ensemble
What it checks
When no single strategy clearly wins, we combine the best three TA strategies with the best three alpha-family strategies — half-and-half — and use that blend as the champion. On stocks where every individual strategy was failing, the blend rescued 2 out of 3 of them.
Mechanism
Two-lane allocator that combines the Sharpe-weighted top-K from the TA-strategy bench (RSI/MACD/Bollinger/etc., 22 components) with the Sharpe-weighted top-K from the alpha-family bench (91 components, diversity-guarded). Each lane independently produces a blended position series; the allocator combines them with weight w_A (TA) + w_B (alpha) = 1, default 0.50/0.50. Reduces signal-specific failure modes by averaging over heterogeneous information sources at the same execution layer — TA captures price-pattern alpha, alpha-families capture event/fundamental/sentiment alpha, and the ensemble harvests both. Surfaces when neither bench produces a dominant single-strategy champion (cup_tier ≥ 2). On the May 2026 true-struggling cohort (46 tickers, no source cup_tier > 1) the ensemble + per-bench blends lifted cup_tier by an average of +1.43 levels on 65% of (ticker, profile) pairs.
Signal rule
position[t] = w_A · TA_blend(top_K_TA_strategies)[t] + w_B · alpha_blend(top_K_alpha_families)[t], clipped to [-1, +1]
Data dependencies
user_strategiesWorker data table — see services/worker schema.
Expected edge
- Paper alpha
- Variance reduction via diversification; +0.21 avg cup-tier lift system-wide
- Paper window
- 2026-05-15 production calibration on 615-ticker universe
Reduces variance vs single-strategy picks; on cohorts where no single strategy dominates, raises cup_tier by +0.2-1.5 levels.
Illustrative pattern only
NOT a backtestIllustrative pattern only — see /app for live backtests and the actual current equity curve.
Related families
meta equal weightMetaDeMiguel-Garlappi-Uppal (2009, RFS) "Optimal vs Naive Diversification" shows 1/N portfolio weighting is shockingly hard to beat out-of-sample once you account for estimation error in fancy optimization. We apply this at the alpha-family level: K=5 canonical sub-signals (SMA50/200 cross, RSI(14) mean-reversion, 20d-high breakout, 1m-vs-12m TSMOM-lite, 5d return reversion), each returns {-1, 0, +1}, the meta position is the mean clipped to [-1, +1] — buy when ≥3/5 agree long, short when ≥3/5 agree short.
meta regime routerMetaDaniel-Moskowitz (2016, JFE) "Momentum Crashes" + Asness-Frazzini-Israel-Moskowitz "Factor Timing": different alpha sleeves work in different regimes. We classify the regime as TREND (VIX<20 AND SPY>200d MA), REVERSION (VIX>25 OR SPY<200d MA), or CRISIS (VIX>35 AND falling SPY) and route the position: trend follower when TREND, 5d-reversion when REVERSION, short-only TSMOM when CRISIS. Distinct from regime_overlay which only gates a single signal.
champion overlayMetaTake this ticker's most recent user_strategies champion, replay its base position from the recorded indicator templates, then apply each Alpha Discovery V2 overlay filter on top: low-vol regime, earnings blackout, low FINRA short-pct, VIX contango, SPY uptrend, sector uptrend. Goal: surface cases where the existing champion would benefit from a regime/event/flow filter that the old finder doesn't apply.
champion overlay disagreementMetaBryzgalova, Pelger & Zhu (2025) Journal of Finance "Forest Through the Trees" — when an ensemble of model-derived signals disagrees on direction for the same asset, the disagreement itself is informative. High disagreement flags high regime-uncertainty rows where position size should be reduced or extremes faded. We compute per-ticker sign of every alpha-v2 champion's most recent backtested position, count sign-flips, and use that disagreement count as a long-side mean-reversion signal.
Explore ensemble on alphactor.ai
See which tickers this family is currently firing on, with live signals and rankings.