Mudder Report: ATS Q4 Forward Test Update
Reflecting and mirroring our way to the holy grail.
Important: There is no guarantee that ATS strategies will have the same performance in the future. I use backtests and forward tests to compare historical strategy performance. Backtests are based on historical data, not real-time data so the results shared are hypothetical, not real. Forward tests are based on live data, however, they use a simulated account. Any success I have with live trading is untypical. Trading futures is extremely risky. You should only use risk capital to fund live futures accounts and if you do trade live, be prepared to lose your entire account. There are no guarantees that any performance you see here will continue in the future (good or bad)—that’s what makes the hunt for the holy grail so difficult. This is why the best way to trade is with a simulated account on live data. I recommend using ATS strategies in simulated trading until you/we find the holy grail of trade strategy.
I hope everyone had a great summer. I had some truly inspiring conversations, listened to some great live music, doubled my bird count, and got lots of sun on the beach, but now I’m back in the office and ready for what promises to be an exciting quarter. I hope you are too.
Okay, let’s get started.
Due to a number of cyclical patterns that you can read about here, I knew 2024 was going to be a wild ride. I’ll provide a quick NQ technical update at the end of this post to show where I think the market’s going for the rest of 2024. I’ll also provide you with a first draft of the ATS Q4 Portfolio, which is the best yet. I know I say that every update, but this is truly a special quarter. I’m also going to provide a quick update on where we are with the new system build, which is based on the efficiency of communication with the most accurate and capable LLM.
Why am I structuring our quest around an LLM? There can be only one reason. I believe it can help us to create the holy grail of automated trading strategy. In my opinion, there is no better team member.
I want to take a few minutes to revisit what Ray Kurzweil said about LLM’s during a Joe Rogan interview in March of this year. He said that LLMs should actually be thought of as much more. Instead of large language models, Kurzweil said we should be calling them large event models, because “they do all kinds of things”. You can listen to the conversation yourself below starting at ~21 minutes below.
When I asked the LLM Claude to provide a more descriptive ‘term’ for its technology, it said that the best term would depend on which capabilities the AI is being used for. Fair enough.
Kurzweil intends to use LLMs to achieve immortality. I intend to use it to find the holy grail of automated trade strategy. The challenge with LLMs is accuracy, but with the help of techniques like reflecting and mirroring, the gap is quickly closing.
AI Sapience: "Fake it till you make it"
Reflection and mirroring in the context of AI and LLMs are related, but distinct concepts. Reflection is about AI's ability to analyze and think about its own processes and outputs, while mirroring refers to AI's ability to emulate or copy specific behaviors, styles, or patterns. This could be in language use or problem-solving approaches.
Just as a chef might adjust a recipe based on previous outcomes and feedback, a reflecting prompt may allow an LLM to fine-tune its trading parameters based on past results. Likewise, similar to how a skilled dancer mirrors their partner's movements, a mirroring prompt may allow an LLM to synchronize its actions with the rhythm of the market. Put together, these two algorithms make a powerful ensemble, especially as a prediction model.