Built a Forex Tool With AI Because I Can’t Code

Where’s that meme…

Ah there it is!

As you may or may not know I am a bit of a currency strength and weakness nerd and I have been trying to get AI to build me currency strength meters in order to develop my understanding as well as open myself up to new trading opportunities.

We have already built a macro currency strength meter which uses such things as:

• GDP

• Interest rates

• Real yields 

• Core CPI

But this time I wanted to explore something a little different.

So I opened my old pal chatGPT again and got to work. 

1. Theory

I wanted to see if we could build a currency strength meter based on economic surprises. 

What do I mean by this? 

The actual vs the forecast. 

I have kind of done this already but not for all the G10 currencies. You may be familiar with my fundamental tracker which tracks all the releases of the US and UK data, takes into consideration the actual data vs the forecast to see if the averages are wrong. 

In my theory it could appear that the data is not as bad as first thought so the market may begin to reprice their expectations. 

For example if the US retail sales comes in at 0.1% beating forecasts of 0%, then it’s better than expected. If we then have multiple of these key data points coming in better than expected, could this then lead into a positive outcome for the currency? 

2. Sentiment Indicator ONLY

When I talk about building all these indicators, I want to add a caveat. These are only to be used for added layers of confluence and never as a sole excuse to trade the financial markets. We also need to test this going forward for a long period of time in order to see if it actually is useful as backtests can only do so much. 

I say this because the data could come out better than expected but still worse than the previous reading, so in theory the data could still be coming in lower. 

But still, this could be the making of an interesting sentiment reading indicator. As markets may be a little more upbeat on the fact that things aren’t as bad as it seems. 

If I know anything about markets this year, it's that they are being driven by sentiment over macro right now. 

3. What does it look like?

So I did add another little piece into it which was the interest rate differentials, which is probably why we are seeing the Japanese Yen as the weaker currency, as the rate differences favor the USD in this. 

To caveat that I created two one with the interest rate differentials and one without just in case the data was being skewed. 

Here’s how they look. 

As you can see it changes drastically depending on the inputs we use. I like using interest rate differentials when looking for flows, but it depends on the narrative as well. Is a central bank hiking or is it cutting, then we need context of why. 

Again, this is why I would use this as a bit of a sentiment reading more than anything else.

4. AI tools should be used as such

I have been using AI to build these tools as I am an amateur when it comes to coding. I have done some bits here and there, but I am far from being well educated. 

This is where AI for retail traders could really shine.

If we know what we want to ask and we know how the outcome can impact markets, then we can get the AI to do everything in between. 

Like I said, I have been monitoring the surprises and building them into a manual tool, but this could now halve my time and produce the outcome for multiple markets not just two. 

Final Thoughts

What excites me the most is the potential to use AI to simplify, refine and create different versions of ideas. The market is always changing and as traders we should too. We should not just stand still, we just continue to understand and learn. 

So if you have an idea, give it a go. Just be aware it won’t change things overnight for you, but it could change the way you view or understand the market. 

See you in the lab again soon.