Do AI trading bots actually work? An honest answer from someone who builds them
I asked Claude to build 1,000 trading strategies. Every backtest looked brilliant. Then I corrected the subtle errors line by line, and all 1,000 lost their edge. Here's why that keeps happening, and what it actually takes to build a strategy that survives.
Published
|
This is some text inside of a div block.
Research
All
Why the best trading models refuse to learn
Every AI investing product now promises agents that continually learn and adjust. Strip away the marketing and that describes continuous re-fitting to recent data, which is the mechanism of overfitting. There is a right way to build adaptive models, and it is narrower than the pitch.
Published
|
This is some text inside of a div block.
Research
All
Judge the behavior, not the returns
In 2020, the most successful quant fund in history lost more than 20 percent and told its investors it would not be changing the models. The investors who left anyway converted a temporary drawdown into a permanent loss. Every systematic strategy has years like this. What decides your outcome is the yardstick you use to judge them.
