Trend Following — The Strategy That's Been Winning for 100 Years

Trend following is the oldest documented edge in trading — rules like "buy strength, cut losses" have been earning money on data going back to the 1920s. And at the same time, it's the strategy most people abandon after three months. That paradox is no accident: the math of trend following favors the trader, but the psychology works against them.

In this article we show the numbers from long-term tests, concrete step-by-step rules — and honestly explain why a strategy that's "worked for 100 years" still works: because almost nobody can actually stand to hold it.

What Trend Following Actually Is

Trend following doesn't predict the market. It doesn't guess tops or bottoms, and it doesn't analyze "why" a price is rising. It does exactly one thing: it gets on board a move that's already happening and rides it for as long as the move lasts.

The philosophy rests on three observations:

• markets trend more often and for longer than pure randomness would suggest — capital flows in in waves, information spreads gradually, and the crowd joins the move with a lag,
• nobody knows in advance which move will become a big trend — so the system takes every signal and lets statistics sort the wheat from the chaff,
• you cut losses fast and let profits run — a single trade doesn't matter; what matters is that losses are small and winners can be many times bigger.

The practical tools are almost embarrassingly simple: moving average crossovers (e.g., the golden cross), breakouts to N-day highs (the Donchian channel, known from the turtle system), price's position relative to a long-term average. Trend following is a close cousin of momentum — the nuance being: momentum compares the strength of a move (often between assets), while trend following looks at the direction of a single market and follows it.

What the Numbers Say — 100 Years of Data

The strongest cited result comes from QuantifiedStrategies backtests on US industry-sector portfolios going back to 1926. A simple trend rule (be in a position only when the sector is in an uptrend) produced:

Two things in that table matter more than the return figure. First, the horizon: a hundred years of data covers the Great Depression, a world war, stagflation, the dot-com bust and 2008 — this isn't a result fitted to one decade. Second, the drawdown: −33% instead of −84% is the difference between a painful year and losing a lifetime's worth of wealth. Trend following historically earned more mainly because it lost less in catastrophes — the trend rule pulled it out of the market before the bear market did its worst damage.

The picture is consistent in crypto: in the cited test of a Donchian system (20-day breakout, 55-day exit — a turtle variant) on daily BTC since 2017, the strategy was profitable through both bull and bear markets, with a win rate of only 30–40% and wins 3–5x bigger than losses.

Mandatory caveats: the numbers come from summaries of published backtests — verify them directly at the source before citing them. The sector tests cover the US stock market; transferring the parameters to BTC/ETH requires its own test. And the overriding rule: past results don't guarantee future ones.

The most important number here, though, is a different one: a 30–40% win rate. A system that's been earning for decades is wrong on most of its trades. That's not a flaw — it's the design. And it's exactly what breaks most people's psychology.

How to Apply It Step by Step

An example rule set for the crypto market (D1, Donchian variant + trend filter):

1. Regime filter: trade long only when the closing price is above the SMA200. This filters out attempts to catch trends in the middle of a bear market.
2. Entry signal: price closes at a new 20-day high (the upper Donchian band). No waiting for a pullback, no haggling — the breakout is the signal.
3. Initial stop: e.g., 2x ATR(20) below the entry price. Wide, because crypto is volatile; a tight stop in a trend system generates a string of unnecessary losses.
4. Exit: price closes at a 10–20-day low (the lower band) or breaks the trailing stop. No take-profits — you let the winner run, because you don't know which trend will 3x.
5. Risk: a fixed fractional stake per trade, e.g., 1% of capital (the position-sizing formula).

Numerical example on BTC (illustrative): a $10,000 account, 1% risk = $100 per trade. BTC is above the SMA200, closes the day at $64,000 — a new 20-day high. ATR(20) = 2,000, so the stop = 64,000 − 4,000 = 60,000. Risk per unit: $4,000, position = 100 / 4,000 = 0.025 BTC (notional $1,600, no leverage). Scenario A: the trend runs to $80,000, the trailing stop exits at $74,000 — profit 0.025 × 10,000 = +$250 (2.5R). Scenario B: a false breakout, stop hit at $60,000 — loss −$100 (1R). At a 35% win rate and an average win of 2.5R, the system nets a profit — but notice: out of 20 trades, you'll typically see 13 losses, sometimes 6–8 in a row.

Donchian is just one variant — the core could just as well be a moving-average crossover (e.g., 50/200) or a simple condition like "the monthly close is above the SMA10 of monthly data." Trend-following research consistently shows that the specific signal is a secondary detail: most reasonable rules catch the same big trends, and results differ mainly by cost and by the number of false signals. What matters more than the choice of indicator is diversification across markets — professional trend funds run the same system on dozens of markets at once, precisely so that a losing streak on one instrument gets offset by wins on another. On a retail crypto account, the minimum is a handful of independently behaving pairs instead of everything piled into BTC.

Psychology — the Real Cost of the Strategy

Here's the core that courses don't say out loud. Trend following is psychologically brutal for three reasons:

• Long losing streaks. At a 35% win rate, a streak of 7 losses in a row is statistical normality, not a system malfunction. Most people "fix" the rules after the fifth loss — and that's exactly when they stop having a system.
• Giving back paper profits. A trailing stop, by definition, gives back part of the peak. Watching +40% shrink to +25% before you close hurts more than a plain loss — and it's built into the method.
• Months of boredom. In a consolidation, the system generates a string of small losses and does nothing spectacular. FOMO whispers that you should "add something." Every such addition is usually the beginning of the end.

That's why trend following still works after a hundred years of publication: the edge is protected not by secrecy, but by pain. Arbitraging it away would require a mass of people to sit through losing streaks — and almost nobody does.

When It Doesn't Work and the Most Common Mistakes

• A sideways market is a tax on the strategy. In consolidation, every breakout reverses and the system bleeds small losses. This is unremovable — filters (e.g., ADX, channel width) ease the problem but don't erase it. If a market trends about 30% of the time, for the other 70% the system mostly waits and loses small amounts (in consolidation, a different approach performs better).
• Sharp V-shaped reversals. Trend following exits with a lag — a crash followed by a week-long bounce can take back the profit without giving a timely re-entry signal. Crypto makes moves like that more often than indices do.
• Stops that are too tight. Carrying day-trading intuition ("a 1% stop below entry") into a trend system kills it — a normal pullback wave stops you out of a position that would have been a big winner a month later.
• Cherry-picking signals. "I don't like this signal, I'll skip it" — and it's exactly the ugliest, most uncomfortable breakouts that tend to start the biggest trends. A system takes everything, or it isn't a system.
• Too little capital to survive a streak. At 3% risk per trade, a streak of 8 losses is −22% of the account and near-certain panic. Stakes of 0.5–1% look boring, but they let you live long enough for the trend that pays.
• Confusing trend following with "buying because it's going up." Without a defined exit and stop, that's not a strategy, that's hope. The entire edge sits in the asymmetry of cutting losses and letting profits run — not in the entry itself.

Trend following is the mirror image of mean reversion: there you win often and small, here you win rarely and big. Both families have documented edges in different markets and horizons — and both fail when used in the wrong regime.

The moral: a strategy that's been winning for 100 years doesn't win because it's clever — it's practically primitive. It wins because it's paid for in a currency most people don't have: patience, and a tolerance for being wrong most of the time. Before you play it, check the historical data for the longest losing streak — and honestly ask yourself whether you'd survive it with real money on the line.

FAQ

Rafał — Krypto Bez Ściemy

Trader and founder of Strefa Tradingu. He’s been breaking crypto down on YouTube for years — no hype, no signals, with a focus on market structure and risk management.

kryptobezsciemy.plInstagramYouTubeKlub Zero