An AI hedge fund is a fund whose trading decisions come primarily from machine-learning models rather than from human judgement or from hand-built statistical rules. The models learn patterns from data instead of being told what to look for.
How it differs from a quant fund
Every AI fund is a quant fund; not every quant fund is an AI fund. Traditional quant strategies encode a human hypothesis — value stocks outperform, trends persist — and test it. Machine learning starts from the data and finds the hypothesis, which can surface patterns no one thought to look for.
The cost is interpretability. A factor model can tell you why it bought. A neural network often can’t.
What the models do
Predict short-horizon returns from features — price history, order flow, volatility, cross-asset signals. Classify market regimes. Optimise execution. Detect when a strategy’s own predictions have stopped working.
The frontier is using models to generate strategy candidates — automated research — and then testing them the way a human researcher would.
Where the label gets misused
Any fund that uses a computer now calls itself AI. The test is whether models are making decisions that a rule couldn’t, and whether the fund can show that the models add return over simpler approaches. Most can’t, and don’t try to.
Why it matters for individuals
The methods are no longer exclusive. What was exclusive was the capital and the infrastructure, and on-chain venues have made the second one small.