Our house ensemble for stocks and ETFs — Golden Cross, EMA Trend Bias and OBV volume confirmation in a 2-of-3 vote. Deliberately filter-free, because that is what our data says.
The Wuguan Master is not a new indicator — it is a curated ensemble of strategies that already run in the Arena, selected purely from our own backtest evidence (26,000+ runs, evaluated with Deflated Sharpe Ratio checks).
What the data showed for stocks (medians across all tested symbols and periods):
The ensemble (v1, frozen):
K1 = SMA(50) > SMA(200) — the Golden Cross level
K2 = EMA Trend Bias(30/60) bullish — price trend with ATR neutral zone
K3 = OBV-MACD up — volume flow confirms the move
Vote = K1 + K2 + K3
ENTRY: Vote >= 2
EXIT: Vote <= 1
No regime gate, no filters — the stocks data is unambiguous on this. All component parameters are the unchanged live defaults of the source strategies; nothing was re-fitted.
Why these three? They look at different things: K1 is pure long-term price trend, K2 is medium-term trend with a volatility-aware neutral zone, K3 is the only component that reads volume — a move without volume behind it doesn't get its third vote.
Bias protection: Component parameters are frozen in v1 and cannot be tuned. Signals form only on confirmed closing data; once given they are immutable. Median results mix many symbols and periods — they are aggregate evidence, not a per-stock guarantee.
The pre-baked mini-backtest is refreshed daily — check back soon or start a live run in the Arena.
Run in Arena →// Components (v1, frozen — not user-tunable)
k1 = SMA(close, 50) > SMA(close, 200)
k2 = emaTrendBias(30, 60, ATR(60) * 0.30).state == 'bullish'
k3 = obvMacd(1, 9, 26, 2).tChannel == 'up'
vote = k1 + k2 + k3
// State machine (level-based)
if position.is_flat and vote >= 2:
BUY
if position.is_long and vote <= 1:
SELLBecause we follow the evidence, not the symmetry. In crypto, the 200-week SMA filter added +6 to +15 percentage points of median CAGR to trend strategies. In stocks the same filter was neutral to harmful (−0.8 to −2.5pp). Stocks spend far more time above their long-term average and their drawdowns are shallower — a hard regime gate mostly creates missed re-entries. So the crypto variant gets the gate, the stocks variant doesn't.
On median CAGR across our data: usually not — Buy & Hold on stocks sat at ~16% median while the components sat at 7–12%. What the ensemble buys you is the risk side: the strongest component configurations showed Sharpe ratios up to 1.04 and drawdowns around −11% where Buy & Hold drew down −32%. If you only care about maximum end value and can stomach any drawdown, Buy & Hold is honest competition. If you care about the path, that's what this is for.
No. Signals form only on confirmed closing candles — never on a forming bar — and once emitted they are immutable. New data changes future votes, not past ones. There is no repainting: if you backtest the same period twice, you get the same trades.
The classic trend-following signal — when the 50-day SMA crosses above the 200-day SMA, the trend has flipped bullish.
Two EMAs plus an ATR-based neutral zone — like the commercial Larsson Line, but tunable, transparent, and backtested. Choose your bias.
Apply MACD logic to On-Balance Volume — combining volume confirmation with trend signals. Volume tells you the truth that price hides.
Check out our Strategy Insights Reports — pre-baked deep-dives with historical results, comparisons, and market context.