am not sure that even I own 9 other books about systematic trading!
Although I'd originally intended to write a book which was accessible to smaller retail traders my publisher and I decided that I'd be better off writing something more technical that was aimed at institutional traders. There is a bit of stuff about minimum capital in the latter part of the book but mostly I assume you have lots of money and lots of knowledge about the markets.
The second source of idea was my habit of watching "trading guru" youtube videos, or at least having them on in the background whilst I worked. I find these videos fascinating, but I also find it absolutely terrifying that these people are influencing tens or even hundreds of thousands of people. The more I watched these videos, the more I noticed when just walking around or surfing the internet the increasing levels of advertising from brokers peddling dangerously leveraged products. This was all pretty bad already, but when the Crypto currency hype got going in 2018 it reached another level. Over the last few years I've noticed more and more "ordinary" people asking me for advice which inevitable involves trading "four-ex" or Bitcoin.
I don't write books primarily for the money, I write them as a philanthropic gesture to help as many people as possible avoid the plethora of ways to get ripped off in the financial industry. It struck me that relatively inexperienced retail traders using leveraged products are a very large group of people who needed a large amount of help. Maybe I should write a book for them. So I did.
What is the main focus of the book and who is it for?
So the book is about safe leveraged trading. Regular readers will know that I frequently talk about the three main errors of systematic trading: overbetting (taking on too much risk), overtrading (trading too often), and overfitting. This book focuses mainly on dealing with the first two: getting the right level of leverage and the right frequency of trading. It does this by suggesting that novice traders use a system (quelle surprise!). The system is calibrated to avoid excessive risk or trading costs. I also explain how I designed the system to avoid the perils of overfitting.
But it's also a book for those who don't necessarily want to trade purely systematically. In my first book I introduced the idea of a "Semi-Automatic Trader"- someone who chooses which position to hold in some non systematic way, but sizes and closes positions systematically. I take this idea much further here, and explain in more detail how you can combine human intuition with the most useful parts of a trading system. Importantly, I also explain how you should calibrate your risk and trading frequency depending how well you have performed in live trading.
It's a book for relative beginners. As such it doesn't assume much knowledge and also goes into significant amounts of detail about specific leveraged products. I chose five products, because one is common in the US (margin trading), two are common in Europe but illegal for retail traders in the US (spread-bets and CFDs), and two are traded pretty much everywhere (futures and spot FX).
It's a book for traders without much money. A key theme throughout is answering the question "What is the best use of my scarce capital?".
What specific advice is there for smaller traders?
Firstly I talk a lot about product choice. For reasons I explain in the book* there is generally an inverse relationship between the minimum capital required to trade something, and how much it costs to trade.
* If you must know: within product type and across instruments it's vol scaling. But also the more 'institutional' the product, the larger the capital required, the cheaper it is to trade. So futures are cheaper than CFDs for example.
So the optimum choice for a smaller trader is to find a relatively cheap product which they can afford to trade. That rules out most futures (too big) and also a lot of OTC products (too expensive).
If I assume that traders don't have much capital, and are relatively inexperienced, then it makes sense for their first trading system to be binary (no 'forecasts', which requires more capital to do properly), discrete (trades are opened and closed without any adjustment to position size), and close trades using a stop loss (which most people understand).
Then suppose you have slightly more than the bare minimum to trade: what next? Should you diversify by adding another instrument to your portfolio? Should you make your system more complex by adding new trading rules? Should you start trading a non binary system? All of these decisions are discussed, and viewed through the prism of a trader with limited capital.
Why the title, and the focus on leveraged trading?
A few reasons. As I've already said I think that the leveraged end of the OTC retail broker spectrum is one of the most dangerous parts of the trading world, even after the EU introduced restrictions on retail margin levels the availability of leverage is still far too generous. It's also a world to which naive punters are naturally attracted, due to the lottery like payoffs available ("invest £100 and win £10,000!") with massive leverage.
Secondly getting your leverage level right is probably one of the most important decisions any trader can make, and one most retail traders get spectacularly wrong. As an institutional trader your risk target is exogenous and usually fairly modest (even the 25% vol target I run at would be considered punchy in most shops). At a product level you're unlikely to have excessive leverage, unless it's something with really low vol (front contract EuroYen anyone?) or negative skew. But telling a retail trader that they should only use leverage of 1.5 rather than the 50 their broker will allow them is another story.
Thirdly, and very cynically, I didn't want to make this a "system trading" book as this would limit the audience. Of course I want more people to read this from a philanthropic point of view, and the money is a nice bonus (note book is also priced 50% lower than my first two books). I think fewer people are likely to pick up a book with "Systems" in the title than something that name-checks the CFDs and spread-bets that they've been hearing about in the news.
Finally from a technical point of view it's much easier to explain position sizing using inverse vol weighting if you can use leverage. You don't need to worry about risk appetite and the book can be 50-100 pages shorter.
I've already read "Systematic Trading". Should I bother with this?
That is an excellent question, and one I address in the book:
"You will see from my website that I wrote another trading book a few years ago: “Systematic Trading” (ST). Perhaps you are browsing on-line or in your local book shop and trying to decide which of these two books you should buy. Maybe you already own ST, and are considering adding this book, “Leveraged Trading” (LT), to your collection.
To help you decide, the main differences between the two books are:
- As the title suggests, ST is mostly aimed at traders who are enthusiastic about systems trading. LT helps new traders learn how to trade by using a system, but then explains how to combine the system with their own human intuition; the method I’ve named “Semi-Automatic Trading”.
- The trading systems in ST require large amounts of money (at least £100,000; around $130,000). The Starter System in LT needs just £1,100 or $1,500. I spend a lot of time in LT discussing how smaller traders can make best of their scarce capital.
- ST is written for relatively advanced traders with some prior knowledge of certain financial concepts. LT is suitable for novices.
- ST is a generic book which doesn’t go into much detail about individual markets. LT explains how to trade specific leveraged products.
- ST explains the various components of a complex trading system one by one; it isn’t until the book is finished that you can see the entire picture. In LT I introduce a simple system in its entirety which you can start using right away. I then go on to explain how, and why, you could make it more complicated.
- ST explains how to design trading systems from scratch, which requires using software to simulate historical system performance (a process called back-testing). In LT I present a system I have already back-tested. I then explain how you can modify the system for different types of trading, and to cover different markets, without needing any further testing.
Because I have designed the trading systems in this book with the same principles in mind there are some ideas that readers of ST will find familiar, although there is no duplicated content in this book. I would recommend that you read “Leveraged Trading” (LT) if:
- You tried to read ST and didn’t get it.
- You read and understood ST but are struggling to build a simple system from scratch.
- You have not read ST and are an inexperienced trader who is unfamiliar with financial theory and back-testing software.
- You are specifically interested in trading leveraged products: FX, CFD, margin accounts, spread-bets, and futures.
- You do not have enough cash to trade the systems in ST.
- You are interested in combining your own trading intuition with a trading system: semi-automatic trading."
I am the guy who complained there weren't enough formulas in "Systematic Trading"
Just for you there are so many formulas in Leveraged Trading I lost count.
In all seriousness because this book is more 'hand-holding' it does go into more explicit detail and includes formulas. So for example where in Systematic Trading I might have just said "Take a weighted average" here I include the actual formula for a weighted average.
But don't panic, there are no greek letters* in the book, just simple formula that an 11 year old could understand.
* Full disclosure: there might be a few capital Sigma to indicate a summation. But definitely no integrals.
I am the guy who complained there weren't enough numerical examples in "Systematic Trading"
Again, you will love this book! It's my bet that beginners prefer to see concrete tangible examples rather than pages of theory.
I notice there is the usual obsession with costs
Yes! I make no apology for it. Costs are the only thing you can control and forecast (almost) perfectly. If you get your leverage right but trade too often then you will still go bust just a bit more slowly. Also there is a lot of misunderstanding and misinformation about costs out there. In particular brokers in the OTC world will usually promote the low levels of execution costs on non dated derivatives ("zero commission! 0.5 point spread!") whilst quietly hiding the very high levels of holding costs (17 clicks into the website you may just find that you are paying LIBOR+3% in overnight funding charges).
And talking about costs allows me to talk about one of my favourite bugbears: day trading. The idea that you can be profitable day trading, whilst paying retail level commissions and crossing the trading spread, is an idea that deserves to be rebutted constantly (if only to try and match the constant stream of BS from brokers and trading gurus promoting the concept).
There's also a bonus section in this book on reducing your costs through optimal execution tactics. You can find it in the appendix (it was a complete chapter, but was too short).
You really have it in for brokers and trading gurus, don't you?
Yes! Partly this makes the book more entertaining, for the same reason it's more fun to read a novel with a decent villain in it. But yes, I really hate the whole retail trading culture which enriches the shareholders of spreadbetting firms and a bunch of charlatans at the expense of emptying the accounts of generally less wealthy people who don't know any better. This strikes me as worth a different circle of hell compared to the institutional buy and sell side who are effectively payed through a pro-rata tax on everyone's managed money (a tax which you can reduce if you are smart and only invest in passive funds, or avoid overtrading: c.f. Smart Portfolios).
Of course I don't hate all brokers - only the ones that are expensive or dodgy. And not everyone who has ever written a book about trading is automatically dodgy, or that would clearly include me. But you should be extremely skeptical of any trading gurus who claim outlandish returns, who don't tell you what their trading returns are, who aren't open about their methodology, and whose videos mostly consist of them prancing around a villa in Thailand whilst occasionally pretending to trade.
You can rent this villa for your youtube videos here |
Presumably as this is a 'beginners' book there is nothing in here about statistical uncertainty?
You couldn't be more wrong! I had to bring in statistical uncertainty so that people would understand why costs are important when selecting instruments and trading rules, and pre-cost returns irrelevant (since they are not statistically different from each other). I needed the concept to explain why diversification is so important (it gives you a massively statistically significant improvement in returns), whilst adding more trading rules and making your system more complex are good to do but not so valuable (the improvements are more modest and at times skate on the verge of insignificance).
Of course there is nothing in here about the mechanics of fitting, or in and out of sample periods, and you will not see the formulas for parameter uncertainty which I punish my students with every year. But I firmly believe that no trader will survive unless they have a good intuitive understanding of how uncertainty affects financial markets generally and trading strategies specifically.
Who helped you write this book?
The team at my publishers, Harriman House, were their usual brilliant selves. Craig Pearce was the commissioning editor and Stephen Eckett did the editing. I understand why people self publish, but I'm far too lazy and realistically I won't be able to match the professionalism of these guys. It's extremely difficult to be self critical enough to do a decent job of writing a book outline that is coherent and will reach the right audience. It's even more difficult to find someone who is able to read your stuff and give you intelligent feedback on everything from structure to content to grammar and spelling (and believe me, almost nobody can edit their own stuff).
(Anyone who compares my unedited blog and my edited books can clearly see the value of a decent editor)
As with previous books I had three "beta readers" who had to read the shockingly awful earlier drafts before Stephen had turned them into a readable book. Riccardo Ronco also read "Smart Portfolios" and did an equally excellent job with this book. James Udall filled in my not inconsiderable gaps in understanding about CFDs and spread bets. Finally Tomasz Mlynowski, whose day job involves (amongst other things) reading my lecture notes and explaining them to baffled students, brought along his fine toothed eye for errors and mixed metaphors.
Finally, it's a cliche, but writing is a job with flexible hours from which it's difficult to switch off. So it's tough living with a writer. So the three young people, one slightly older person, and cat, that live with me deserve a lot of credit.
Okay, maybe not the cat, which has a nasty habit of sitting on my lap when I am trying to finish a paragraph.
So.... book four?
Yes, probably. This will either be "Smart Portfolios for dummies" in the same way that this book is kind of "Leveraged Trading for dummies", or a more cerebral book on uncertainty in financial markets (which I know already my publisher will be less keen on). Before then however I will be working on another project- a new trading strategy - details to follow in the next blog post.
Congratulations on publishing your third book! I am the proud owner of a copy of the ST book. I learned an awful lot from it. Both on what trading futures entails, as well as how to structure a software program to run an automated trading system.
ReplyDeleteCongratulations, I had a chance to flip through LT - got my copy early and have read ST thoroughly. Am wondering whether omission of a discussion on drawdowns and leverage is deliberate. I tend to apply jelly leverage to the drawdown target in mind. How appropriate is that approach in your opinion ? Regards, Kunal
ReplyDeleteHi Kunhal
DeleteI don't use a drawdown for reasons that are worth a whole blog post to explain (coming soon!). However in LT there is a concept of 'maximum prudent leverage' which is based on the largest concievable loss for an instrument; however this is different from the largest d/d seen for a strategy in a backtest.
Looking forward to reading that blog. Yes I did see maximum prudent leverage. Am not able to get my head around the fact that what if the system fails and true sharpe falls. Then cutting risk at a pre determined drawdown level seems rational and then drawup risk when system repairs. Influenced by Aaron Brown (Ed AQR) on this. Thanks
DeleteIf your true SR is falling, or rather your estimate of it based on backtest and live trading, then you would automatically reduce your leverage according to Kelly. And indeed, that is discussed in the final chapter of LT.
DeleteMany thanks !
DeleteHi, I read the last chapter of LT. Suppose I am running a continuous automated system like ST ; then at what frequency should I compare the realised sharpe with the backrested sharpe to manage the risk, especially if I get a streak of negative runs ?
DeleteDo you mean how often should you compare? I guess you could compare after every trade, but the ratio won't change so much.
DeleteRob could you please elaborate the technical note 145 accompanying Table 72. I need to recalculate the table for markets in India but not sure how are you converting performance ratio to sharpe. For futures trades would you still need to subtract risk free rate in sharpe ? Also, I understand the sampling distribution formula but N is no of periods but we have no of trades in the table above ?
ReplyDeleteIt isn't trivial to recalculate these tables, and there is no reason why you'd need to do this. The effect of the risk free rate changing is almost zero.
DeleteHi,
ReplyDeleteThank you for your new book. I just got it a few days ago. And I look forward to read it. I skimmed it and I'm already questioning day trading which I had begun.
Your other book, Smart Portfolios was a life saver. I found a decent all world etf and continue adding to it.
Thank you for your books and sharing expertise.
Matt
Thanks hope you enjoy the new book.
DeleteHi Rob, One of the issues I have with the Forecasts both in LT and ST is that they jump around from day to day. For eg a +10 would go to -1 the next day. The changes are not smooth. I have to admit that I am using Mean Abs Dev calulated over a small sample to convert raw forecasts to scaled forecasts (as explained in ST and a blogpost). Apart from using a 'wrong' scaling factor, is there something else that could be causing the jumps and leading to large losses, especially in trying to do continuous trades. Thanks
ReplyDeleteWhich trading rule is this?
Delete5/20 Crossover on a cumulative normalized series and a 9 period Breakout system (Un-smoothed)
DeleteDo you have a .csv with the forecasts for each individual system in?
DeleteIf so you can email me at rob AT systematicmoney.org
DeleteThanks, sent you an email
DeleteHi Rob, I'm an avid reader of this blog and own all your three books (I got my copy of LT in mid september and I read it in a weekend!). My knowledge of finance was pretty much 0 before, but I always wanted to do something about it, and being very technically minded your books were a godsend. Not only I opened my portfolio of ETFs as in Smart Portfolios, but I'm now testing a small trading account. I just wanted to thank you for your work! I also have some specific questions but I'm not sure if writing them on the blog is the best way to contact you. Thanks!
ReplyDeleteThat's great to hear. You can email me at rob AT systematicmoney.org
DeleteCongrats Rob! A great achievement. I am your loyal reader, had bought your 2 books and really enjoyed reading.
ReplyDeleteHi Rob, since not all the capital is used while trading, where would you put the spare capital? Just leave it as cash in the account? That seems to be a bit of a waste, no? Thanks.
ReplyDeleteGreat question
DeleteI assume you mean 'not use for trading' to mean 'not required for margin'.
You will always need to leave some surplus capital in the account to cope with losses that occur before you get a chance to top it up, but let's put that to one side. Let's also put to one side the fact you might want to minimise your exposure to your broker failing; if your broker is that dodgy you shouldn't be using them.
If you're going to put the cash somewhere else it will need to earn a return, or there isn't much point. Brokers do pay some interest on cash balances, and in my experience it's similar to what you can get from a bank. So to earn more you will need to take risk, and buy a higher yielding asset. That means you could be in a position where money you have earmarked for trading capital, itself falls in value. Not ideal.
The other problem is that if you keep some of your capital outside of your account, you will get in the habit of topping the account up when you have losses. My concern for many traders is that they will then continue to top up the account even when they have depleted the capital they meant to use for trading.
Thanks Rob. Another question: you mention in your book that your backtest indicates FX risk is minimal, and it's not worth hedge FX exposure. However, look at the GDPUSD :) - I'm a Canadian, and I hold a significant portion of my portfolio in USD for trading. So if I dare ignore your advice about FX exposure and insist on hedging FX exposure, what will be the most optimal way? Thanks for sharing your knowledge with the world!
ReplyDeleteWell, I have a GBP trading account, and then I only convert what I need for margin into other currencies. My excess cash stays in GBP (my home currency). I guess if you forced me to hedge I would use futures, but that's only valid if your account size is big enough. You can quite easily work out the cost of doing it (carry plus slippage costs).It looks like the carry is in your favour if you sell USD and buy CAD, so that's good.
Deletebook 4? Simple, Python for Systematic and Quantitative Trading PSQT. There you go :)
ReplyDeleteHi Rob
ReplyDeleteOn pages 25 & 26 of Leveraged Trading, talking about using CFDs, you say "My broker has a minimum margin requirement of 3.33%, equating to a leverage factor of 30. As I will discuss later in the book, this is too high. To use a more sensible leverage factor of 10 we’d need to deposit £ 1,000"
I've checked out IG and Saxo and I don't see any way to change the leverage factor. It seems it's not a minimum margin, but rather the actual margin used. Am I missing something or is this just the way CFDs are structured?
Hi,
DeleteThis might be an example of where you should read the entire book before asking the question :-) Nevertheless, I will briefly explain. Suppose you want to buy a £3,000 CFD with a margin requirement of 3.33%, i.e. £100. If you only had £100 in your account, then you'd be operating at a leverage factor of 30. If instead you put £300 in the account then you will be on a leverage factor of 10. Of this £300, only £100 would be required for margin. And you're right, this figure cannot be changed. The rest is there to cushion you against further losses.
Ah right. Many thanks for taking the time to explain this to a newbie. Cheers.
DeleteHi Rob,
ReplyDeletequestion on your book leveraged trading. In the book you got 6 different "strategies" for the breakout and 6 for the MA model with different lookback periods. Let's say trading costs are cheap enough for me to trade them all. Shall I equal weight then or what weights shall i put on each of the strategies when calculating the non-binary combined forecast.
Equal weight is fine.
Deletethanks Rob. much appreciated. One more question.. in "Leveraged Trading" you use (simple) moving average whilst in "Systematic Trading" exponential weighted moving average. With me understanding and being able to calculate both, shall i go for ewma ?
DeleteYes go for EWMA. I use SMA to make it more understandable
Deletethanks. is this correct then in python for the 64day ewma?
Deletema_short = data.ewm(span=64, adjust=False).mean()
Yes, but I don't use adjust=False
Deletehttps://github.com/robcarver17/pysystemtrade/blob/master/systems/provided/futures_chapter15/rules.py
also, when I want to implement your system without carry. Would you place 50% or your risk on the MA system and 50% on the breakout. Or more on breakout as I can see from you chart that this has a slightly higher sharpe ratio?
ReplyDeleteNo 50% each is fine.
DeleteThere is no statistically significant difference in performance between the two systems.
Hi Rob,
ReplyDeleteone more question about the book. You say for oil you usually trade the december contract as it's a seasonal commodity. For STIR you might trade a contract 3 years out.
Do you backtest on these contracts then as well? or is your backtest based on front month contracts?
I backtest on the contracts I will trade
DeleteIs there a typo on page 314? Third spreadsheet formula from the top says:
ReplyDeleteD26=D26*16, E27=D27*16, …
should be
E26=D26*16, E27=D27*16, …
correct?
Yes well spotted.
DeleteI love all your 3 books. I was a long only invester, and purchased your Smart Portfolio. Since it was too good, I purchased your another 2 trading books. They are also too good to make me feel like starting trading, and in fact I started my trading trial for about 2 months. I found it is a fun excesise (and also dangerous if I don't do this properly).
ReplyDeleteHowever, I recently noticed that I have not taken your advice seriously about the cost calulation, and the trading rules selection by instruments. For the last 2 months, I have been using your cost classfications of 'cheap' and 'expensive' for instruments, assuming it can apply to my case (I use Interactive Brokers which I believe is same as yours). Yesterday I finally started my calculation, to find that the result seems very different from the examples shown in your book (yes they are just examples!).
Here I got questions on the cost calculation on the appendix B, particularly for the futures, which I hope you can provide some answers.
1) Transaction Cost: Do the 2 parameters I use for corn futures look alright? The numbers are different from the numbers in the book, although I believe we use same brokers - Interactive Brokers.
commission: $2.82 ($1 in your book)
spread: 0.25 (0.01 in your book)
2) Holding Cost: Is my understaing correct for roll trade multiplier N?
N = 1:
when you can use spread orders,
or, when you can wait contract expiry date for your instrument to be liquidated
N = 2:
the rest of the cases
3) Determining Right Rules by Instrument: do you dynamically change the rules to apply by instruments? Volatility change in time can play a big role here. For example the corn volatility is about 6%, which is about half the number in the book (12%). And this can happen in any instruments (and I think it does today). My calculation says that, for corn futures, I need to remove mac2-8, mac 4-16, breakout10, breakout20, and even breakout40.
Finally, I hope you write another new book in the near future. I definately buy one!
1) I think when I wrote the book I must have been getting a special rate! I'm currently paying $2.37 in commission on Corn. And yes the spread is 0.0025 or 0.25 depending on how you express the price (if it's ~$400 then it's $0.25)
Delete2) Yes
3) Yes volatility changes over time. Incidentally I have corn at around 9% right now but let's not quibble. However I take the view (rightly or wrongly) that risk adjusted costs are roughly stable over time. I'd rather not be constantly taking rules in and out of my system (I've not changed the rules I use in 7 years). A big problem here is if you fit the system when vol is unusually low or high, so you may want to use an average of vol over the last 5 years when decidiing which trading rules to use.
Thanks for the kind words about the next book. I will definitely write one at some point!
thank you for the answers Rob! It makes sense to use the recent years average volatility. I take your advice. And also I will check my volatility calculation...I now use 25days normal standard deviation, but will see how it changes if I use the exponentially weighted measure (guess this is not the reason).
Deletejust to let you know that I got 9% vol when I use the ewm. I will change all my vol calculation to ewm.
DeleteThis comment has been removed by the author.
ReplyDeleteThis comment has been removed by the author.
ReplyDeleteHello,
ReplyDeleteI already sent this message by mail but I suspect some problems with my internet provider. My apologize if some redundancy here. All your books are inspiring on how to tackle financial problems and incertainity.
After reading LT, I tried to put calculation costs in practice with FX. The broker (XTB) does not specifically mention interest rate for lend or depo but in place provides daily swap points for long and short.
How may I adjust formula in the cost calculation sheet to get the holding cost percentage.
Thank you.
Hi Rob,
ReplyDeleteCould I know why the "scaling factor" for moving average crossover shown in your 2 books are different, Systematic Trading and Leveraged Trading? The scaling factors shown in Leveraged Trading are about 14 - 17 times greater than those in Sytematic Trading.
Systematic Trading: 10.6 / 7.5 / 5.3 / 3.75 / 2.65 / 1.87
Leveraged Trading: 180.8 / 124.32 / 83.84 / 57.12 / 38.24 / 25.28
I don't think it is due to the different moving average techniques used in the 2 books, or is it? (EWMA in Systematic Trading, simple MA in Leveraged Trading). I came across this scaling factors in pysystemtrade introduction docs also. Numbers quoted there are the ones shown in Systematic Trading.
Thank you in advance for helping me understand your books better!
That is weird. I'd need to go back and check there, but my gut feeling is that the ST numbers are definitely correct for EWMAC as that is what I'm using in my own system.
DeleteThank you! It could be because the formula in LT is return vol based, whereas in LT it is price vol based? I am not so sure.
ReplyDeleteWhat makes the difference seems to be simply because the vol used in the 2 books are different (the 16 effect!). In ST it is daily vol, and in LT it is annaulized vol, which I find is much easier to understand for novices like me. Sorry I should have read the book more carefully...
DeleteI also noticed that the ratio between the 2 sets of scaling factors are not consistent acorss speed variations (slight difference). This could be due to the EWMA vs simple MA, or something else. Anyway I will change my formula to use ST numbers (with proper annual conversion), as I now use EWMA thanks to your advice. This should reduce the weight of faster variations, which I believe is not a bad direction. Thank you!
Ratio of Scaling factor - LT / ST:
17.1 / 16.6 / 15.8 / 15.2 / 14.4 / 13.5
Ah! Well done. I really should have seen that myself. Also goes to show that writing a book is such a traumatic experience that I try and forget it as soon as I write it.
DeleteLooking at the figures there, I'm pretty sure it's EWMA vs MA since there is nothing else, and that clear pattern changing over time is one I would expect. But in the long run that is a difference that will make no difference to your systems actual positions so I wouldn't worry about it.
I can only imagine but do understand how hard it would be. Being asked by someone for the work you did long time back is not a confortable experience, and something I would like to avoid where possible.
DeleteYesterday I started reading the book by Andreas Clenow, which is one of the book you recommend. In the book he rightly say that stop worring about the minor details. EWMA or MA, parameter tunings, and so on are exactly the ones he mentioned as minor details!
I find the book a must read. I am determined to use more time in learning programming so that I will be able to run backtesting myself. Howerver I don't think I can stop worring minor details completely so please expect questions from me from time to time... Thank you!
Hi Rob,
ReplyDeleteI'm reading the book and wounder if it will still be profitable to use ETFs instead of CFDs or Futures. as I live in the US I can't trade in CFDs or Spread Bet and I don't have the capital to trade Futures.
Thanks!
The obvious disadvantage of ETFs is that you can't as easily go short or go leveraged. But if the costs check out, then there is no reason why you shouldn't trade ETFs.
DeleteErrata for Leveraged Trading
ReplyDeleteHi Rob,
In case you're planning another edition, or for posting an errata section on the book's website, I've found a few errors in Leveraged Trading:
p.207 should say "Method three: Relative to next contract"
p.277 "Rounding down to a ratio of 0.35 we can see that the expected Sharpe ratio in table 74.."
- should read "table 72"
There is no Table 74. p.276 has Table 73 and p.300 has Table 75 (unless I've missed it)
p.315 Rolling average calc: either remove the "AVERAGE" function and change to (B10+C10)/2 or remove the "/2" (compare with formula on p.199)
Thanks for that
DeleteHi Rob,
ReplyDeleteI'm currently looking at the Spread Betting system, Starter System with 5 Trading rules and 4 Instruments. I've not yet looked at re-calculating costs or capital requirements, and I'm going with what you have in Leveraged Trading. In Chap 7, for Spread Betting on Equities you recommend Nasdaq and Euro Stoxx. Nasdaq is cheaper but requires much more capital (£5,500). In Chap 4, Table 4 has S&P 500 with the same cost as Nasdaq, 0.015, but more lower capital (£1,800). Is there a good reason to go for Nasdaq instead of S&P 500?
In particular, with a multi-instrument set up, where initially you recommend equal capital per instrument, S&P means much lower overall capital requirement.
No reason not to go with the lower capital instrument if costs are not an issue.
DeleteBut please check these figures according to what your broker offers and current volatility as they will change.
Thanks - and will do
DeleteHi Rob,
ReplyDeleteI've calculated costs using the formulae in Leveraged Trading based on Spread Betting at CMC Markets.
I don't know how easy it is to view this, hopefully it can be easily pasted into Excel. I don't think I can format a table in a comment.
All are for Dec 2021 contracts.
Gold, Nasdaq, S&P 500, Euro Stoxx are broadly in line with what you have in the book, although S&P is expensive due its spread (which has moved to 7 since I did the calcs).
Bonds are completely off compared to what you have in the book - and if my calcs are correct, way too expensive to trade using Spread Bets. This comes from their Instrument Risk being much lower, the absolute Price being much lower, and so Spread as a proportion of price is much higher. I'm guessing that when you wrote the book, spreads on Bonds were a fraction of what I'm seeing. I've had a quick look at IG, and their standard spread on Gilt futures is also 2, so this isn't unique to CMC.
I'm keen to know your thoughts.
Gold Nasdaq S&P 500 Euro Stoxx 10 Yr Gilt Bund
Transaction Costs
Point Size 1 1 1 1 1 1
Bet per point £0.50 £0.50 £0.10 £0.60 £0.10 £0.10
Spread 0.4 2 6 2 2 2
Current Price 1761.80 14759 4354.00 3997 124.79 169.65
Instrument Risk 14.9% 15.8% 13.5% 16.8% 4.40% 3.60%
Trades/year 6.8 6.8 6.8 6.8 6.8 6.8
Cost per Trade £ £0.10 £0.50 £0.30 £0.60 £0.10 £0.10
Cost per Trade % of min exposure 0.011% 0.007% 0.069% 0.025% 0.801% 0.589%
Risk adjusted cost per trade 0.0008 0.0004 0.0051 0.0015 0.1821 0.1637
Holding Costs
Rolls/year 6 4 4 4 4 4
No. of trades/roll 1 1 1 1 1 1
Annual Holding Cost 0.0681% 0.0271% 0.2756% 0.1001% 3.2054% 2.3578%
Risk adjusted holding cost 0.0046 0.0017 0.0204 0.0060 0.7285 0.6549
Total Costs 0.0098 0.0046 0.0551 0.0161 1.9669 1.7683
Minimum exposure £880.90 £7379.50 £435.40 £2398.20 £12.48 £16.97
Target Risk 15% 15% 15% 15% 15% 15%
Minimum capital £872.33 £7773.07 £391.86 £2685.98 £3.66 £4.07
Can you put it into a google sheet or something?
Deletehttps://docs.google.com/spreadsheets/d/e/2PACX-1vTICNmIa8bhW3XIXcQWr74tj4UPSfT2WRSODzq9tc51mwsCdescDttuo9PmgTMIBmiP6MQStRlnaIZL/pub?output=xlsx
DeleteThanks. Figures look about right to me, at least your calculations look okay, obviously only you know if the input data is right.
DeleteThanks for checking - it does highlight though, that it's important to review costs from time to time, if you have a favourite list of intruments. If risk falls and spreads widen then costs can make an instrument unviable... and if the price is a relatively low number (e.g. 150 for bonds vs about 15,000 for Nasdaq) then it will be more sensitive to spreads widening.
DeleteHow come you don't have the same issue with your futures portfolio, if you have Gilts or Bunds? The risk and price will be the same - the only difference will be in the spread and the trades/per year. For these bonds to be below a speed limit of .08, I need a spread < 0.1
ReplyDeleteFrom here https://github.com/robcarver17/reports/blob/master/Costs_report the bund half spread is 0.005 so spread is 0.01 and the SR cost per trade is about 0.002
DeleteA 2 point spread on bonds sounds insane... is the quote really 169 to 171?
I don't trade gilts.
I just opened a cmc demo account and the spread is 0.02, not 2. So there is your problem.
DeleteAh - rookie mistake. I was looking at their "Product Overview" pop-up (and their Quote Panel Grid). For Gold it says the spread is "0.4", and that is the difference between Buy and Sell. For Gilts and Bunds the popup says the spread is "2.0", but yes, the difference between Buy and Sell is 0.02
DeleteIs this a market convention?
Anyway - many thanks for sorting that - I was thinking I needed to give up on entire asset classes!
I'd never use the brokers 'product information' for spreads, as they're usually biased downwards. Have a look at the price quote during liquid hours and see what the difference is between bid and offer. Ideally do this over several days and take an average.
DeleteNot sure it's 'market convention' but broker convention is to be as confusing as possible and make it very difficult to make fair comparisons or calculations. Spreads are quoted in a variety of ways: 'pips', 'points', 'ticks'... the list is endless, but what you want are the same units as the price or it won't make any sense.
Yes, lesson learnt regarding the brokers :-) I appreciate the time you've spent on this - thank you
DeleteAfter playing around in the Demo account, I realise now that the confusion was caused by Point Size.
DeleteGold & Nasdaq have a Point Size of 1
S&P has a Point Size of 0.1
Gilt and Bund have a Point Size of 0.01
I haven't been able to find where the Point Size is described in CMC for each product, it was only by entering values into the Order Ticket in the Demo account that I could see this. So their reported Spreads are correct, factoring in Point Size. I now have better numbers for my Total Costs, and also corrected numbers for Minimum Capital (which was previously looking suspiciously low for Bonds)
dr1ver - it seems we're on a similar journey (and I too am using CMC).
ReplyDeleteHandy tip: when you open an Order Ticket, the font size of each digit in the price immediately clues you in to the value (in *points*) of the instrument. For instance, it's no accident that Gold quotes "1846.80" with the ".80" in a much smaller font.
One frustrating aspect of spread betting for me is that I'm not sure where the prices are actually sourced from.
Take Gold Dec 2021:
On Friday, CMC Markets has last price at 1847.05.
On Friday, CME/Globex has last price at 1846.8 (GCZ1)
On Friday, ICE has last price at 1847.7 (ZGZ21)
My best guess is that CMC Markets may take a weighted average of the CME and ICE instruments... but as far as I can see, CMC Markets don't disclose this information anywhere.
BTW dr1ver, if you continue down this journey it may be worth the two of us keeping in touch (e.g. via Rob's comments section LOL) - as we may be able to avoid duplication of effort (I also have access to a Bloomberg terminal which may be handy...).
Rob - great blog... so sad I didn't find this several years ago!
Thanks for the kind comments. Just on the price source query, don't forget your orders aren't usually sent to a market if you're spread betting, just set by the broker. Logically they will be equal to price in the market where the broker hedges, with an adjustment reflecting their current positioning in unhedged stock.
DeleteHi Harry_Hindsight,
DeleteI'm happy to share progress and ideas. I don't know if Rob wants the comments section to be for broader discussions directly between users/traders or if there is a more appropriate location to discuss...
And I echo your thoughts - it is a great blog.
I'm working through the latter chapters of the (excellent) book this evening. A couple of observations:
Delete1) Notwithstanding footnote #132 on page 249... I am struggling to digest the scaling factor of 33.5 for N=80/160/320. If we accept the scaled breakout price can climb as high as 0.5 (but no further), am I right to conclude the maximum "scaled forecast" for these Breakout rules is 33.5*0.5 = 16.75? And if so, is the Breakout rule permanently hobbled in its ability to contribute towards the overall forecast (whereas MAC rules *can* contribute +20s)?
2) Page 253... should the calculation for notional exposure include "Absolute" notation somewhere? In my spreadsheet I am calculating a negative forecast (based on 3 MAC and 3 Breakout rules, all equally weighted for now). Hence my "notional exposure" is negative (which is helpful by itself, I suppose, if it immediately informs me that I should be *short* at this moment). This generates bizarre "deviation from ideal exposure" outputs.
...I'll probably wake up tomorrow with a clearer head and regret sharing what may turn out to be nonsense observations!
...As always, the answer certainly to #2 at least can be found at a later point in the book: pg.263: "note... the ideal exposure is short: the sign of the forecast has changed".
DeleteAnd indeed it seems deviations as large as -255% are not a symptom of a broken spreadsheet as per pg.264.
Hi Rob,
ReplyDeleteI'm heading down the path of migrating my trading from Excel to Python, to benefit from all of the additional functionality.
I'm using Spread Bets, not Futures. Trading will be manual as will updates relating to the account, which is currently with CMC Markets.
At some point, I'll have a mix of Futures and Spread Bets.
Market Data comes from Barchart.com
Are there any particular issues that I'll need to deal with in terms of getting this working in pysystemtrade?
Ones I can think of are:
- Spread Bets are in GBP although the quoted prices are the local currency prices (e.g. WTI is the USD price)
- Instead of the Futures contract multiplier, there is the size of a point
So to if I understand you're getting futures data then executing eithier in spreadbets or futures?
DeleteYes, I think the main thing you will need to do is set up a little spreadsheet (!) that translates futures trades into spread bet points taking account of FX and point size differences. You might want to run pysystemtrade with 100* the notional capital so that you don't get a rounding issue that isn't really a problem.
ok - and thanks for the quick response
Deletedr1ver - after faffing around with Excel to get a feel for things, I too am transitioning to Python. If you start a github repository for your work, consider sharing a link here. At the very least I may be able to provide a massive injection of error-free historical EOD data (can't recall off-hand if this is already available on Rob's!).
DeleteBTW...
I recently had a go at manually stitching (panama method) sequences of futures contracts data. I found negligible difference between my resulting "stitched" datasets and the "backadjusted" datasets that I am able to access from other sources.
Hence, feel quite comfortable skipping the manual stitching in favour of the ready-made backadjusted data.
I also did some comparisons of the backadjusted datasets with the corresponding instruments on CMC Markets.
The correlation in each case was 99.5% or higher. In other words... although the prices on CMC Markets are not identical to those unfolding on the "proper" Futures exchanges, they do seem close enough.
Did you check the correlation of returns, or just prices?
Deletecorrelation of prices only. Intuitively I'd expect 99.5% correlation in price to be reflected in very similar result for the returns.
DeleteBut since you're asking the question, that suggests to me my intuition is wrong - so I will revisit this!
@dr1ver @Harry_Hindsight I have created a fork of Rob's project pysystemtrade which trades spread bets with IG: https://github.com/bug-or-feature/pysystemtrade-fsb
Delete@ageach - thanks - I need to take a look at this
DeleteHi All,
ReplyDeleteI was wondering if someone may be able to help me please.
Going through the excellent leveraged Trading book, superb. Can't rate it highly enough.
One thing i'm slightly confused on is what type of product i'm using with my broker - CMC Markets. I signed up to a CFD account and the product document says CFD but i when i come to calculate some of your formulas in the book, i don't know if i should be using CFD per contract or CFD per point. Eg p59...I'm never sure which formulas apply to me
Many thanks for any time and help,
Tom
They are basically the same thing, it's just framing. You want to know how much in £££ you will make or lose for a given price movement, with a given position. If you bet £1 a point, and a point is 0.10 of the price, then you will make £10 if the price goes up by 1 unit. As long as that holds you have done the calculation correctly.
DeleteThanks Rob, I think i'm getting it a bit more now. The thing that still confuses me on CMC is that I buy in 'units' - say i bought 10 units, is that like saying i've bought 10 contracts? Or does 1 unit mean i'm risking a £1 a point (if i was buying say the FTSE 100).
DeleteSorry if these seem like beginner questions, i'd just rather make sure i know exactly what i'm doing.
It's good to know what you are doing. I'm not a CMC customer and to be honest I found there examples confusing. However this feature is useful; "The deal ticket in the platform doubles as a CFD calculator and will show your expected profit and loss for a trade before you execute, based on the number of units and stops/limits you’ve specified." if you specify a stop that is one price unit below the current level you can work out your p&l for a given price movement for some number of units.
DeleteYou can use the Demo Account in CFC, which uses their live prices. It will give you a Valuation for each Instrument that you have a position in, which is equivalent to what the book calls Current Exposure (at least, that's how it works for Spread Betting, not sure about CFDs)
DeleteHi Rob,
ReplyDeleteIf you have time could you please see if I’m on the right track here with my calculations for p58/59 of the Leveraged Trading book please? Struggling to understand parts of it.
Broker - CMC Markets ( Which I think uses CFD’s per point)
Market - US SPX Dec 2023 - lets say the price is price 4260.00
My starting capital - £10,000
Formula 1 - Annual risk = notional exposure home currency x annual std dev
So, first I need to calculate my exposure = Amount per point x price / point size
- My question with this is, how do you decide what the amount per point should be? How do you find this number? £5 a point? £1 a point?….I don’t understand
Lets say I choose £1 a point, I would do :
Exposure = £1 x 4260 / 1 = £4260
Annual Risk = 4260 x 12.79% = £544.85
Many thanks
Tom
"how do you decide what the amount per point should be?" perhaps read the whole book or at least the next few chapters before asking a question that will be answered ...
DeleteHi Rob,
ReplyDeleteOn p105 when working out the minimum capital required, we are required to find our minimum exposure first, I noticed it says ‘your exposure will be equal to the minimum bet per point, multiplied by the price’. I just wanted to check, does this mean the minimum units I can buy for an instrument? Eg on CMC (CFD per point) with Nasdaq as my instrument, the minimum units I can buy is 0.02 units, so would I then do 0.02 x 14300 (nasdaq price) ?
I only ask because on p59 it doesn’t mention anything about ‘minimum bet per point’ so just wanted to check I was doing it correctly?
Many thanks
Tom
I also use CMC, but for Spread Betting.
DeleteIn the UK, Spread Betting is tax free, whereas CFDs are subject to Capital Gains. I'm wondering why you've decided to use CFDs.
There are downloadable spreadsheets in the Resources section of the book website that contain the various formulae.
"I just wanted to check, does this mean the minimum units I can buy for an instrument? Eg on CMC (CFD per point) with Nasdaq as my instrument, the minimum units I can buy is 0.02 units, so would I then do 0.02 x 14300 (nasdaq price) ?" yes
DeleteThanks both.
DeleteI chose CFD's on CMC because i didn't know the difference between the two to be honest. Yes will check out the sheets, already had a look through them
I've got myself to p161, end of section 2 now and i'm fairly confident with the system, but it's the position sizing i'm still not 100% on with some instruments, i think the whole 'units, points, contracts' thing is still not straight in my head with how my broker CMC does it.
For example, if i trade say GBP/USD - there are no units, my minimum i can order is 400 GBP - so in the book would i do the formulas that are under the 'Spot FX' titles?
Thanks for your time and help :)
Are you trading GBP/USD as spot FX or a CFD?
DeleteIt's ok i think it's all coming together now. It's a spot fx CFD i believe. Thanks for the help on the other posts too. Really appreciate that.
DeleteHi Rob,
DeleteQuick question if I may please when calculating my ‘fx exposure’ with spot fx.
Seeing as my account is in GBP, if I was trading the GBP/USD pair, would I just use 1 as the fx rate?
For example :
GBP/USD
Capital - £2000
Annual std dev - 7.35%
Target risk - 12%
Notional exposure = (12% x £2000) / 7.35% = £3265
FX exposure = £3265 x 1
If I was trading EUR/USD I would do the following :
EUR/USD
Capital - £2000
Annual std dev - 7.18%
Target risk - 12%
Notional exposure = (12% x £2000) / 7.18% = £3343
FX exposure = £3343 x 1.14 (GBP/EUR rate) = €3811
I’m pretty sure that’s correct but just making sure I’m on the right path.
I also just wondered why on p136 for the AUD/USD example you have wrote that the AUD/GBP rate on 18 June was 1.781 - don’t you mean the GBP/AUD rate?
If I use the AUD/GBP rate (approx 0.56) it doesn’t seem to total correctly as I’d expect.
Thanks for any time and help,
Tom
Hi Rob,
ReplyDeleteIn ‘Leveraged Trading’ p140 I’m seeing quite different prices to what you have listed in the book for Gold futures, Corn futures and Euro Stoxx futures.
I’m using TradingView for my prices, What am I missing here?
I did see yours says price shown reflects paying the spread, but even still they would be quite far off still (accord to the current spreads for these on my broker CMC)
Gold :
Yours - 1,268
Mine - 1,278
Corn :
Yours - 379
Mine - 353
Euro Stoxx :
Yours - 3,391
Mine - 3,427
Your AUD/USD and SPY prices are the same as mine, it’s just the others that are off and I can’t think why?
Many thanks for any time and help, I appreciate you must be busy so any help would be gratefully appreciated.
Thanks
Tom
Again, probably spot versus futures. I use futures data. But these are just examples, so it doesn't matter if the levels are different.
DeleteHey all,
ReplyDeleteWould i be right in thinking that to calculate the annual standard deviation of Bitcoin i would multiply the daily standard deviation by 19? Because 19 is the square root of 365 and BTC trades every day?
As opposed to things like forex etc where you multiply the standard deviation by 16 because its the square root of the roughly 256 trading days in a year.
Thanks
yes
DeleteWould be much easier for me if you joined my ET thread https://www.elitetrader.com/et/threads/fully-automated-futures-trading.289589/ then it's easier for me to reply and others will also chip in if I'm busy.
DeleteThanks Rob, i have just joined yesterday and posted but it says my post is waiting to be accepted by the moderator
Delete