Backtest

Replays the forecast models through up to 10 years of real history, as if they'd been running all along: each day they see only the prices up to that day. Then it checks whether following their top picks would have beaten simply buying the S&P 500, after every trading cost.