Which Strategy Is The Best?
ATS Forward Test Update
Important: There is no guarantee that these strategies will have the same performance in the future. I use backtests to compare historical strategy performance. Backtests are based on historical data, not real-time data so the results shared are hypothetical, not real. There are no guarantees that this performance will continue in the future. Trading futures is extremely risky. If you trade futures live, be prepared to lose your entire account. I recommend using these strategies in simulated trading until you/we find the holy grail of trade strategy.
By far the most common question I get is:
Which automated trading strategy is the best?
The question used to irritate me because it seemed lazy and this pursuit is not for lazy people. It requires a kind of obsessive diligence that you either have or don’t have for a thing. And, as much as I like to grow my portfolio with subscription revenue, I have no interest in helping anyone lose money—karma is a nasty business. That said, it’s a valid question. It’s the kind of question I might ask someone with a newsletter called Automated Trading Strategies, especially if I’d just purchased a subscription.
I also know that many of you have other things that require your time and attention (read: family, jobs, the general demands of life), and while you would spend more time on this if you could, it’s hard to carve out that time. So, while I still advise creating your own forward test to answer this question, I also want to do what I can to shorten your learning curve. That is the goal of this post.
The ATS Forward Test
ATS started with a simple nine strategy backtest published in January of 2021. Those initial backtests led to the development of better strategies, but copious backtests get you nowhere. So in January of 2023, the best performing strategies in the backtest were used to create the ATS Forward Test.
The ATS Forward Test is based on live data rather than historical data. Yes, the forward test takes longer to run because it’s in real-time, but the results are more reliable. That does not mean that the forward test is indicative of future results, only that it is a better model of how a strategy performs over a given test period because you aren’t relying on a simulation to accurately portray market data. This becomes particularly problematic when creating bar formations. The more complex the calculation, the more opportunities for error—good luck getting accurate backtest results with Renko bars.
Another issue is the frequency of bar formation. In general, the greater the frequency, the more calculation is involved in the simulation, which increases backtest error. Volatile instruments amplify this effect.
The best way to avoid these issues is to only use minute (or day) data on a time series of 60 minutes or higher. Keep it simple. That said, I am of the belief that more anomalies exist at higher frequencies (1, 2, 3 minute) as opposed to lower frequencies (45 min, 60 min, 270 min). I also believe that over the next 5 years, as the “rendering” of a market gets easier for the simulation to handle, it will create an opportunity to refine the hunt. I don’t know if any of you have tried the most recent version of ChatGPT, but it’s pretty remarkable. If advances in simulation technology follow suit, the next five years will be very interesting.
There’s also evidence to suggest that high frequency anomalies exist with intraday momentum or breakout strategies. I’m working with a few people on this strategy now. In particular, I’m trying to use this article (thank you Klemen) as a foundation to create a momentum strategy based on high frequency, intraday trades.
To that end, and before I get into some of the findings from the forward test, I want to make a quick detour to request help from those of you with the time or inclination. If you’re interested in helping to develop this strategy let me know what you think of this article. What are the main takeaways and what kind of strategy does it inspire you to create? Send an email or direct message to: AutomatedTradingStrategies@protonmail.com
I promise to respond to all submissions. Please be brief. I’m looking for people that can add value to a question. To be clear, the question is: how do we create the most profitable intraday momentum strategy based on the lessons learned/best practices presented in this article?
Now, let’s get into the results and lessons learned from the forward test.
The forward test I’m about to share with you was created from the Q1 and Q2 2024 Backtest. It shows forward test results by both instrument and contract. It also shows the quantity of trades, profit and profit per trade.
Getting back to the original question: ‘Which automated trading strategy is the best?’, I look for strategies with the highest profit per trade, at least five trades per contract and a confirmation of some kind. An example of a confirmation is a strategy that is profitable on more than one instrument or a strategy that is profitable on consecutive contracts (i.e., 3-24, 6-24). I’ll give some examples of what I mean in a moment.
With that said, here’s a link to ATS forward test results from 1/1/23 to 5/24/24: