Thesis & Universe
The mean-reversion strategy operates on a deceptively simple premise: large-cap equities that become technically oversold tend to snap back. The agent buys when RSI drops below 30 and exits when RSI climbs above 70 — a classic contrarian signal applied to a 24-stock universe spanning mega-cap tech (AAPL, MSFT, NVDA), financials (JPM, BAC, V, MA), healthcare (JNJ, UNH, PFE, ABBV), consumer staples and discretionary (PG, KO, WMT, COST, MCD, NKE, HD), energy (XOM, CVX), and industrials (CAT, HON, DIS).
The breadth of the universe is a deliberate design choice: sector diversification means the agent isn't purely at the mercy of any one macro regime.
Backtest Performance
Over 451 days ending May 2026, the strategy returned 14.73% total (7.98% annualised), finishing with a final equity of $11,473 on a $10,000 starting balance. The win rate of 70.6% across 38 trades is notably high for a rules-based system, and total fees of $38 (one dollar per trade) are negligible.
The Sharpe ratio of 0.58 is modest — acceptable for a low-turnover momentum-agnostic approach, though well below the 1.0 threshold often used as a benchmark for live deployment. Max drawdown of 15.64% is the sharpest edge of concern; at roughly twice the annualised return, it implies recovery from a trough could take longer than the strategy's track record.
Turnover of 879% deserves a note: calculated against the full position history, this reflects the concentrated burst-and-wait nature of RSI trading rather than a genuinely high-churn book.
Walk-Forward Validation: A Mixed Verdict
The four-fold walk-forward results are where the picture becomes more complex.
| Fold | Period | Return | Sharpe | Max DD | 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 |
Three of four folds are profitable, and Fold 2 is genuinely impressive. But the most recent fold — the one that matters most for forward-looking decisions — is the weakest. The out-of-sample Sharpe of -0.33 and a return of -2.84% drove the validation gate to a failed status.
The Probabilistic Sharpe Ratio (PSR) of 0.785 and Deflated Sharpe Ratio (DSR) of 0.304 tell a consistent story: after adjusting for the number of trials (6) and the backtest's finite length, the odds that the observed edge is real rather than noise fall short of the threshold required for automatic deployment.
Recent Activity: Patience in a Neutral Market
The agent has run daily since June 17 without executing a single trade. Portfolio value has drifted between $10,030 and $10,083 on cash of $7,569 — the last position added was 21 shares of WMT at $115.75 on May 31. This inactivity is expected behaviour: RSI extremes on large-caps are relatively rare in a range-bound or trend-following tape, and the strategy correctly stays flat when no signal fires.
Takeaways
Strengths: High win rate, diversified universe, low trading costs, and two strong positive folds suggest the underlying signal has real statistical content in certain regimes.
Risks: A 15.64% max drawdown against a 7.98% CAGR is uncomfortable leverage on patience. The DSR of 0.30 is a reminder that six trials across a 451-day window isn't enough data to dismiss the possibility that Fold 2 carried the entire result. The strategy remains on watch — not retired, but not yet trusted with live capital.