4 Regime Checks for Retail and Algo Traders: Trend or Mean Reversion
- Steven Hartwell
- 1 day ago
- 7 min read
Trend following bets that a moving price keeps moving; mean reversion bets that a stretched price snaps back to its average. Neither wins outright: a century-long study of time-series momentum found trend following pays off across asset classes when positions are volatility-scaled, while Investopedia’s mean reversion research shows the opposite style dominates range-bound markets. Pick based on the regime you’re actually trading, or use an indicator like Big Move Algo to filter for one automatically.
TL;DR:
Trend-following strategies perform best in markets with strong directional movement, confirmed by ADX above 25 across multiple timeframes.
Mean reversion thrives in low-volatility, range-bound markets where ATR is below the 30th percentile and ADX signals chop.
Combining multiple lookback periods and regime filters helps avoid false signals and adapt to changing market conditions.
Automated tools like Big Move Algo can assist in managing regime filters, volatility scaling, and alerts, reducing human error.
Proper position sizing and risk management depend on whether the market regime favors trend or reversion, not on a fixed strategy choice.
Table of Contents
Trend Following vs Mean Reversion: Core Differences and the Evidence
The two styles start from opposite assumptions about how prices behave. Trend following assumes that once a market starts moving in a direction, it tends to keep moving there for longer than random chance would predict. Mean reversion assumes the opposite: that prices are elastic, snapping back toward a statistical average once they stretch too far from it. Everything else, from signal choice to holding period to how you’ll feel three trades into a losing streak, flows from that fork in the road.
Trend traders lean on tools built to confirm persistence: moving average crossovers, the ADX for measuring trend strength, and time-series momentum models that simply ask “is this asset up or down over the last N months?” Mean-reversion traders reach for tools that measure distance from normal: z-scores, RSI readings above 70 or below 30, and Bollinger Bands that flag when price has wandered outside its usual range. Investopedia notes that RSI and Bollinger Bands remain the two most common mean-reversion signals among retail traders, largely because they’re easy to visualize and backtest.
Lookback and holding periods diverge sharply between the two approaches:
Trend following typically uses lookbacks of one to twelve months and holds positions for weeks to months, riding the trade until the trend exhausts itself.
Mean reversion typically operates on much shorter lookbacks, from a few days to a few weeks, and exits quickly once price reverts toward its average.
Blending both frequencies, short-term reversion, medium-term momentum, and occasionally long-term reversion, is a known method for smoothing an equity curve across market cycles.
The statistical profiles are almost mirror images. Trend-following systems often win a minority of trades, but the winners tend to be large multiples of the losers. Mean-reversion systems often win a majority of the time, but each win is small and a single bad reversal can erase a week of gains. Both profiles produce similar expectancy over time; they just deliver it on completely different emotional schedules, a distinction The Capital Process ties directly to psychological cost.
The empirical record for trend following is unusually long. Researchers tracing time-series momentum back to 1880 found that a strategy volatility-scaled to a 10% annualized target produced persistent excess returns across equities, bonds, currencies, and commodities for over a century. AQR extended that work across 67 markets and found the edge holds up, though performance is sensitive to cross-asset correlation and the specific lookback window chosen. Mean reversion doesn’t have quite the same century-spanning dataset, but pairs trading and statistical arbitrage research going back decades shows a comparable, if more execution-sensitive, edge.
How Do You Know Which Regime You’re In?
You don’t need a PhD to tell trend from chop.
Average True Range (ATR) percentile tells you how compressed volatility is relative to its own history: low ATR percentiles often precede reversion-friendly conditions, while expanding ATR often accompanies a fresh trend. ADX above roughly 25 generally signals a market with directional strength worth following; ADX below 20 usually signals chop, where reversion trades have more room to work. Checking whether the trend agrees across two or three timeframes, say daily and 4-hour, filters out a lot of false starts. And watching correlation dispersion across a basket of related assets can hint at whether a broad macro trend is forming or individual instruments are just noise.
A simple, codeable checklist looks like this:
Check ATR’s current percentile against its trailing 100-day range. Below the 30th percentile, lean toward mean reversion setups.
Check ADX on your primary timeframe. Above 25, with the higher timeframe trend pointing the same direction, lean toward trend-following entries.
If ATR is elevated but ADX is weak, treat the market as unstable and reduce size on both styles until one confirms.
Reconfirm the regime every few sessions. Regimes shift, and a filter that isn’t rechecked becomes a stale assumption.
Different instruments lean toward different regimes more often than others. Crypto assets frequently trend hard once they break a range, since retail momentum and liquidation cascades reinforce direction. Major forex pairs, especially in low-volatility macro periods, spend long stretches range-bound, which is why reversion strategies have historically found a home there. Commodities tend to sit quietly until a supply shock hits, then trend aggressively for weeks. None of this is a law of physics. It’s a tendency worth building a filter around, not a guarantee worth betting the account on.
Pro Tip: When a regime filter is ambiguous, don’t force a full-size trade in either direction. Combine volatility scaling with a smaller starter allocation until price action confirms the regime, then scale up. Practitioner guidance on transitioning between styles backs this up: half-size positions during regime uncertainty cut drawdown without sacrificing much upside.
Building the Trade: Entries, Exits, and Position Sizing
A workable trend-following recipe combines three ingredients: a trend filter (a blend of 1-month, 3-month, and 12-month price momentum works well and is the same structure used in the long-run academic studies), a breakout or pullback entry once the filter confirms direction, and a volatility-scaled position size so that a $50 stock and a $5 crypto token contribute roughly equal risk to the portfolio. Exits run on a trailing stop tied to ATR, letting winners run rather than capping them at an arbitrary target. Trend-following practitioners have used this constant-volatility sizing approach for decades specifically because it keeps risk contribution stable as instruments move in and out of high-volatility phases.
A mean-reversion recipe looks different at almost every step:
Entry triggers off a normalized z-score or a Bollinger Band touch, not a breakout.
Stops sit tight, since the entire thesis breaks if price keeps moving against you instead of reverting.
Holding periods run days, not months, and positions get closed on a reversion to the mean rather than a trailing stop.
Transaction costs matter far more here: frequent small-edge trades get eaten alive by spread and slippage in a way trend trades rarely experience.
Pairs and stat-arb variants need careful pair selection and ongoing stability testing, since a correlation that worked last year can quietly break down.
Before risking real capital on either style, run out-of-sample testing on a period your model never saw, model realistic slippage rather than assuming perfect fills, and stress-test how sensitive your results are to small changes in the lookback window. A strategy that only works with a 47-day lookback and falls apart at 45 or 50 days isn’t a strategy. It’s an overfit curve.
Where Automation Fits Into the Decision
Running a regime filter, a volatility-scaled position size, and a trailing stop by hand, on multiple charts, every session, is exactly the kind of repetitive judgment call where humans slip. An indicator like Big Move Algo builds trend filters, volatility scaling, and alerting into one TradingView tool, with a built-in Fake Trend Detector designed to flag low-quality conditions where neither trend nor reversion trades tend to work cleanly.
Automation removes emotional hesitation and missed entries; it doesn’t remove the need for judgment around news events, exchange outages, or execution failures. A practical rollout:
Link your alerts to execution so signals don’t sit unread in a notification tray.
Backtest indicator signals against realistic transaction costs before trusting them live.
Start on a demo account or small allocation, using automated signals to build confidence before scaling size.
The One Rule Worth Remembering
Trade the regime, not your preference: trending conditions favor trend following, range-bound conditions favor mean reversion, and when you can’t tell which you’re in, size down or run a filter until the market decides for you. Next steps: backtest with realistic transaction costs, code a simple regime filter like the ATR/ADX checklist above, and size positions with volatility scaling from day one. For deeper implementation detail, see how buy and sell signals get calculated and how to detect trend reversals before the regime shift catches you off guard.
Why Most Traders Pick the Wrong Side of This Debate
The mistake I see most often isn’t choosing trend following or mean reversion. It’s choosing one permanently, as an identity, instead of as a response to current conditions. The academic evidence for trend following is genuinely strong across a century of data, and that strength gets misread as “trend following is the better strategy” rather than what it actually shows: trend following works when volatility scaling and cross-asset diversification are done correctly, and it goes quiet or draws down hard in choppy, correlated markets.
Mean reversion gets dismissed too quickly by traders who’ve only seen it fail in a strong trend, without acknowledging it was never designed for that environment in the first place. The honest answer to “which is better” is neither, measured in isolation. What separates traders who survive from those who blow up is whether they built a regime check before picking a side. If you take one thing from this comparison, let it be that: build the filter first, then choose the strategy it points you toward. Diversifying across styles the way you’d diversify a portfolio isn’t a hedge against being wrong. It’s an admission that markets don’t hold still long enough for either style to work forever.
— Steven Hartwell
This article is general information, not a substitute for advice from a qualified financial advisor. Consult a qualified financial professional about your own circumstances before acting on anything here.
Sources
Recommended
Comments