VAMR BACKTEST
VOLATILITY-ADJUSTED MEAN REVERSION · REAL MARKET DATA · REGIME FILTER COMPARISON
ALPHA VANTAGE · REAL DATA
Free key at alphavantage.co — 30 seconds, no card needed. Saved in your browser automatically.
✓ Key saved
3yr
20
1.8σ
0.2σ
0.15
1.5σ
1.5x
Off by default — uses API credits. Enable when you want Claude's read on a result.
◈ AI ANALYSIS — RUN A BACKTEST FIRST
Enter your Alpha Vantage key above, select a pair and press Fetch & Run.
◇ PARAMETER CHEAT SHEET
— what each slider does & how to tune it
LOOKBACK PERIOD10–60 bars
Rolling window for mean and std dev — the model's memory. How far back it looks to define normal price behaviour on real data.
Higher (40–60): Smoother z-score, fewer trades, misses fast reversions. Better for Gold and Silver.
Lower (10–20): Reactive, more trades, noisier. Works better on liquid forex like EUR/USD.
ENTRY Z-SCORE1.0–3.0σ
Standard deviations price must deviate from rolling mean before a trade fires. Your selectivity dial — higher means only the most extreme real dislocations.
Higher (2.0–2.5σ): Fewer trades, higher win rate, better R:R. Start here on real data.
Lower (1.0–1.5σ): More trades, lower quality. Watch drawdown spike — trading noise.
EXIT Z-SCORE0.0–1.0σ
How close to the mean price must return before profit is taken. 0.0 = wait for full mean reversion. Higher exits early and locks in partial profit.
Higher (0.5–1.0σ): Early exits, shorter holds, lower profit per trade but faster turnover.
Lower (0.0–0.2σ): Waits for full reversion. Higher profit potential, longer exposure.
EMA VOL ALPHA0.05–0.50
Smoothing weight on EMA True Range vol estimate. Controls how quickly the vol model reacts to new real price movement.
Higher (0.3–0.5): Fast vol detection, catches real spikes but prone to false regime flips.
Lower (0.05–0.1): Smooth, stable vol. Lags real spikes but avoids whipsaw.
VOL REGIME THRESHOLD1.0–2.5σ
Z-score of current EMA vol vs its own rolling history that trips regime to ELEVATED. Core filter gate — how aggressively you sit out high-vol real market periods.
Higher (2.0–2.5σ): Only extreme spikes excluded. More trades but moderate vol still hits.
Lower (1.0–1.3σ): Aggressive filter. Fewer but cleaner trades. Watch for over-filtering.
STOP LOSS (ATR x)0.5–3.0x
Stop distance as a multiple of EMA vol. Volatility-normalised — wider real markets get wider stops. Position size scales inversely, risk fixed at 1% equity per trade.
Higher (2.0–3.0x): Survives real noise, lower stop-out rate but larger losses when stopped.
Lower (0.5–1.0x): Tight stops — gets clipped by normal daily noise. R:R deteriorates fast.
▲ KEY INTERACTIONS: Tune Entry Z x Lookback as one pair (signal quality) and Vol Threshold x EMA Alpha as another (filter sensitivity). On real data the filter matters most around central bank decisions, NFP and geopolitical shocks — visible as SPIKE regimes on Gold and GBP/USD.
ENTER API KEY  ·  SELECT PAIR  ·  PRESS FETCH & RUN