Hey,
I think it's a good time to launch one of my side projects catchfire.finance ! (It's been running for months, so "launch.") I started it a couple of years ago after my Graph Theory class with Professor Radu Marculescu @ UT Austin, where I did some research into transfer entropy graphs for equities. In addition, a lot of inspiration came from one of my Econ Professors @ Rose-Hulman, Dr. Jong Hun Kim, who talked about impulses and how economic shocks can travel through economies. (Also where I got my love for Corporate Finance and Macroeconomics)
Starting out, it was a fun way to get back into coding, since I'd spent so much time doing ASIC development. It was fun to see how far software development has come with open-source tooling and easy access to commercial-level software. Honestly, it made me very jealous, since much of this openness and support for getting something off the ground didn't really exist on the ASIC / hardware design side at the time. Since I started, though, the open-source hardware ecosystem has grown a great deal, and it's been great to see the work behind so many projects like OpenRoad. It's also been great to see programs like TinyTapeout, where you can print your own ASICs and make interesting projects that you couldn't just years ago. This project was a fun way to keep learning about market modeling, machine learning, and feature engineering which all have been interests of mine.
Anyway, back to the point. I went through a couple of revisions of the site and eventually got to the style it has today. I would say "pre-AI" I essentially got the main "Stock" page up and running, with the backend able to run and update this graph, where it could provide interesting graphics. But "post-AI" I was able to flesh out so many of the smaller things that I think were needed to get to a point I would say is worth showing people. Adding the Forex and Crypto markets was basically trivial, even with older models popping up a year ago. With newer models, I could hone down the underlying algorithms some more. Then, with the recent models actually able to add tooltips and the volatility index without the page layout going insane. Frankly, now it's at the point where I just have a Claude Code window open, managing everything from an AWS machine. So I would say the era of this being my hand-coded side project is over, whether it be toiling away in Google Colab Sheets running models or looking up React docs to try and get the graph not to refresh every millisecond.
Now we get to the launch. One reason I was hesitant about launching it is whether this was just a neat data visualization tool or could actually provide some value to someone. I think it would still be worthwhile if it was just a visualization, but I wanted to strive to a point where it actually could show some useful information. In the background, I've been developing a signal that helps give a volatility index of a market or a certain stock, which proves useful. Now maybe in the future I will give more detail about how exactly to get that signal from this information, but it's proving to have somewhat good alpha and is different from typical signals you can get from a financial API. Again, I'm not saying you should invest anything based on this data, and it's not investment advice. But I'm at least convinced it's more than a pretty picture at this point.
Also, look into my other articles about these transfer entropy graphs, and maybe you can gather what exactly I did:
Detecting Market Regimes: https://www.imjustageek.com/blog/looking-at-structural-breaks-in-the-crypto-market-after-trumps-tariff-announcement
Model Construction and Feature Engineering: https://www.imjustageek.com/blog/revisit-analyzing-the-stock-market-with-transfer-entropy-graphs
Underlying Ideas: https://www.imjustageek.com/blog/using-causation-networks-to-predict-future-network-robustness
Anyways, hope you all enjoy catchfire.finance!
Also, a shout-out to Jonathan Earles, with whom I made the very first version of this site:
