A trend-following system is a rule-based trading strategy that buys instruments whose price has been rising and sells (or shorts) instruments whose price has been falling, betting that recent direction persists rather than mean-reverts. It makes no forecast of fair value and no attempt to predict turning points; instead it reacts to realized price, riding established moves and exiting when they reverse. The defining tension is structural: trend-following accepts a low hit rate — most individual trades lose small amounts during sideways, choppy markets — in exchange for capturing a handful of large, sustained moves that more than pay for the losers. It is, in effect, a long-volatility, positively-skewed strategy that profits when markets make big directional moves and bleeds when they don't.

How it's formed

Every trend system is built from four decisions, and most public and CTA systems differ only in how they parameterize them:

1. Signal / entry rule — how a trend is detected. The two dominant families are: - Moving-average models: go long when a fast MA crosses above a slow MA (or when price sits above a single MA). Industry-standard lookbacks are 20-, 50-, 100-, and 200-day, per StockCharts/QuantifiedStrategies overviews. - Breakout (channel) models: go long on a new N-day high, short on a new N-day low. The famous Donchian channel / Turtle rules used 20-day and 55-day breakouts (Richard Dennis's 1980s Turtle program). 2. Position sizing / risk normalization — the unglamorous core of professional systems. Positions are scaled inversely to each market's volatility (commonly via 20-day Average True Range) so each holding contributes roughly equal risk — CTAs typically target on the order of 0.5%–2% of portfolio value risked per position (industry rule-of-thumb, TradersPost). A portfolio-level volatility target (e.g. an annualized vol target) then levers the whole book up or down. 3. Exit / stop rule — trailing stops (e.g. ATR-multiple stops, opposite-channel breakouts, or MA re-crosses). "Cut losses short, let winners run" is the mechanical expression of capturing positive skew. 4. Universe — classic managed-futures trend trades dozens of liquid futures across equities, bonds, commodities, and currencies, which provides diversification across uncorrelated trends.

The academic abstraction of all this is time-series momentum (TSMOM): sign(past 12-month excess return) determines the position, sized to constant volatility.

How it's used in practice

Trend-following is the dominant strategy of the managed-futures / CTA industry. Practitioners run it not as a stock-picking tool but as a whole-portfolio engine across asset classes, prized for a property no other liquid strategy reliably offers: crisis alpha. Because it can go short, a trend book can profit during prolonged equity bear markets — it tends to be short stocks and long bonds by the time a crash matures, making it a diversifier that pays off precisely when 60/40 portfolios suffer.

Most CTAs blend multiple speeds (short-, medium-, and long-term lookbacks) to reduce sensitivity to any single parameter, and combine MA and breakout signals. Risk management — vol-targeting and per-position risk parity — does more of the work than signal cleverness; this is the part retail imitations most often omit. In equity-only contexts the same logic appears as simple "above the 200-day MA" timing overlays, which the literature finds modestly reduce drawdowns at the cost of upside in strong bull markets.

Adoption, debate & evidence

Trend-following has unusually strong long-horizon academic support relative to most chart-based methods. Moskowitz, Ooi & Pedersen (2012), Time Series Momentum (JFE), examined 58 futures over 1985–2009 and reported the 12-month-lookback strategy delivered positive, significant returns in every asset class, with a composite Sharpe of roughly 1.28 versus ~0.38 for buy-and-hold — and noted the effect persists ~12 months then partially reverses (AQR/SSRN). Hurst, Ooi & Pedersen (2017), A Century of Evidence on Trend-Following Investing, extended the data back to 1880 (67 markets) and found positive average returns in every decade and strong performance in 8 of the 10 worst crisis periods for a 60/40 portfolio (AQR/SSRN). Researchers also find no clear capacity constraint, consistent with deep futures liquidity.

The honest counterweight: a documented "lost decade." Industry write-ups describe the SG Trend Index as roughly flat over 2009–2018 (commentators cite a total return on the order of a few percent for the full ten years, with a peak-to-trough drawdown in the low-20% range), a far cry from its strong pre-GFC track record (Man Group, Trend Following and Drawdowns). Sharp whipsaw years — 2011 and 2018 — saw the majority of constituents finish red. Critics argue the post-GFC low-vol, central-bank-anchored, fast-mean-reverting regime starved trends of the sustained moves the strategy needs; AQR-aligned researchers counter that such droughts are within historical norms and that fees, lower yields, and crowding explain much of the shortfall. The base-rate folklore is real and important: trend systems commonly win on a minority of trades (win rates frequently cited in the ~30–45% range for trade-level systems) and earn their returns from large outlier winners — a positively-skewed profile confirmed across studies. The "edge" is genuine and robust over a century, but it is lumpy, regime-dependent, and demands patience few discretionary traders possess.

Strengths & limitations

Works when: markets make large, sustained, low-noise directional moves — inflation shocks, commodity supercycles, prolonged bear markets, currency regime shifts. Its diversification (low/negative correlation to equities in crises) is its strongest, best-evidenced feature.

Fails when: markets chop sideways or mean-revert quickly (range-bound, low-vol regimes), producing strings of small losses and "death by a thousand whipsaws." It is structurally late — it never catches tops or bottoms and always gives back open profit at reversals.

The #1 misuse: treating it as a high-accuracy signal and abandoning it after a normal losing streak, or running it without volatility-based position sizing. The signal is the easy part; survival comes from risk normalization, diversification across many markets, and the discipline to hold through the multi-year flat stretches that are intrinsic to the strategy's skew.

Sources

• Moskowitz, Ooi & Pedersen (2012), Time Series Momentum, Journal of Financial Economics — SSRN
• Hurst, Ooi & Pedersen (2017), A Century of Evidence on Trend-Following Investing — AQR/SSRN
• QuantPedia, Why Did Trend-Following Underperform in the Last Decade? — quantpedia.com (causes of the post-GFC drought)
• Man Group, Trend Following and Drawdowns: Is This Time Different? — man.com (SG Trend Index ~flat 2009–2018, drawdown depth, 2011/2018 whipsaws)
• The Hedge Fund Journal, Trend Following with Managed Futures (crisis alpha, positive skew) — thehedgefundjournal.com
• QuantifiedStrategies / TradersPost — practical system components, MA/breakout, ATR sizing — quantifiedstrategies.com, blog.traderspost.io
• Graham Capital, Trend-Following Primer — industry construction overview — grahamcapital.com

Disputes flagged: the cause and significance of the 2010s underperformance is genuinely contested (regime change vs. crowding/fees vs. normal variance). Trade-level win-rate figures (~30–45%) are commonly cited and vary by system parameters — treated as qualified, not exact.