Mean Reversion Futures

Yes, mean reversion is counter-trend trading which carries inherent risk. That's why: 1) We only trade at statistical extremes where probability favours reversion, 2) We use smaller position sizes (1-1.5% vs 2%), 3) We require confirmation signals before entry, 4) We avoid trending markets (ADX > 30) - especially important with Australia's resources-led trends, 5) We accept that some trades will fail when trends continue. Mean reversion isn't about fighting every move - it's about identifying overextended conditions with high reversion probability.

Multiple measures identify overextension: 1) Bollinger Bands - price touching or piercing the outer bands (2 standard deviations), 2) RSI - below 25 (oversold) or above 75 (overbought), 3) Distance from VWAP - more than 0.3-0.5% for the SPI 200 intraday, 4) Distance from a moving average - significantly above/below the 20 or 50 period MA. It's best when multiple measures agree. Single-indicator overextension is a weaker signal than multiple indicators confirming the extreme.

Targets are based on the return toward the mean: 1) Conservative: 50% of the deviation distance (e.g. if 200 points below the mean, target 100 points of profit), 2) Standard: return to the mean or moving average, 3) Extended: if momentum shifts strongly, you can target the opposite extreme. Most traders use partial exits: 50% at halfway to the mean, trail the rest. Don't expect price to reach the mean every time - taking partial profits ensures you capture something from trades that reverse early.

Avoid mean reversion when: 1) ADX > 30 (strong trend), 2) Price is consistently making new highs/lows (trending), 3) Major news is driving the move (a fundamental shift - RBA, APRA, a commodity shock), 4) Higher timeframes are also extended in the same direction, 5) No confirmation signals despite extreme readings, 6) The first 30 minutes of trading (opening noise), 7) Quarterly expiry week (increased randomness). Mean reversion is a ranging-market strategy - recognise when conditions aren't suitable and stay flat.

Hold until: 1) Target reached (return to mean), 2) Stop hit (trend continued), or 3) Time stop expires. Typical durations: intraday VWAP reversion - 1-4 hours; daily Bollinger reversion - 1-3 days; weekly extreme - 3-7 days. If there's no reversion progress in the expected time, exit at market. Extended holds suggest your thesis may be wrong - trends can continue longer than expected. Use time stops as discipline, and remember the day-session-only sector futures can gap overnight against a held position.

Use both for confirmation: 1) Check the VWAP deviation - is price significantly above/below? 2) Check the Bollinger position - is price at or beyond the bands? 3) The best signals come when both confirm: price above VWAP AND at the upper Bollinger Band = strong overbought. Entry approach: use the Bollinger touch as the signal, VWAP as the target. Additional confirmation: RSI at an extreme. Multi-indicator agreement increases probability significantly.

Yes, but with strict rules: 1) Only if the original thesis remains valid (the extreme getting more extreme), 2) A predefined scaling plan (e.g. add 30% at 2.5 SD if entered at 2 SD), 3) Maximum additions limited (typically 2 adds maximum), 4) Stop fixed for the entire position (don't move the stop further away), 5) Total position must stay within risk limits. This is controlled scaling, not desperate averaging. If the position exceeds planned risk, exit - don't keep adding indefinitely.

SPI 200: Best for beginners - most liquid, moderate volatility, reasonably predictable reversion, though the resources weighting means it can trend on commodity cycles. Financials future (AF): bank-dominated, reverts cleanly around rate/credit equilibrium, but can trend on RBA/APRA catalysts. Resources future (AR): faster, commodity-driven moves with larger reversion targets, but it can extend much further on commodity trends - more aggressive and less reliable. Single-name shares: vary widely - some mean-revert well, others trend persistently. Before trading any instrument, analyse its historical mean-reversion behaviour: calculate half-life, test Z-score strategies, check ADF stationarity.

Catching falling knives is buying during downtrends hoping for a reversal - it usually fails. Mean reversion is buying at statistical extremes in ranging markets with confirmation. Key differences: 1) Mean reversion requires a ranging market (ADX < 25); falling knives occur in trends, 2) Mean reversion waits for confirmation; falling knives enter immediately, 3) Mean reversion uses statistical measures; falling knives use gut feel, 4) Mean reversion accepts that some trades fail; falling-knife traders hope every trade works. Discipline and statistics separate the two.

Volatility impacts it significantly: 1) Low volatility - smaller deviations from the mean, smaller profit targets, but more predictable reversion. 2) Normal volatility - optimal for mean reversion, standard parameters work. 3) High volatility - larger deviations, larger targets, but extremes can extend much further, so you need wider stops. Adjust the strategy: in a low A-VIX, use tighter deviation thresholds; in a high A-VIX (above ~18-20), use wider stops and require stronger confirmation. Track performance by A-VIX regime.

Z-score calculation: Z = (Price - N-period Mean) / N-period Standard Deviation. Implementation: 1) Calculate the rolling 20-period SMA and SD, 2) Compute the Z-score for each bar, 3) Entry when Z crosses below -2 (long) or above +2 (short), 4) Exit when Z returns to 0, 5) Stop if Z exceeds ±3. Refinements: use exponential calculations for faster response, adjust the N-period for instrument characteristics, combine with a regime filter. The Z-score provides an objective, standardised measure across instruments and timeframes.

Half-life is the expected time for a mean-reverting deviation to reduce by 50%. Calculation: 1) Run a regression of the price change on the lagged price level, 2) Half-life = -ln(2) / regression coefficient. Interpretation: if the half-life = 5 days, expect the deviation to halve in 5 days. Use: instruments with a shorter half-life are better for mean reversion (faster reversion). Typical values: liquid indices 3-7 days, some shares 10-15 days. If the half-life is very long (>30 days) or the regression coefficient suggests non-stationarity, avoid mean reversion on that instrument - a common situation for trend-prone resources names.

Detection methods: 1) Rolling ADX - when ADX crosses above 25-30, reduce mean-reversion exposure, 2) Rolling autocorrelation - negative autocorrelation suggests mean reversion works; positive suggests trending, 3) Track strategy performance - a declining win rate or increasing losses signal a regime change, 4) Bollinger Band behaviour - bands expanding with price trending along them = trending regime, 5) A-VIX regime - an extreme A-VIX (very high or very low) can disrupt normal mean reversion. Build a regime score and adjust allocation accordingly.

Key considerations: 1) Sufficient data - a minimum of 5 years covering different regimes, 2) Transaction costs - mean reversion has many trades; include realistic slippage and commissions (and the wider spreads on the sector futures), 3) Walk-forward testing - mean-reversion parameters can overfit; use walk-forward validation, 4) Regime separation - test performance in ranging vs trending periods separately, 5) Risk metrics - drawdown, time in drawdown and consecutive losses matter as much as total return, 6) Statistical significance - ensure enough trades for statistical validity. Expect a 55-65% win rate with smaller wins than losses but positive expectancy.

Integration approach: 1) Allocate 40-50% to mean reversion (works in ranging markets), 40-50% to trend-following (works in trending markets), 10-20% to breakout/other, 2) Regime-based allocation - increase mean reversion in low ADX, decrease in high ADX, 3) Correlation monitoring - ensure strategies aren't taking the same positions, 4) Performance attribution - track which strategy contributes in which regime, 5) Rebalancing - periodic adjustment based on performance and regime. This diversification smooths the equity curve as strategies perform well in different conditions.