British American Tobacco Mean Reversion Strategy

BAT has low beta (it moves less than the market), relatively stable demand for nicotine products regardless of the economy, a high dividend yield providing income, and globally diversified cash flows. During market falls BAT typically declines less than higher-beta stocks; during rallies it rises less. That stability makes it defensive - it tends to defend portfolio value during volatility.

Bollinger Bands consist of a middle line (the 20-day average) and upper/lower bands at 2 standard deviations. They show when price is significantly above or below average. At the lower band price is oversold (2 SD below average); at the upper band it is overbought. Mean reversion trades these extremes expecting a return to the middle.

Price can stay at, or break through, the bands during strong trends - and BAT has at times suffered a multi-year structural de-rating. The reversal candle (hammer, engulfing) confirms that buyers have actually stepped in and momentum is shifting. Without that confirmation you risk catching a falling knife - buying as price keeps falling.

The primary target is the middle Bollinger Band (the 20-day moving average) - the mean that price reverts to. For strong setups the extended target can be the opposite band. BAT typically takes about 5-10 days to mean revert, making the middle band a realistic target.

BAT's high dividend yield (~7-8%) is very attractive to income investors. When the price drops, the yield rises (the same dividend divided by a lower price equals a higher percentage). Income investors buy for that yield, creating buying pressure at lower prices. This interest acts as a natural floor, supporting mean reversion from oversold levels - though it is not guaranteed if the market fears a dividend cut.

Check the ADX indicator: below 25 = range-bound/mean-reverting, above 30 = trending. Also check the band-width percentile: below 30 = squeeze (expect a breakout). Look at price action: multiple crosses of the middle BB = ranging, while staying above/below the middle = trending. Only apply mean reversion in a confirmed mean-reverting regime.

The UK Autumn Budget (late October/November) sets tobacco duty via the RPI+2% escalator and, from October 2026, a new Vaping Products Duty. A larger-than-expected measure can move the shares and blow through technical stops. This binary event risk overrides mean-reversion signals - avoid positions in the week before the Budget. Afterwards, if there is no negative surprise, mean-reversion opportunities often emerge from the oversold bounce.

For directional trades, ATM or slightly ITM options at the band extremes; a bull call spread (buy at the lower BB, sell at the middle BB) captures the defined move efficiently. For income, an iron condor sells options at both bands to profit from range-bound behaviour. But remember UK single-stock options trade on ICE Futures Europe with 1,000 shares per contract and thinner retail liquidity than the US, so spreads are wider - many retail traders use a tax-free spread bet instead.

Band width = (Upper - Lower) / Middle, showing volatility. Normal width = standard mean reversion. High width (after volatility) = larger moves possible, wider stops needed. Low width (squeeze) = expect a breakout, avoid mean reversion. Calculate the width percentile over 50 periods to classify the current regime.

Weekly Bollinger Bands show macro overbought/oversold zones; the daily generates signals. The best setups: daily oversold + weekly approaching oversold = a strong support zone. Both timeframes confirming raises the win rate. A weekly downtrend can override a daily oversold signal - the weekly provides context for the daily.

Optimise the parameters (BB period/SD, RSI period/thresholds) with walk-forward testing. Use the Z-score for a statistical framework. Classify the regime (ADX < 25 for mean-reverting). Create a signal quality score (band penetration, RSI, candle, volume, sector/rates). Backtest 5+ years net of UK frictions (0.5% SDRT, spread, financing) targeting: win rate > 55%, profit factor > 1.6, Sharpe > 1.0.

Calculate the spread ratio (BAT/IMB) and its historical mean and SD. When the spread deviates > 2 SD, go long the undervalued name and short the overvalued one (beta-adjusted for neutrality). Test cointegration to validate the relationship. This captures relative mispricing while hedging market direction. Retest cointegration quarterly as relationships can break - for instance if either company resets its dividend or New Categories trajectory.

High-importance features are typically RSI momentum (the direction of change, not just the level), band-width percentile (regime), Z-score extremity, days to results (event risk) and the change in gilt yields (BAT's bond-proxy sensitivity). ML captures non-linear interactions such as oversold + RSI turning up + falling yields = high probability. Use it as an ensemble with traditional analysis, never as a standalone screener or auto-trader.

Target delta 0.55-0.65 (ATM/slightly ITM). BAT's gradual moves mean gamma is less critical. Use 15-20 DTE minimum to manage theta over the 5-10 day expected duration (UK monthly options expire on the third Friday). BAT's low-to-moderate IV keeps long options affordable, but UK spreads are wide - a bull call spread for longs optimises for the defined target at the middle BB and reduces execution cost.

Base allocation 8-10% (BAT's low beta allows larger). Maximum consumer-staples sector 15%. Factor the large dividend income into returns, and shelter it where possible inside a Stocks-and-Shares ISA (or express the view via tax-free spread bets). Drawdown limit -8% (lower than high-beta). Track the strategy separately - win rate, profit factor, Sharpe - and compare with a buy-and-hold BAT total-return benchmark. The strategy should generate alpha to justify active management.