What I Wish I’d Known About Scaling Before Building with AI

Slow backtests, blown memory, and the coding habits I wish I’d caught sooner.

This Model Beat FanDuel AND DraftKings on NFL Totals. Then the “Best” Bets Lost Money.

Two MLB Betting Models Made Money. That Was Not Enough.

Both models finished profitable in a corrected 2022–2025 backtest. That is a lead worth following, not proof of an edge.

I Deconstructed a Failed Strategy. Was Anything Worth Saving?

Why the Best Backtest Metric Might Be the Wrong Optimizer Target

A walkforward optimizer-objective test from Alpha Lab.

Last call: Forge Learn starts tomorrow at 11:00 AM ET

Before alpha, there must be trust

The first job of a serious research system is not to find alpha. It is to create the infrastructure where alpha can be trusted.

I haven't been absent. I've been building.

Not tutorials. Not notebooks. Real systems, built daily.

One question before I ship this

30 seconds. Decides what I build first.