My AI trading agents disagree with each other constantly — turns out that might be the most useful part
Been building AegisProject (multi-agent trading system — technical, sentiment, and volatility agents each vote on a trade), and the most interesting part turned out to be where the agents disagree, not the final decision.
When they all agree, the trade is usually straightforward. The interesting cases are when they split — technical sees a clean breakout and says BUY, volatility looks at the same move and says it's too fast, sit out. Those disagreements seem to flag something real that a single-model system would just smooth over.
I checked whether trades with more agent disagreement actually turned out messier. Early signs say yes, but not nearly enough data to trust that yet.
Curious if anyone here has seen disagreement between models/agents actually mean something, or if it's usually just noise dressed up as insight.