Trend following versus mean reversion

What trend following means

A trend-following rule assumes that a move can persist long enough to trade. It does not need to predict the final top or bottom. It waits for evidence of direction, enters after some movement has already happened, and exits when that evidence weakens or reverses.

A moving-average crossover is a typical example. When a faster average crosses above a slower one, recent prices are stronger than the longer baseline. A Donchian breakout uses a new recent high as evidence. MACD, PSAR, Vortex, and Aroon can also express trend ideas.

The cost of confirmation is delay. A long strategy will not buy at the exact low, and it will usually give back some profit before an exit signal appears. That is part of the design, not proof that the calculation failed.

What mean reversion means

A mean-reversion rule assumes that an unusually large move away from a recent reference is likely to reverse. The reference might be a moving average, a volatility band, or the balance between recent gains and losses.

An RSI rule may buy after momentum becomes unusually weak and exit after it recovers. A Bollinger Band rule may enter when price moves below a lower band and leave when price returns toward the middle. Stochastic, Williams %R, CCI, and MFI are also commonly used to describe stretched conditions.

The central danger is confusing “far from average” with “about to reverse.” During a strong decline, the average follows price downward and an oversold indicator can stay oversold. Buying every dip is not automatically mean reversion with controlled risk.

Different market conditions, different pain

Trend rules often struggle when price moves sideways. Small moves trigger entries, reverse, and trigger exits. The result is a string of small losses and transaction costs—whipsaw. Their payoff may depend on a smaller number of sustained moves paying for those failures.

Mean-reversion rules often look comfortable in stable ranges: they buy near the lower part and exit after recovery. Their painful period comes when a range becomes a new trend. Frequent small wins can be followed by one much larger loss.

No label guarantees performance. A slow mean-reversion rule can resemble a trend filter, and a short breakout can reverse quickly. Read the exact entry, exit, and position logic.

A simple price example

Suppose price moves 100, 102, 105, 108, 111. A trend rule may enter at 105 after confirmation and remain long. A mean-reversion rule may avoid buying because price looks extended—or may sell under a two-sided design.

Now suppose price moves 100, 97, 94, 98, 101. A mean-reversion rule may buy at 94 and benefit from recovery. A trend rule may remain flat because direction was negative. If the sequence continues from 94 to 80 instead, the mean-reversion entry becomes dangerous while the trend filter’s delay is protective.

The same entry can look intelligent or foolish depending on what happens next. A backtest measures how the rule handled many such sequences, not whether one chart example is persuasive.

How TickRun represents both styles

TickRun’s strategies are long-only and all-in/all-out. A valid buy moves position from flat to long; a valid sell returns it to flat. The strategy earns the next close-to-close return after a closing signal, and position changes incur the selected cost.

This common engine makes comparisons easier, but it also limits the meaning of some labels. There are no short positions, leverage, scaling in, volatility-based sizing, or stop orders. A “sell” means close the long position, not open a short trade.

Signal delay can test whether a rule depends on immediate timing. Minimum holding period can reduce rapid exits, though it may also force a strategy to remain in a bad position. During optimization, TickRun searches delays but holds minimum holding period at zero.

Turnover and trading costs

Both styles can trade frequently. A fast crossover whipsaws in a range; a sensitive oscillator repeatedly trades small moves. Costs are paid on every entry and exit, so a small average edge can disappear even when the gross chart looks attractive.

Test more than one plausible cost. Also remember that TickRun’s single basis-point field is a simplified model. It does not estimate changing spread, market impact, partial fills, or gaps. A rule that survives only at zero cost is fragile.

How to compare them fairly

• Use the same ticker, dates, initial capital, and transaction cost.
• Check time in market; a mostly flat strategy is taking less exposure.
• Compare maximum drawdown and the full equity path, not only return.
• Count completed trades and inspect the open final position.
• Identify whether results depend on one trending or ranging period.
• Change nearby parameters rather than trusting one winning value.
• Validate the unchanged rule on later data.

Which approach should you choose?

Start with a market behavior you can explain. If the idea is that persistent information or investor behavior creates continuation, test a trend rule. If the idea is that short-lived pressure pushes price away from a stable reference, test mean reversion. Do not choose solely because one optimized result ranks first.

You can also study both without combining them. Their different failure modes teach more than a crowded rule containing six indicators. If you later add a regime filter, state precisely how it is known at the time and include the additional choices in the overfitting assessment.

Where to continue

Use the guide to all 40 strategies to see exact TickRun signals. The indicator introduction explains the main families, and the robustness guide shows how to investigate settings without worshipping the best result.