Who wins? The war of trading edges - Mean Reversion vs. Trend Following - live session on the StrategyQuant channel

Two strategies, two completely different market experiences

At first glance, both "toolbox" strategies might look similar: stocks from the S&P 500 index, trading on daily data, entries and exits defined by simple rules. However, trend following and mean reversion are two completely different ways of making money, both in terms of logic and trader psychology.

In the article, we will examine two specific systems:

• trend following based on breakouts above Bollinger Bands,

• mean reversion based on the classic RSI2 pattern.

Using these examples, it is easy to see what truly distinguishes these approaches—and why, instead of choosing a "winner," it's better to think in terms of a portfolio of strategies.

1. Trend Following – buy strength, hold winners

Trend following strategy, a system inspired by Nick Radge's book Unholy Grails. The idea is straightforward:

We buy stocks that break upwards from a long-term volatility range and hold them as long as the trend lasts.

Main Assumptions in Practice:

• Universe: S&P 500 stocks.

• Entry: breakout above the upper Bollinger Band (e.g., 100 days, +3 standard deviations).

• Exit: drop below the "tighter" band (e.g., +1 SD) or another logical trend expiration criterion.

• Market Filter: e.g., SPY above a long moving average (SMA 75) – we only play when the entire market is in an uptrend.

• Position Ranking: e.g., ROC(30) – we select the strongest stocks from recent weeks.

• Portfolio: maximum number of positions at once (e.g., 15), all balanced equally.

• Orders: entries and exits at the session close.

How does such a strategy perform?

• An average annual return of ~12% CAGR over approximately 20 years of data.

• Several hundred transactions throughout the entire period (relatively few).

• Maximum drawdown (open DD) over 33%.

• The transaction success rate is relatively low, but individual outliers (e.g., Tesla after a breakout) drive most of the profit.

This is a typical trend-following profile:

• many small losses and small gains,

• a handful of huge, rare winners,

• long periods of "nothing special," interrupted by years when the strategy makes a really huge leap forward – often after major bear markets.

  
  

2. Mean Reversion – Buy weakness in an uptrend

The second strategy is the classic RSI2 mean reversion in the Larry Connors style – but with some practical adjustments.

Logic in one sentence:

We buy temporarily "oversold" stocks within a long-term uptrend and sell on a quick rebound.

Main Assumptions:

• Universe: again, actions from the S&P 500.

• Long-term trend: price above SMA(200) – we are only interested in companies in an upward trend.

• Entry: RSI(2) < 10 (strong, short-term oversold).

• Ranking: we select companies with the lowest RSI(14) – "most beaten down" in the short term.

• Portfolio: up to 10 positions simultaneously, equal weight.

• Exit:

  • RSI(2) > 70 or

  • max. number of days in the trade (e.g., 15),

  • plus technical stop loss (e.g., 30%).

• Orders: entry and exit at the session close.

Typical results of such a system:

• ~14% CAGR over a similar historical period.

• Approximately 9,000 transactions, which is incomparably more than in trend-following.

• Success rate around 70%.

• Maximum drawdown of about 33% (comparable to trend following).

• Average transaction duration is about 5 days.

The average profit from a single transaction is less than the average loss (payout ~0.6) – the result stems from a high success rate, not from single gigantic hits.

This provides a completely different trader experience:

• frequent transactions,

• many small profits,

• losses are less frequent but larger,

• equity grows more "stepwise" and reacts faster to market changes.

However, transaction costs are much more significant in MR system – with thousands of transactions, we need a very reasonable broker, a good fee structure, and automation.

3. Key differences between these approaches

In these two specific examples, you can clearly see what really distinguishes mean reversion from trend following:

a) Source of profits

Trend following:

• Earns mainly from a few to a dozen or so huge trends.

• The rest of the transactions are noise, small profits, and minor losses.

• One missing "outlier" can significantly change the outcome.

Mean reversion:

• Earns from thousands of small rebounds.

• Doesn't need individual stars – statistics matter.

• Each transaction is small, but together they add up to a solid result.

b) Psychology and comfort

• Trend following has a lower win rate but large wins. You need to be able to live with a series of losses and "boredom" between big trends.

• Mean reversion provides frequent "small wins," which are psychologically more pleasant, but occasionally there is a larger, painful loss.

c) Costs, technology, execution

• Trend following has relatively few transactions – it can even be conducted semi-manually.

• Mean reversion practically forces automation – with hundreds of transactions per year, manual clicking doesn't make sense, and every tick in costs starts to matter.

4. Who wins? Really... the portfolio

Looking solely at CAGR, mean reversion wins in this example (14% vs 12%). Looking at maximum drawdown, both systems are similar (around 30+%). The accurate picture emerges only when we combine these strategies in one portfolio:

• Trend following shines after major bear markets and in strong index trends.

• Mean reversion can earn "in the noise" when the market moves sideways, making it harder for trend followers.

• Their equity curves are partially independent, which means that when combined:

  • equity is smoother,

  • the return/drawdown ratio usually improves,

  • the weakness of one style is mitigated by the other.

    
    

Therefore, instead of asking:

“Which strategy is better – mean reversion or trend following?”

a more sensible question is:

“How to build a portfolio where both approaches work for me simultaneously?”

Then:

• mean reversion provides frequent, statistically based profits,

• trend following captures large moves and a "bonus" after a large period of panic,

• the whole is more resilient to market regime changes than a single system.

Attention! The above article compares trading concepts rather than final transactional systems. You can use the guidelines provided here as a basis for your system to test various filters, position scores, SL and TP levels, and different methods of entering and exiting positions in your final system.