When AI compresses alpha’s half-life, the edge is the system that outruns the decay.
LEAP 01 is a private roundtable for thirty senior systematic investors on the question every desk is now living: how to use AI to find alpha, test it honestly, and mutate it faster than the market arbitrages it away. Hosted by SimicX with Google DeepMind. No pitch. No recording. Just the conversation your competitors wish they were in.
Signal half-life has fallen from years to months. Compute alone made it worse.
As AI adoption crosses the majority of systematic capital, the same models converge on the same trades. Crowding accelerates decay; published estimates now put effective signal half-life near eighteen months, down from five-to-seven years. Throwing more GPUs at a homogeneous search only deepens the problem. The desks that compound through this are not the ones with the single best signal — they are the ones with the fastest, most disciplined loop from idea to validated, differentiated, in-production alpha.
AI that searches the whole space
Search across the six axes of alpha — forecast structure, frequency, methodology, asset class, data source, risk — at a scale no research team can cover by hand, surfacing candidates where you are not already crowded.
Validation that survives reality
Walk-forward, deflated Sharpe, probability-of-backtest-overfit, realistic costs. The discipline that separates a validated alpha from a backtest that got lucky — and kills the 90% that should never reach capital.
Alpha kept orthogonal to your book
Mutate and recombine surviving signals so each new one is deliberately orthogonal to what you already trade — the cheapest defence against crowding and the fastest path to capacity.
Small enough to be honest. Senior enough to matter.
Thirty seats, one table. The room is built from systematic PMs, CIOs and heads of quant across hedge funds, multi-strategy platforms, asset managers and proprietary desks — primarily equities and multi-asset. Everything runs under the Chatham House Rule, so the people who never share, can.
Introductions over coffee. Badges show first names only.
SimicX and Google DeepMind on finding, testing and mutating alpha in a homogenising market.
Moderated, off-the-record discussion driven by the room's own questions. No slides.
Good food to grab, and the candid conversations that only happen once the day winds down.
Held under the Chatham House Rule.
“Participants are free to use the information received, but neither the identity nor the affiliation of any speaker or participant may be revealed.”
Invitation-only. Every request is reviewed.
Three short steps. Your answers save as you go, so you can finish later from this device. Submitting is a request, not a confirmation — we will be in touch.