Claude coding update
For anyone interested this is beyond amazing. I don’t even check code anymore. I collaborate, define technical specifics, and run the tests. My argument for why this was an epic fail for agentic ai was lack of determinism, but I’ve become so confident in the output I never even look at the code. It does more or less exactly what I tell it to do.
That tells me the wider agentic approach with ai work flows has legs.
I also realized my backtesting engine is effectively a stock screener after all a backtest identifies trades based on characteristics and a screener identifies stocks by characteristics.
I haven’t wired in fundamental data to my pipeline, but when I do I will likely have a stock screener more powerful than literally anything I could get free or even pay for
I’ll have it in about an hour, and it will be free.
Its insane