Verify Panel Test

How this result is verified

Computed by the MarketHeist engine v1.0.0 from yahoo total-return-adjusted data — 80 bars over 2019-01-01 to 2019-03-21.

Reproduce: config hash b109f8d738a311ea. POST /api/backtest with ticker=SPY, frequency=1d, indicator_id=rsi, and the same config; engine 1.0.0 over 2019-01-01…2019-03-21 reproduces these numbers.

Caveats

• Single in-sample run — not walk-forward or out-of-sample validated. In-sample metrics can overstate real performance; treat as a starting point, not a proof.
• No transaction costs modeled (0 bps). Real-world returns will be lower, more so for high-turnover strategies.
• ⚠ Only ~0.3 years of data in the measured window. Metrics are statistically thin and sensitive to the exact period.
• Probabilistic Sharpe ≈ 95% — marginal confidence (just under the 95% bar) that the true Sharpe > 0 after adjusting for sample length and return shape. If this was selected from many trials, the true (deflated) confidence is lower still.