Strategy Overview
The mean-reversion agent operates on a straightforward thesis: equities that become oversold (RSI below 30) tend to snap back, and equities that become overbought (RSI above 70) tend to cool off. The strategy runs across a 24-ticker universe of large-cap U.S. names spanning technology, financials, healthcare, consumer staples, and energy — the kind of liquid, widely-followed stocks where short-term sentiment extremes are most likely to revert rather than persist.
The logic is classic and well-studied. In range-bound or low-volatility regimes, RSI-based mean reversion has historically generated consistent edge. The question, as always, is whether that edge holds in live conditions.
Backtest Performance
Over 451 days and 38 paper trades, the strategy produced a 14.73% total return and a 7.98% CAGR, with a Sharpe ratio of 0.58 and a 70.6% win rate. Those headline numbers read reasonably well for a rules-based approach with no leverage and flat $1 fees per trade.
The max drawdown of 15.64% is the figure worth watching. For a strategy trading large-caps with RSI guardrails, a drawdown nearly equal to the total return suggests the edge is thinner than the win rate implies. High turnover — 879% over the backtest period — also indicates the strategy is recycling capital aggressively, which amplifies both opportunity and friction in live markets.
Walk-Forward Validation: Where the Concern Lives
The four-fold walk-forward breakdown tells a more nuanced story:
| Fold | Period | Return | Sharpe | Drawdown | Trades |
|---|---|---|---|---|---|
| 1 | Aug 2024 – Jan 2025 | +2.06% | 0.57 | 4.24% | 8 |
| 2 | Jan 2025 – Jul 2025 | +11.10% | 1.32 | 10.53% | 10 |
| 3 | Jul 2025 – Dec 2025 | +2.21% | 0.46 | 8.80% | 6 |
| 4 | Dec 2025 – May 2026 | −2.84% | −0.33 | 14.96% | 10 |
Fold 2 was the standout — a Sharpe of 1.32 and double-digit returns. But the most recent fold (Dec 2025 to May 2026) inverted: negative return, negative Sharpe, and a 14.96% drawdown. That fold is the out-of-sample window, and it's the one that matters most for forward-looking confidence.
The validation result is a formal fail. The Probabilistic Sharpe Ratio sits at 0.785 and the Deflated Sharpe Ratio at 0.304 — the latter adjusting for the fact that this is one of six tested configurations. A DSR below 0.5 means there's meaningful probability the observed edge is noise.
Recent Activity: The Market Isn't Offering Entries
Since June 1, the agent has run daily and executed zero trades. Cash remains at $7,569.25 against a portfolio value hovering just above $10,000. The last actual trade was a 21-share buy of WMT at $115.75 on May 31.
This quiet stretch isn't necessarily a flaw — it means the strategy is respecting its own rules and not forcing trades when no ticker crosses the RSI thresholds. In a trending or momentum-driven market, mean-reversion signals simply don't fire.
Risks and What to Watch
The core risk is regime sensitivity. Mean reversion thrives in sideways markets and suffers when trends persist — whether driven by macro catalysts, earnings cycles, or sector rotations. The recent fold's underperformance aligns with that: if the market has been trending rather than oscillating, RSI extremes become continuation signals, not reversal ones.
Additionally, the small trade count (38 over 451 days) means each fold's statistics carry wide confidence intervals. A single bad trade in fold 4 disproportionately shapes the out-of-sample verdict.
The strategy remains live and under observation. Before any capital allocation decision, the validation gate should clear — ideally with a positive out-of-sample fold in the next review window.