Strategy Lab

策略上线前，先过 SPY 和 QQQ 双基准。

这里记录策略 gate、研究来源和下一轮实验队列。历史通过不等于未来承诺，但至少不把明显跑输基准的策略包装成产品。

Publication Gate

当前策略通过情况

Gate: CAGR 为正、相对 SPY/QQQ CAGR 为正、最大回撤好于 -35%、Sharpe >= 1.0。

Strategy	CAGR	vs SPY	vs QQQ	Max DD	Sharpe	Gate
平滑相对强度增强	33.0%	12.8%	7.8%	-23.1%	1.37	PASS
质量低波相对增强	32.4%	12.2%	7.1%	-20.8%	1.52	PASS
QQQ 均衡 Alpha	32.0%	11.8%	6.8%	-22.0%	1.49	PASS
QQQ 相对质量低波	31.3%	11.2%	6.1%	-19.9%	1.47	PASS
均衡相对 Alpha	29.8%	9.7%	4.6%	-22.1%	1.40	PASS
平滑相对强度	27.7%	7.6%	2.5%	-22.7%	1.30	PASS
稳态赢家	27.1%	7.0%	1.9%	-21.1%	1.33	PASS
抗回撤动量	27.1%	6.9%	1.9%	-21.4%	1.33	PASS
质量低波复利	26.8%	6.6%	1.5%	-21.2%	1.31	PASS
基准相对低波	25.8%	5.6%	0.6%	-22.1%	1.22	PASS

Research Sources

Jegadeesh and Titman, Returns to Buying Winners and Selling LosersRanks stocks by medium-term relative strength, but adds trend and drawdown filters before weekly rotation.
Moskowitz, Ooi, and Pedersen, Time Series MomentumRequires positive own-trend confirmation instead of buying weak absolute trends only because they outperform a benchmark.
Fama and French, A Five-Factor Asset Pricing ModelKeeps strategy families diversified across momentum, quality/profitability proxies, and conservative risk controls.
Frazzini and Pedersen, Betting Against BetaPenalizes high-volatility and deep-drawdown names so the model is not just a high-beta QQQ clone.
vectorbtInspires future parameter-grid and batch experiment design for faster strategy iteration.
btUseful reference for strategy trees, weighing logic, and benchmark comparison reports.
backtraderUseful reference for more realistic order, slippage, and broker simulation later.

Next Experiments

Walk-forward parameter plateau testTest nearby weights/windows instead of one hand-picked parameter set. Gate: Median out-of-sample excess CAGR vs SPY and QQQ stays positive; no single parameter point carries the result.
Regime-aware cash filterReduce equity exposure when both SPY and QQQ are below long moving averages. Gate: Improves max drawdown without destroying CAGR in bull regimes.
Sector concentration capAvoid ten recommendations collapsing into one sector during momentum bursts. Gate: Maintains positive excess CAGR with lower single-theme exposure.
ETF/stock sleeve splitSeparate broad ETF rotation from single-name stock rotation for cleaner member use cases. Gate: Each sleeve beats its relevant benchmark independently before blending.

Disclosure: Strategy Lab is research documentation, not personalized investment advice or a promise of future returns.