Welcome to NexusFi: the best trading community on the planet, with over 150,000 members Sign Up Now for Free
Genuine reviews from real traders, not fake reviews from stealth vendors
Quality education from leading professional traders
We are a friendly, helpful, and positive community
We do not tolerate rude behavior, trolling, or vendors advertising in posts
We are here to help, just let us know what you need
You'll need to register in order to view the content of the threads and start contributing to our community. It's free for basic access, or support us by becoming an Elite Member -- see if you qualify for a discount below.
-- Big Mike, Site Administrator
(If you already have an account, login at the top of the page)
I felt the same way about @kevinkdog 's BWATS, but I am glad I bought. For this book, I got the Kindle version (~$30) which I can reference easily on my desktop.
Laurent wrote the book before adding any Python, so it stands alone. The title is misleading, because it it not just about short selling using Python. I find value without needing the Python, but it is there for once I learn and start building my systems in Python.
From the preface:
Who this book is for
This book is written by a practitioner for practitioners. It is for advanced to expert market participants. Even if you have never coded a line in Python, this book is still for you. It was originally written without the source code. This later addition is meant to help readers implement the concepts in real life. If you are an experienced coder but new to the markets, you will pick up concepts that will help you on your journey. You may however want to supplement your market education with further reading.
Even if you choose never to sell short, this book is still for you. The tools and techniques developed for the short side are built to withstand extreme conditions.*If you can survive the arid environment of the short side, imagine how you will thrive on the long side. If you are in the long/short business, the question is not whether you should read this book or not. The real question is can you afford to not read this book. You may disagree with some ideas, but they will provoke thoughts and spark conversation. The ideas we originally resist are the ones that makes us grow, so welcome to the space beyond your comfort zone.
I have been doing some fascinating work on systems in ETF's and John Ehlers' RSI with Hann windowing (from January 2022 TASC magazine). Given my current limitations in trading, I need to develop systems that follow the strict rules set forth by my wife's employer.
Just some testing on 18 random instruments (9 ETF's and 9 equities) has resulted in some interesting results. I am not permitted to take short positions and I have 30 day minimum holding periods. Although I consider the short side just as important as the long side, it does not apply to me. My systems have always been long and short.
Equities and ETF's are biased to the long side, but I have been fixated on having a system that performs well both short and long. Anyhow, back to the Ehlers RSIH, as I call it, I unwittingly built a trend-following system by excluding shorts and using a Williams fractal as my exit.
Using @kevinkdog's Monte Carlo tool, here are the results of Monte Carlo analysis for a portfolio of 14 of the 18 instruments (4 did not pass walk-forward testing), for a period from Jan 1, 2010 to Dec 31, 2020:
Moving into a full year of 'incubation' on unseen data, here is the result of the portfolio for all of 2021 (closed trades only):
It looks like I have my first system for equities. I am looking at ETF's mainly because the ones I am using do not require preclearance.
Here is the shuffled trades analysis, 2010-current for the entire portfolio (closed trades only):
Anyhow, this has been quite enlightening, especially since I picked a random selection of instruments. I will share more as I keep working through various instruments.
This week has been interesting and have a few things to share regarding my algorithmic trading systems. I received an update from my wife's compliance department on trading rules and restrictions. The list of approved ETF's are now exempt from minimum holding periods and pre-approval, i.e. I am mostly free to trade, though still bound to long only positions (go figure).
With that in mind, I went immediately to some of the systems I have built and have passed our stringent requirements. Before I share what I have, I have to mention the benefits of the work I have done over the past couple years. Firstly, with all the work already done on a lot of my trading systems, particularly development and testing, I can take any of a few dozen systems and apply them to new timeframes and instruments.
Secondly, doing this work allows me to have an inventory of tradable systems. Even if I do not trade them, they are ready to pull off the shelf and start using when the opportunity arises. I think this is something that is not really talked about with respect to systems, where a lot of focus is on building that one great system, instead of several very good systems. Right now, I have several very good systems, with the tricky part being assembling them into a decent portfolio. It is a good problem to have.
Top Traders Unplugged posed this question in a recent podcast to the panel of guests: Would you rather trade 100 markets with 1 system, using 10 different timeframes, or would you rather trade 50 markets, using 3 different systems, using 5 different timeframes?
This is part 2 of a two-part podcast, which I recommend. Great stuff.
On to the work I have done this week.... I dusted off a mean-reversion system I built last year, which I tested only on a basket of equities, both long and short: Trading Idea #11 - Connor's RSI 2. I applied this to a subset of approved ETF's, and the results were good, just as they were with the equities. Here is the system on the symbol IVE (daily):
Since the system was built and tested, I simply built a portfolio using MultiCharts Portfolio Trader, ran a backtest from 2006 through 2020, long positions only. I kept the instruments that passed these two criteria: Profit Factor >= 1.5 and Adjusted Profit Factor (worst case) >= 1. With that set, I went through incubation, which covered the period from Jan 1, 2021 until today. Here are the results:
Here are some notes on this system:
High win rate for the system, as expected for mean-reversion
It did well on unseen data
Time in market, 31.8% over 16+ years
High correlation between some instruments (see below), probably making some of them redundant
The system makes money, but annual returns are low and may not be worth taking money away from other more profitable systems
Leverage may help boost returns, but there is a cost for that
Daily Correlation
There you have it. My next steps are to see where this may fit with the Hann system I mentioned in my earlier post. I will post some results for that system tomorrow or next week.
I know you can only trade long, but have you looked at the "short" ETFs (ones you go long of, but you are actually short)? I realize there is a decay issue with these for longer term holds, but these might have a place in your portfolio.
Great idea! The short (inverse) ETF's are not on the approved list, but we can still trade them. They require pre-approval and are subject to the longer holding periods. I have played around a little with the inverse ETF's, but have not had any success with them, though to be honest I have not spent enough time on them. I will post results here when I have had a chance to run them through my systems.
One additional note regarding the mean-reversion system: I did no optimization to the systems. The results were good despite that. I ran optimization for IVE, just to see what would happen, and my default parameters were the best. I found that interested (the system on equities is optimized).
@kevinkdog , the aforementioned podcast has a discussion that might interest you. One of the guests, I think it is Rich Brennan or Mark Rzepczenski, mentioned their method for testing a new system: parallel incubation. This is doing your traditional incubation period, for 1 year in his case, but also incubating in a small live account at the same time, then comparing the results to assure live execution matches expected results.
I think you have done some similar things (Renko experiments maybe?). Anyhow, I thought this idea was an interesting twist on what you teach.
I cannot remember if it was in the first part of the podcast or the second, but here is part one:
I was able to successfully take the system I have been working on through my entire trading system development process, trading a basket of 14 equities and ETF's. It is a dead simple process:
RSIH provides the entry, fractals provide the initial and trailing stop. This is a long-only system for two reasons:
I wonder if the entries are really that important for this type of system (go ahead, prepare your flaming arrows). It is a trend-following system, so I wonder if this type of system would be successful with any type of break-out entry. I suspect it would be.
There is one thing that I struggle with, though I am not alone in this: can I beat the S&P 500 (i.e., attain alpha)? With this system and instrument selection as a portfolio, the short answer is 'no'. My system name is SAT-027, which I compare against the SPY ETF, from Jan 1, 2010 through Jan 28, 2022:
(Note: all trading costs, including margin, are included in these numbers)
At first glance, buy and hold on the SPY easily beats this system against the system on SPY and the portfolio I have constructed. However, there is more to the story. Firstly, the max drawdown for the SPY during that period was 52%, which is difficult for any trader to stomach. The portfolio had a max drawdown of 36% marked-to-market (I trade this weekly), with a close-to-close max drawdown of 25.8%. This lower risk and equity curve smoothing may be more desirable to some traders.
In addition, buy and hold ties up capital 100% of the time and has additional margin costs associated, if I am leveraging, though lower transaction costs come into play. The portfolio had 311 entries, to one for just SPY buy and hold. The portfolio was in market 73% of the trading period.
My final thought here is the trading period: 2010 to early 2022. I would definitely see a much clearer picture going back further. The reason for not using a longer lookback period is that not all the instruments have enough data that far back, specifically ETF's that did not exist. The SPY results definitely look a lot different over 16+ years. This is definitely something to keep in mind, but I may make myself cuckoo over it if I think about it much longer.
There is a lot to unravel here and I could analyze this until the cows come home, but I will stop here. I publish the results tomorrow on my site (noon ET), so I will drop a link here for those interested.