Can AI Really Build a Macro Trading Tool? Here's What Happened (Part 2)

AI Helped Me Finish My New Macro Strength Meter. Here's 4 things I learned.

On April 27th, I set out to see if AI could help me build a macro currency strength meter. The first version wasn't very fancy, it mainly discussed indicators I was already familiar with such as interest rate differentials, GDP differentials and some sentiment analysis. But I felt that it was very surface level and I wanted to dive deeper. 

So I did.

I wanted to share what I had learned through working with AI to build the tool and what it sparked in me. 

1. AI is a tool, not the solution. 

I quickly learned that AI was not going to re-invent the wheel here, and I expected it to (not really, but you get what I mean). I thought I would be able to tell it some simple prompts like 'what's the best way to identify strength and weakness of currencies' and it spit out some solution that would change my edge forever. It doesn't. Instead I started to ask it questions. Such as 'if you were a hedge fund manager, and you had to identify the strength and weakness of currencies, what would you do and why?'. 

This helped identify a couple of things for me:

- You need to ask the correct prompts.

- You must vet what the AI tells you.

- If you want, you can ask it deeper questions.

2. Fundamentals matter.

When I asked the initial prompt about being a hedge fund manager, the AI suggested looking at a few fundamental drivers which include:

- Interest rates

- 5Y breakeven inflation

- 2y swap rate

- CORE CPI y/y

- PMI Composite

- GDP y/y

- Current Account %GDP

There was nothing about technical, and I am not one to go against them, but when we want to understand the WHY behind the market, I think fundamentals and narratives drive the markets. Technical analysis helps us with entries and risk management.

3. Backtesting is limited.

I found that once the tool was created, it lacked the backtesting that comes with it. This was down to the data we are using, the 5Y breakeven inflation rate is difficult to find, so it's not something that the AI can just draw on, I had to add this manually. So the hard work now begins to track and monitor it from now on to see if it works to spot trading opportunities. 

This week it highlighted the following:

Weak currencies:

- CAD, EUR, SEK

Strong currencies:

- NOK, AUD, GBP and USD.

My goal is to now add a history table, monitor week to week progress, the currency pairs it suggests and how they have performed. Hopefully is six months time we may be on to something. 

4. It's isn't just about the output, it's about the process too.

The real benefit isn't the final tool. I have no idea if it works yet. It was learning to think about macro data more systematically. I had long discussions with the AI about why it included each macro data point. Why it would be meaningful. I then went away and researched each point. Honestly it really helped me gain a deeper understanding of some points. 

So all in all, the macro strength meter is here. Or at least a version of it is. I will continue to forward test this and give you some updates on the way, in terms of its performance. 

Make sure you join our mailing list to be the first to know if it works or if it sucks.