Z-Score Mean Reversion

±2.0 is the standard threshold. At Z = -2, price is 2 standard deviations below average, which statistically occurs only about 5% of the time. More conservative traders use ±2.5 (even rarer), while aggressive traders might use ±1.5 (more common).

Typically 5-20 days for a 20-period Z-Score. The goal is for Z-Score to return to zero. If it doesn't revert within 15-20 days, consider exiting as the thesis may be wrong or a trend has developed.

Yes, Z-Score mean reversion works in TFSA accounts. The moderate holding period (1-3 weeks) and frequency are appropriate for registered accounts. Pairs trading also works in TFSA if you can short.

Start with 20 periods for short-term mean reversion. Use 50 periods for medium-term, or 100 for major moves. The period determines what 'average' you're measuring deviation from.

Mathematically yes! Z-Score of +2 equals being at the upper 2-SD Bollinger Band. The difference is presentation: Z-Score gives you a number (like -2.3), while Bollinger Bands show bands on the price chart.

Use a trend filter: only take oversold buys when price is above the 100-day MA. Check ADX - if ADX > 25, it's trending and mean reversion is risky. Weekly Z-Score can also provide context.

Pairs trading trades the spread between two correlated assets. Z-Score measures when the spread is statistically extreme. Spreads between correlated assets (like RY/TD) mean-revert more reliably than individual prices.

1) Calculate spread: Spread = Price_A - (Beta × Price_B), where Beta is from regression. 2) Calculate mean and standard deviation of the spread over your lookback. 3) Z = (Current Spread - Mean) / SD.

Tiered entries (e.g., 33% at Z=-2, 33% at Z=-2.5, 34% at Z=-3) can improve average entry if price continues. They're optional but reduce the risk of catching a falling knife with full position.

Most common: exit when Z-Score returns to zero (mean). Alternative: partial at Z=-1, rest at Z=0. Time stop: exit if no reversion in 15-20 days regardless. Stop loss: if Z goes to -3.5 or more extreme.

Calculate the Hurst exponent: H < 0.5 indicates mean reversion, H = 0.5 is random walk, H > 0.5 indicates trending. You can also run an Augmented Dickey-Fuller test or analyze autocorrelation of returns.

Half-life = -ln(2) / ln(ρ), where ρ is the autocorrelation coefficient from regressing price change on lagged price deviation. A half-life of 10 days means deviations halve every 10 days.

Use the Augmented Dickey-Fuller (ADF) test on the spread residuals. A p-value < 0.05 suggests cointegration. You can also use the Johansen test for more robust results. Re-test monthly.

In high volatility (VIX elevated, ATR high), use stricter thresholds (±2.5 or ±3). In low volatility, ±1.5 or ±2 may be sufficient. Can automate: threshold = 2 + (ATR percentile - 50)/100.

It depends on the instrument's mean reversion speed. Calculate half-life and use ~3x half-life as lookback. For most liquid equities, 20-50 days works. Test with walk-forward analysis to validate.