This was developed with the help of Claude and it did an astounding job
After clicking the link Claude can help you modify the “Artifact” to your liking
The app takes into account the magnitude of improvement. So you can find the IB algo that improved that average fill price compared to a benchmark (like the opening price). Not just the frequency of improved fills.
State of the art Bandit algorithms. Choices include Upper Confidence Bound 1 and Gaussian Thompson sampling.
Beautiful demo
TLDR: After your own modifications (which is easy to do with Claude) this might be a serious contender for the best way to keep track of your trades. For anyone.
Awesome, First impression is that it looks like the best approach I’ve ever seen for evaluating effects of different trading methods that include slippage. Did require a Claude subscription and log in to view. Unfortunately for many of us using Fidelity or Schwab we would need to modify for their trade format data and their respected algorithms.
As you noted a free Claude subscription is needed. For those not aware already Claude will still customize it for you at the free tier level and you should not have trouble with limits
If anyone is interested, I am happy to share the code and you can use Google AI Studio which does a decent job. Google AI Studio limits the number of apps that can be published with a link so not the best for me to publish there. I could post a GitHub too if it would help anyone.