Only strategies that survive honest testing graduate toward live capital.
Most trading hypotheses are false, and a research process that cannot demonstrate this about its own output is not a research process. Systematic validation is the gauntlet every Bountify hypothesis must survive before it comes anywhere near a capital decision. The stages are fixed, the criteria are registered in advance, and no strategy, however elegant its logic or distinguished its origin, is exempt from any of them.
Causal Backtests Without Hindsight
Validation begins in an event-time simulation built on point-in-time data. The engine sees only what a trader could have seen at each moment: no look-ahead, no restated figures, no fills that assume liquidity the market never offered. Transaction costs are modeled inside the simulation rather than deducted afterward, so the economics under test are the economics a live deployment would actually face, friction included.
Out of Sample, Walked Forward, Stressed
A strategy that performs on the data used to find it has proven nothing. Survivors face out-of-sample evaluation on untouched history, walk-forward analysis that re-fits and re-tests across successive market regimes, and transaction-cost stress tests that widen spreads and slippage until the edge either holds or breaks. Multiple-testing discipline runs throughout: because our AI research agents evaluate thousands of candidates, every performance statistic is deflated to account for the size of the search that produced it.
Pre-Registered Criteria and Paper Trading
Promotion thresholds are written down before results exist, which removes the oldest temptation in research: deciding what success means after seeing the outcome. Strategies that clear the statistical gauntlet then trade on paper against live markets, where latency, liquidity and behavior under real conditions are observed without capital at risk. Only after live paper results confirm the simulation does a capital decision arise, and that decision follows its own risk governance. The gauntlet is deliberately expensive, and most hypotheses die inside it. That mortality is the point: what emerges is not a promising backtest but a strategy whose behavior has been confirmed under the most hostile conditions we know how to construct.