Strategy Backtest Results
33-year comprehensive analysis (1993-2026) with zero-fee assumptions
🎯 Executive Summary
🚀 Best Absolute Returns
⚖️ Best Risk-Adjusted
💰 Best Tax Efficiency
Key Findings
- Reddit strategies dominate in absolute returns (3-27x advantage) with ultra-low turnover
- Tolerance bands reduce trades 66% (714 → 243) but lower returns 28%
- Tax efficiency is critical: High-turnover strategies lose ~75% of returns in taxable accounts
- ⚠️ Regime-dependent performance: All strategies excelled in 2011-2026 bull market but struggled 1993-2010
- reddit_200sma_tqqq LOST MONEY 1993-2010 (-0.8% CAGR) but dominated 2011-2026 (+35.8% CAGR)
- ETF liquidity matters: Stick to TQQQ/UPRO/TMF for live trading
📊 Full 33-Year Performance
All strategies tested with zero transaction costs (realistic for modern brokers)
| Strategy | Total Return | $10k → | CAGR | Sharpe | Max DD | Trades | Trades/Year |
|---|---|---|---|---|---|---|---|
| regime defensive rotation 3x Highest Return | 54792.6% | $5,489,258 | 34.5% | 0.86 | -63.9% | 714 | 21.6 |
| reddit 200sma tqqq Reddit Strategy | 26960.4% | $2,706,036 | 30.1% | 0.69 | -66.0% | 23 | 0.7 |
| regime defensive rotation with bands 3x | 22905.0% | $2,300,499 | 29.1% | 0.74 | -72.3% | 243 | 7.4 |
| regime defensive rotation 2x | 9784.9% | $988,488 | 24.1% | 0.90 | -47.3% | 714 | 21.6 |
| reddit 200sma spy Reddit Strategy | 8808.9% | $890,895 | 23.5% | 0.69 | -58.4% | 17 | 0.5 |
| regime defensive rotation with bands 2x | 5370.3% | $547,032 | 20.7% | 0.78 | -55.8% | 243 | 7.4 |
| top2 relative strength rotation Best Sharpe | 1638.9% | $173,890 | 14.4% | 0.93 | -26.6% | 1706 | 51.7 |
| buy and hold | 1106.3% | $120,629 | 8.8% | 0.43 | -59.3% | 1 | 0.0 |
| regime defensive rotation | 1106.0% | $120,600 | 12.4% | 0.93 | -26.0% | 714 | 21.6 |
| dual momentum rotation | 949.3% | $104,933 | 11.7% | 0.56 | -51.4% | 998 | 30.2 |
| regime defensive rotation with bands | 791.8% | $89,176 | 10.8% | 0.82 | -32.2% | 243 | 7.4 |
⚖️ Leverage Comparison: Timing vs Buy & Hold
SPY 3x: Timing vs Buy & Hold
QQQ 3x: Timing vs Buy & Hold
SPY: Leverage Scaling
QQQ: Leverage Scaling
🎯 Key Takeaway
Timing adds significant value at 3x leverage:
- SPY: +58% better returns, 87% better Sharpe, avoids -95% catastrophic drawdown
- QQQ: -8% lower returns BUT 45% better Sharpe and avoids -94% catastrophic drawdown
- Leverage scales returns but DESTROYS Sharpe ratio without timing
- Only 17-23 trades over 33 years = ultra-low tax drag even in taxable accounts
🔬 Additional Strategy Comparisons
Reddit 200 SMA vs Regime Rotation
Verdict: Reddit wins for taxable accounts (tax efficiency). Regime wins for IRA accounts (better risk-adjusted returns).
Tolerance Bands Impact
Verdict: Bands reduce trades 66% but also reduce returns 28%. Trade-off isn't worth it for most investors.
Leverage Impact (Regime Defensive)
Verdict: 2x leverage offers best risk/reward balance. 3x is extreme but works for aggressive risk tolerance.
⚠️ Regime Sensitivity Analysis
Performance across different market periods: 1993-2010 vs 2011-2026
reddit_200sma_spy
reddit_200sma_tqqq
regime_defensive_2x
Critical Understanding: These strategies use fixed parameters (200 SMA, 50 SMA, etc.) - not optimized on historical data. The improvement from 1993-2010 to 2011-2026 is primarily due to more favorable market conditions, not validation against overfitting.
1993-2010: Dot-com crash (-50%), 9/11, GFC (-56%), slow recoveries = Difficult environment
2011-2026: Longest bull market, QE infinity, fast recoveries, AI boom = Easy environment
The Real Question: Will the future resemble the 2011-2026 bull market period, or return to 1993-2010 style severe bear markets? Your answer determines which strategies to use.
💡 Strategy Recommendations by Account Type
🏦 Taxable Accounts
Why: Ultra-low turnover (17 trades in 33 years) means minimal tax drag. Long holding periods qualify for LTCG rates.
Alternative: reddit_200sma_tqqq for more aggressive approach (+26,860% but higher volatility).
📈 IRA/Roth Accounts
Why: Best risk-adjusted returns (high Sharpe). No tax drag in IRA means frequent trading doesn't hurt performance.
Alternative: regime_defensive_3x for aggressive growth (+54,693% but -64% drawdown).
🛡️ Conservative / Near Retirement
Why: Lowest drawdowns (-26% vs -47%+ for leveraged). Still beats buy-and-hold SPY on risk-adjusted basis.
Note: Matches SPY absolute returns but with better Sharpe and lower volatility.
⚠️ Critical Caveats
1. Regime Dependency: These strategies excelled during the 2011-2026 bull market period (QE, fast recoveries, AI boom). The reddit_200sma_tqqq LOST MONEY during 1993-2010 (dot-com crash, GFC). Future performance depends on whether markets resemble 2011-2026 or 1993-2010.
2. Paper Trade First: All strategies should be paper traded for 6-12 months before deploying real capital. This validates that backtest assumptions (execution quality, slippage, regime detection) match reality and that the bull market continues.
📋 Methodology
Test Parameters
- Time period: 33 years (1993-2026), daily EOD data
- Transaction costs: 0 bps (realistic for zero-fee brokers with limit orders)
- Tickers: SPY, QQQ, GLD, TLT, XLE, XLU (all ETFs to avoid survivorship bias)
- Leverage: Simulated using 2x/3x leveraged ETFs (SSO, UPRO, TQQQ, etc.)
- Validation: Walk-forward testing (train 1993-2010, test 2011-2026)
Strategy Descriptions
- reddit_200sma_spy / reddit_200sma_tqqq
- Simple 200-day SMA with 5%/3% asymmetric tolerance bands. Buy when price > 5% above SMA, sell when price < 3% below SMA. 100% allocated when in, 100% cash when out.
- regime_defensive_rotation
- Multi-factor regime detection (50 SMA + momentum). Risk-on: rotate SPY/QQQ/XLE. Defensive: rotate GLD/TLT/XLU based on relative strength.
- regime_defensive_rotation_with_bands
- Same as above but with 5%/3% tolerance bands to reduce whipsaws.
- Leveraged versions (_2x, _3x)
- Apply 2x or 3x leverage using leveraged ETFs. Captures higher returns but increases drawdowns.
Data Sources
All code and data available in the repository. Backtesting engine is custom-built in Python with full transparency. Results are reproducible using the provided configuration files.