The trend-following school

Definition

Trend following is the school that abandons prediction entirely: it doesn’t forecast where markets go, it systematically follows where they are going, cutting losers fast and riding winners as long as they run. Its claim is among the most heavily evidenced in this KB — Hurst-Ooi-Pedersen document positive trend returns in every decade back to 1880 across dozens of markets — and its culture (rules over opinions, losses as tuition, the system IS the edge) built much of modern systematic trading, including the famous Dennis-Eckhardt “Turtle” experiment demonstrating the method was teachable.

How it works / structure

• The mechanical core: entry on strength/weakness (breakouts — indicator-donchian-channels, the original Turtle rule — or moving-average crossovers), exit on reversal (trailing stops — mgmt-stop-loss in its ratchet form), volatility-normalized sizing (risk-volatility-targeting — equal risk per market, the school’s structural invention), applied across MANY uncorrelated markets simultaneously (strategy-futures-trend-following carries the engine-executable form).
• The return shape (the school’s psychology): most trades lose small (documented win rates near 30-40%); profits concentrate in a few large trends — the distribution is unbearable without the discipline, which is why the school’s texts are half rules, half temperament (the Turtle experiment’s documented finding: the rules were easy, following them was the filter).
• The evidence record: century-scale positive returns (Hurst et al), crisis-window outperformance (“crisis alpha” — trend profits concentrated in sustained dislocations: 2008’s documented CTA year), and the post-2010 flattening (crowding and chop-regime debates, labeled live) — the record and its recent softness carried together.
• The philosophical stance: markets trend because information diffuses and crowds herd (bias-herding monetized rather than resisted) — trend following is behavioral finance run in production.

When it applies

Multi-market portfolios (the edge is the ensemble — single-market trend following forfeits the diversification that carries the record); regime persistence windows (sustained macro moves are the harvest — episode-rates-shock-2022 was a documented banner trend year); portfolio diversification (trend’s crisis-alpha profile is the documented complement to long-only equity sleeves).

Risk profile & failure modes

• Chop consumption (the base cost): rangebound regimes whipsaw every entry — the documented drawdown state, arriving in clusters and lasting years; the school’s survival mechanism is small per-trade risk and many markets, never conviction overrides.
• Discipline decay: overriding signals after loss strings is the documented practitioner failure (the Turtle finding restated); the system’s edge lives in the trades that feel worst.
• Crowding at scale: managed-futures AUM concentration produces synchronized entries/exits in the same breakouts (documented flow studies) — capacity is a live question at institutional size, minor at retail scale.
• Late-trend entries: the school enters AFTER moves start by construction — the cost is structural giveback at turns; trailing-exit parameterization is the give-take dial.

Evidence & limits

Hurst-Ooi-Pedersen (2017) is the century-scale peer-reviewed anchor; time-series momentum (Moskowitz-Ooi-Pedersen) is its academic formalization; the Turtle experiment is documented practitioner history. Post-2010 return softness is documented and its causes debated (labeled). The school’s evidence is cohort-level and ensemble-dependent.

Falsifiable-thesis examples

Illustrations only, not signals:

• “A 40-market ensemble (Donchian entries, ATR sizing, trailing exits) produces positive returns over the next 5 years with crisis-window concentration (school-claim replay)” — falsified by the ensemble ledger.
• “Win rate stays under 45% while average-win/average- loss exceeds 2.5 (distribution-shape check — the school’s signature)” — falsified by the trade statistics.

Cross-references

• The executable form: strategy-futures-trend-following (and strategy-momentum cross-sectional cousin)
• The tools: indicator-donchian-channels, risk-volatility-targeting, mgmt-stop-loss
• The temperament layer: bias-herding (monetized), disc-process-vs-outcome
• The discretionary neighbor: style-global-macro