Stock Performance

We’ve been making moves towards selecting fewer stocks per day. The default is now five, as opposed to six (however there are still days where six or seven are being selected). Being able to remove one stock and have that improve the results compared to last month is a great direction to be moving in. Five is perhaps a little too many and we will look at cutting one more off before the end of the year.

Tuesdays have historically been a poor day for the algorithm and we’ve been making some progress on that front. Here’s a breakdown per day:

Surprisingly it was the best day of the month, though I think ICCT and its 275% increase is doing a lot of heavy lifting there.

AI Performance

For context, here is the performance on July 2nd when we first started:

💡 A HALF READ is a correct entry leading to the first price target, a FULL READ is a correct entry leading to a the second price target. We then expand the stocks performance by X% to see how accurate the predictions were.

At the time, I described it as “exciting, full of potential but not yet usable”. So how did it perform in August?

The most obvious thing to mention is that has been a dramatic increase to the entry price. Another thing to point out is that although the HALF READ and FULL READ are a couple points above where they used to be, they’re more accurate predictions overall as seen by the expanded ranges. When we get those dialled in, that’s going to be a pretty massive boost to usability.

We generate the analysis ~30 minutes before the market opens with the goal of predicting price movements during market hours. Here’s the results if we start measuring immediately after the analysis is generated:

That’s a pretty sizeable bump in performance. I’ve mentioned a few times that we’re working on a WhatsApp* integration that will allow you to analyse any stocks that you want, and it’s good to know that it should be somewhat usable off the bat.

* More will be available, too!

However, that’s looking at the things which can be measured objectively. We generate two things: an analysis of the stock and price predictions, and the analysis is the part that most impresses me right now.

Here are some examples from the past few days:

The iCoreConnect Inc. (ICCT) stock has been on a rollercoaster ride recently, with trading halted due to share conversion issues and then resuming, causing some investor uncertainty. However, the stock showed resilience with a strong recovery of 70.4% after a 20% dip.

VinFast Auto Ltd. (VFS) has been on a remarkable run, with a 312% surge in its stock price over the past week. The company's expansion into the U.S. market and recent public listing have fueled this growth. Today's pre-market trading shows a continuation of this upward trend, with the last candle closing at $79.7, up from an opening price of $75.02.

VinFast Auto Ltd. (VFS) has been on a rollercoaster ride since its recent public debut, with gains of up to 94% in a single day. The company's impressive range ratings for its new electric SUV and a significant move by one of its suppliers have fueled this surge. However, the market's efficiency in relation to VinFast's stock price has been questioned by some analysts, suggesting that the price may not accurately reflect the company's value. This volatility and speculation could lead to further significant price movements in the future. The broader market sentiment, as indicated by SPY and QQQ overviews, is currently experiencing volatility due to market anticipation of Federal Reserve Chair Jerome Powell's speech and other economic data.

It still shocks me that these are AI generated. Developing this has given me that feeling of “oh, we’re in the future” on a daily basis.

I’ve been stressing how important it is for the analysis to show why a stock is interesting and in contention for that day. This has been a pretty big step towards that but I do accept that it’s not yet where we want it to be (though we have some ideas on how to get there!).

Recent Updates

• Selecting 5 stocks per day as opposed to 6 to weed out under performers
• Redesigned how we process company news which allows us to consume a lot more information
• Confidence percentages on both price predictions (this was a subscriber suggestion, thank you so much for it!)
• AI has more understanding of the wider market movements and how this may impact the stock (still a work in progress)
• Improving readability for stocks as opposed to a data dump of technical analysis
• Including company financial data into the analysis to introduce the ability to find undervalued stocks

Next Steps

There will be a version of the WhatsApp integration rolled out in the next few weeks. It may have a limited test run as we want to make sure that we’re building on the correct foundations. Feetr currently targets day trading but it’s going to be a whole lot more than that and we need to make sure we aren’t giving ourselves headaches later on.

We will also continue working to improve the AI analysis. The most pressing matter is to get it to understand the direction that the stock will take. That will, I believe, massively improve the success rate of the price predictions.

---

If this is something that you wish to support, you can subscribe over at feetr.io (only $8.99 per month!) or you can follow along for free at @feetr_io