Automated Trading Strategies: Q3 2023 Portfolio Update
Gaining clarity and focus on the forward test.
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 we share 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.
We’re on the hunt for the holy grail of automated trade strategy. There are over 70 strategies here on ATS. Each one provides a technique or a tactic for a market set up that can be (and has been) automated.
There are many different types of strategies. Some include a risk management plan; some only make long trades; some should only be traded with metal futures; while others should only be traded with equities. Each strategy has its own universe of variation as well. You can either download the strategy or recreate it based on the steps provided in the strategy description.
Every few months I like to climb the highest tree in search of a broader view of the HUNT. In particular, I like to review:
where we’ve came from,
where we are now; and,
where we’re going from here.
Then I’ll close with a bit about the economy and our next strategy: Strategy 73.
The goal of this review is to update you on any progress made since the last update. You can view past updates below:
When we first started, I published 10 strategies that barely had a profit factor greater than 1.05. Today, we have 70+ strategies, and the average profit factor is 2.19.
When we first started, we optimized every parameter; today, after being impaled by a few over-fitted unicorns, we only optimize those elements that are known to change market structure based on volume.
When we first started, I thought the simulation had no limitations. Today, we know that using the simulation as a tool to find viable strategies must carry with it an understanding that accuracy holds an inverse relationship with complexity, which requires additional resources in the form of time or computational power.
Is the holy grail of automated trading strategies a real thing?
I’m not sure, but we’re on the hunt. In hindsight, it was more of a ‘shoot for the moon’ kind of thing when we first started. What I’ve come to believe is that the holy grail is real, but it should be thought of as a moving target rather than a single strategy. So the true holy grail is in your ability to approach that moving target. It might take variations of the same strategy along a frequency wave or it might take multiple strategies at the same frequency. At the moment my vote is for a mix of the two; variations of several strategies along a frequency wave that changes based on historical volume levels.
There's no such thing as the goose that lays the golden egg forever.
Jim Simons, October 1996
The Relationship Between Market Volume & Frequency
A few of you have contacted me regarding the framework for selecting strategy variations in the portfolio. I thought I’d share the response with everyone.
Like the number of trading opportunities available for those on the hunt for market based trades, the market is also abundant in its anomalies. Market fluctuations, like a wind, can shift volume out of one market and into another. This has the effect of shifting market based trading opportunities into another ‘frequency’ (marked by indicator fail and divergence). In its wake, however, are a wave of anomalies. It follows that the number of strategies to catch this wave of anomalies is abundant as well.
So the challenge is coming up with a system to 1) identify market based trade opportunities, 2) identify anomalous activity, 3) identify the best frequency/dimension to take advantage of those opportunities.
In my mind, the ideal machine learning tool would have the ability to slide from one dimension to the other. Indeed, I believe the holy grail is a moving target along a frequency. From a market physics perspective that means frequency is a function of the amount of volume in the market.
The relationship may seem apparent, but what does it mean from the context of trading strategies? It means that, in general, high volume markets provide more profitable trading opportunities at higher frequencies than low volume markets. This is advantageous because higher frequencies tend to provide more accurate signals. As the market slides up and down a frequency scale, it finds a sweet spot. In general, the sweet spot for NQ is higher on the frequency scale than it is for RTY on the same strategy.
There are some strategies I’ve noticed that perform well across frequencies; that is, when optimized, the same data series provides the best solution across instruments. I believe this to be indicative of a ‘stronger’ strategy.
One way to tell you’re in the wrong frequency is when indicators stop working and/or diverge with price. In this way, the market is multi-dimensional and the indicators can be used to tell which dimension you’re on.
So, when we’re optimizing a strategy for the Q3 Performance Chart, we’re seeing which frequency was used to find the best trades over the last year. While the frequency may not stay there, it gives us a good place to start as an average.
Q3 Performance Chart
The system is always leaking, and we keep having to add water to keep it ahead of the game.
-Jim Simons
The current Q3 Performance Chart is based on optimizations from September 2022 to September 2023. Subscribers, click here for a link to the full version, which includes the instrument and data series to use along with any parameter notes.
What’s happening?
The portfolio made ~$2.5 million, had a profit factor of 2.19 and a return on min starting value of 421%. Min starting value is calculated by adding 25% (as a cushion of protection) to the maximum draw-down. In total, the portfolio made:
~9k trades (~34 trades per day and ~$290 per trade); compared to
~8k trades (~32 trades per day and ~$257 per trade) in Q2; compared to,
~24k trades (~94 trades per day and ~$219 per trade) in Q1.
We’re getting leaner.
There was a time when I viewed strategy variations that produced under 100 trades as being unreliable—the logic being, the more trades you have, the more you know how a strategy performs. Over time however, I’ve noticed that there’s another side to this coin. It is also possible for low trade strategies to be indicative of higher quality as long as the backtest is accurate.
So what? So instead of a portfolio with a profit factor of 1.48, we had a portfolio with a profit factor of 1.96 for the Q2 portfolio and 2.19 for the Q3 portfolio. Instead of relying on a high trade count to push net income (which generally lowers profit factor), we can use scale. As a rough estimate, if we use 3 contracts to trade the portfolio instead of the 1 being used in the model, our trade count would go back to what it was in Q1, but net profit would be ~$7.5 million instead of $2.5 million.
Where do we go from here…
It’s time to take the forward test to the next level. The forward test is tracked in the Mudder Report, which is generally published to paid subscribers every two weeks or so. The trades are being made on a simulated account, so the results are still hypothetical, but the data driving the simulation is live, which gets us closer to how these strategies will perform once traded on a live account.
This is usually where I tell you that I’ll be adding a few strategies to the forward test. However for Q3, instead of only looking at a handful of strategies, we’ll be conducting a forward test of the entire Q3 portfolio, including a few stocks.
I want to thank Xa for the hard work in pulling this together—I know it hasn’t been easy.
Next Steps & A Few Words About The Economy
I will publish the first Q3 Mudder Report (forward test) within the week to subscribers. I’ve got about 50 strategies in the queue that are still in draft form so I’m working on fine tuning the most relevant and/or time sensitive. Like I said, I’ll be ending the strategy formulation portion of this experiment when we reach 100 strategies.
The next strategy to be published is Strategy 73. Strategy 73 will be an update to the election year event study (tbp Oct. 1-2). 2024 is an election year, which is historically bullish for the market. Strategy 73 will look at some ways to take advantage of this pattern.
As the Fed attempts to dig its heels into the economic quicksand of 2% inflation during a period of high spending and debt monetzation, the market is still begging for lower rates. In last week’s FOMC press conference, at least 75% of the questions went like this: “The economy is doing well. Isn’t it time to lower rates again?” No doubt there are still a few wide-eyed analysts out there pushing the apocryphal belief: ‘the Fed will increase rates by year end’. These are the same analysts that said the Fed was going to raise rates in May. As I told you at the end of 2022, it’s not happening. The reality is that we’re on the precipice of what will likely be unprecedented rate inaction.
Another scenario, though admittedly left field, is that the Fed knows inflation will never normalize back to 2% and is using the situation as a way to attempt a massive consolidation of the banking sector.
Either way, election year spending is incoming and there’s a lot at stake. What will happen when that spending pushes the economy higher? Will the Fed acquiesce and raise the inflation benchmark to 3 or 4% or dig its heels in even more with higher rates? Either way, don’t be fooled by the pullback.
I’ll discuss this more in Strategy 73. I’ll also be publishing the backtest results for Strategy 72 on NASDAQ-100 stocks.
Housekeeping & Thank You
The things we are doing will not go away. We may have bad years, we may have a terrible year sometimes. But the principles we've discovered are valid.
Jim Simons, November 2000
I’m happy to say that over 50% of subscribers have been with ATS for more than one year. There are a few of you that have been with ATS for over two years. I take great pride in that metric, so I want to take a moment to say thank you to those of you that have continued to support this effort over the years. If that’s you, please feel free to reach out to me. I’d like to know more about your experience.
I also want to say thank you to Salvino Sardo for sharing ATS with others. Those of us on the hunt tend to be a secretive bunch, but I’ve learned a lot from sharing these strategies with all of you, so I want to reward those of you with a predilection for sharing by having a quarterly contest. The top three “shares” will receive one month free—this pertains to both paid and unpaid subscribers.
I will announce the winners on the next quarterly update at the end of December.
I’d like to start the contest by giving a free month to Salvino Sardo—please contact me at AutomatedTradingStrategies@protonmail.com and I’ll upgrade your subscription.
If you have any questions, comments or recommendations, please reach out by responding to this post or emailing at AutomatedTradingStrategies@protonmail.com
Click here for FAQs.