Gold Mean Reversion

Buying dips is typically done in uptrends. Mean reversion specifically trades against extremes (overbought AND oversold) in range-bound markets, using statistical measures like RSI and Bollinger Bands to define 'extreme.'

RSI (14-period with 30/70 thresholds), Bollinger Bands (20-period, 2 std dev), and VWAP for intraday. Using multiple indicators in confluence improves reliability.

In trends, overbought/oversold conditions can extend further. Gold can stay overbought for weeks in an uptrend. Mean reversion assumes reversion, but trends continue in one direction, causing losses.

Mean reversion typically achieves 55-70% win rates - higher than trend following (40-55%). The trade-off is smaller average winners. Profit comes from consistency, not big individual wins.

Depends on timeframe. Intraday VWAP trades: hours. Daily RSI trades: 1-5 days. Set time stops based on expected reversion time. If it doesn't revert in reasonable time, the setup may have failed.

Use ADX indicator - below 25 indicates range-bound conditions suitable for mean reversion. Also check moving average slope (should be flat) and absence of clear higher-highs/higher-lows pattern.

Z-score of ±2.0 is standard - price is 2 standard deviations from mean, statistically rare (5% of time). Some traders use ±1.5 for more trades or ±2.5 for higher probability but fewer trades.

For intraday trading, look for price at VWAP ±2 standard deviation band PLUS RSI extreme (< 30 or > 70). This confluence of two indicators provides higher probability mean reversion signals.

Generally NO - this is averaging into losers. Get your size right initially. If the setup is wrong, adding more only increases losses. Mean reversion requires discipline to accept being wrong.

Place stops beyond the extreme - typically 1-1.5× ATR beyond the band touch or RSI extreme low/high. This gives room for normal volatility while defining maximum loss.

Regress price changes (ΔP) on price deviation from mean (P - μ). The coefficient gives theta (reversion speed). Half-life = ln(2)/θ. Re-estimate periodically as parameters change over time.

Use pairs trading (gold/silver, gold/AUD) when you want market-neutral exposure. Use directional when you have a view on gold direction. Pairs trading has less directional risk but more complexity.

Monitor ADX during trade. If ADX rises above 30 (trend starting) while in a mean reversion position, consider early exit. Regime change means your statistical edge is disappearing.

Overcrowding (too many traders using same signals), regime changes (market becoming more trending), parameter drift (optimal settings change over time). Regular re-optimization and monitoring is required.

Use walk-forward optimization: train on in-sample, test on out-of-sample, roll forward. Use simple parameters (fewer is better). Accept median performance, not best. Test robustness across different periods.