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. This is for educational purposes only. I recommend using ATS strategies in simulated trading until you/we find the holy grail of trade strategy.
Affiliate Disclosure: I am an affiliate for Make.com and may receive remuneration for certain links on this post. If you have any questions, please contact Celan at AutomatedTradingStrategies@protonmail.com
Welcome to "The Agent Revolution". I don’t know about you, but this feels like a golden age, especially for new ideas and insights about everything from weather patterns to protein design. AI agents are revolutionizing our world. As Satya Nadella, the CEO of Microsoft, recently said in an interview with Varun Mayya, in the very near future you will be hired for both your work history AND your agentic workflows.
The podcast above is about how traders are using AI agents to transform workflow, data capture and analysis. From using satellites to monitor oil reserves to using social media for sentiment analysis, it has become impossible to have a serious discussion about automated trading systems without including agentic workflows.
What makes agents so special?
They are not bound: AI agents are not bound or restricted to any one database, business logic, platform or application. With the right authorizations, they can write code to navigate these worlds in real-time.
They can analyze and integrate large amounts of data: AI agents can analyze large amounts of data based on a specific goal. That’s the nature of almost all agents. They grab info, send it to an LLM for analysis, and then send that output somewhere else (an email, document or another LLM) possibly for human review or use.
Put together, this allows for the creation of intelligent digital assistants and they are joining the workforce en masse. You can create them yourself (this podcast gives a few ideas and applications I’m personally using), or you can hire someone else to create them for you. I’m sure there’s even an agent that’s designed to create other agents. I’ll talk about agent exchanges (Fetch.ai and Solana) and investing in agents like AI16z (shout out to
for that insight) in future podcasts.Imagine—a team of intelligent digital assistants working around the clock—analyzing data, making decisions, and executing complex tasks with remarkable precision. That's the world of AI agents, and it's not science fiction. This is what’s happening today. For example, I am in the process of creating an agentic portfolio manager for ATS strategies that not only recommends strategies based on historical performance and upcoming events (earnings, fomc, etc), but learns from its own success rate. As Mayya and Nadella said, we are entering a time when your resume/CV will be assessed based not only on your own professional attributes, but by the abilities of the agents that work for you.
But the most important thing is to just get started. Start small. Here’s an example of a simple agentic workflow I created on Make.com. I also like LangChain and CrewAI, but Make.com is the easiest to get started with for non-coders. They also have a good tutorial.
The following agentic framework pulls in data from an RSS feed and then feeds that news to an Open AI module to read and analyze. The LLM prompt is something like: take the incoming articles from the RSS feed and rank them on a scale from 1 to 10 based on xyz; then summarize all ranked 3 or higher and send the output to a Google Sheet. Here’s what the workflow looks like.
For an example of other templates used on Make.com click here. It’s important to note, however, that in all likelihood these workflow applications won’t even be necessary as OpenAI, Anthropic, Gemini, Perplexity and a handful of other popular LLMs continue to add agentic capabilities to their tool set. OpenAI just released The Operator (an OpenAI agent that can use its own web browser to complete tasks for you). In Part 4, we’ll look at some trading platforms that are leading the way in this space.
So whether you're a trader or tech enthusiast, or simply curious about how AI is shaping our future as traders, you're in the right place. Let's dive in and explore the world of AI agents together.
I hope you enjoyed Part 1 of this podcast series. In Part 2, we’ll have a look at additional frameworks and best practices, and then in Part 3 we’ll look at building strategies from the point of view of an agent. In particular, I want to know if it’s possible to put an intelligent agent within the command structure of our strategies.
These are truly exciting times!!
Stay tuned for these upcoming posts:
January 26 - Post: Complex Systems Theory In Trading Strategy Behavior
January 28 - Podcast 2 - Using Agentic Frameworks in Automated Trading Part 2
February 3 - Podcast 3 - Using Agentic Frameworks in Automated Trading Part 3
February 11 - Post: A Framework for LLM Based Automated Trading Information Systems
February 19 - Post: The Impact of A Digital Workforce On Assets & The Cost of Labor (tentative)
Additional Q1 2025 Content
Regular updates on Q1 2025 Forward Test
Strategies 89 & 90 progress (including ML/continuous learning)
Frequency optimization insights
Incubator portfolio (targeted mid-February)
A special announcement planned for March 5
Trading is a journey, not a destination.
If you have any questions, start with the FAQs and if you still have questions, feel free to reach out directly at AutomatedTradingStrategies@protonmail.com.
Share this post