Trend Following — Trend-Riding Systems from the Turtles to Donchian

In January 1983, in a rented office above the trading floor of the Chicago Board of Trade, Richard Dennis welcomed twenty-three people into a five-year experiment. Most had never traded before — the group included an actor, a security guard, a programmer, and a woman with a degree in linguistics. Dennis claimed he could teach them to trade in three weeks, and over five years the group generated roughly 175 million dollars in profit, with an average annual return near 80 percent. He called them the Turtle Traders, and he proved something the market did not want to believe: that mechanical trend following can be codified and run profitably by ordinary people.

The philosophy of trend following — react, do not predict

Trend following rests on a fundamentally different philosophy from most retail strategies. A classic investor looks for chances to „buy low” and „sell high” — trying to predict tops and bottoms. A trend follower does the opposite: they buy when the market has already broken out to new highs and is showing strength, and they sell when it has dropped to new lows. They predict nothing — they react to what the market has actually done, not to what the trader thinks it should do.

That inverted logic has a mathematical justification. Markets move in a clear directional trend only about 15 to 20 percent of the time, and they sit in sideways consolidation roughly 65 to 70 percent of the time. A strategy that tries to catch tops and bottoms in consolidation produces mediocre results most of the year. A strategy that captures trends during their strongest phase — even while it loses in consolidation — has an edge, because the few winning trades are many times larger than the losers.

What the Turtle Traders experiment taught us

Dennis was a legend at the Chicago Board of Trade in his thirties — from four hundred dollars borrowed from family he built a fortune of roughly two hundred million. His partner, the mathematician William Eckhardt, argued that this was an innate gift that could not be taught. Dennis insisted on the opposite, and they settled it with a bet: he recruited two waves of students through newspaper notices and gave each several weeks of training and a live account funded with his own capital. Five years later he had won the bet beyond dispute.

The most important lesson, though, was not about the rules but about discipline. Curtis Faith, the youngest of the group, wrote in Way of the Turtle (McGraw-Hill, 2007) that students who followed the rules to the letter earned far more than those who tried to „improve” the system with their own intuition. Mechanical discipline turns out to be far harder than it looks once your own money is on the line.

Classic entries: Donchian channels and moving-average crossovers

The first family of signals is the Donchian channel breakout. Richard Donchian, a Yale economist, described a simple construction in the 1950s: the upper line is the highest high of the last N sessions, the lower line the lowest low. For twenty sessions, a buy signal fires on a break above the upper line (a new monthly high) and a sell signal on a break below the lower line. I break the full mechanics down in a separate piece on the Donchian channel strategy.

The second family is the moving-average crossover — a shorter average (say fifty-day) and a longer one (say two-hundred-day). When the faster crosses above the slower, you get a „Golden Cross”, flagged as a sign of an uptrend; the reverse is a „Death Cross”. Averages give fewer false signals than raw breakouts, but they are slower — a Golden Cross usually appears only after price has already risen by a double-digit percentage off the low. That is the constant trade-off: faster signals catch more trends but more false alarms; slower ones filter noise but miss the start of the move. Many practitioners combine the two — the average as a direction filter, the breakout as the entry trigger. This is a variant of the classic breakout strategy with an added trend-filter layer.

When the strategy earns, and when it devours capital

In a study spanning more than a century of data, trend-following systems proved consistently profitable, although the gains concentrate in the few years with a strong structural trend — the inflation cycle of the 1970s, the dollar trend of the 1980s, the 2008 crisis, the 2020 pandemic, or the 2022–2023 rate-hike cycle. In those years a mechanical system catches the breakout early and stays in to the end of the trend, and a single trade can produce many times more than one stop loss. What lets you „let the profits run” is the moving exit — a break of the lowest low of the last dozen-or-so sessions acts like a trailing stop that follows price and closes the position only once the trend has genuinely broken.

The problem is that the market consolidates most of the time, and then every breakout can be false: price returns inside the range within a few sessions, knocking out the stops of traders who believed the move. The years 2014–2015 are a textbook example — with low directional volatility, large trend funds posted combined drawdowns of 30 to 45 percent. A „risk-on/risk-off” regime is just as dangerous: the market runs one way for a week and reverses the next, so a buy signal becomes a loss and the following sell signal does too. A few such pairs zero out the monthly risk budget, even though no trade was wrong by the rules.

„The essence of trend following is not the accuracy of any forecast but consistency: you take every signal, you accept the losing streaks, and you let the rare large trends carry the whole result. A trader who picks and chooses which signals to take will sooner or later skip the one that made the year.” — Andreas F. Clenow, Following the Trend: Diversified Managed Futures Trading, Wiley, 2013

Three pillars of risk management

First, small per-trade risk — on the order of 0.25 to 0.5 percent of capital, less than the classic one-percent rule. Trend following generates many more trades and longer losing streaks (five to ten in a row). At one percent per trade, ten losses in a row is a ten-percent drawdown — psychologically hard; at half a percent it is only five. Second, diversification across a dozen or so uncorrelated markets — not to smooth out any single position, but to raise the chance that somewhere in the world a big trend appears and pays for months of consolidation elsewhere.

Third, no manual override of the signals. The classic error sounds like „I know the system says buy, but this move looks weak, so I\'ll skip it”. After a dozen such skips it turns out the entire year\'s profit has been skipped — because those „questionable” signals are precisely the ones that deliver the biggest trades. The difference between the participants who earned Dennis millions and those who washed out lay not in the rules but in the discipline of executing them. The foundations of that discipline are covered in the risk-management section at ForexMechanics.com.

What to do tomorrow if you want to implement trend following

Trend following is one of the best statistically documented strategies in the history of markets — Donchian channels are still the foundation of funds managing billions of dollars. The price of that mechanical character is steep, however: profits in only 15 to 20 percent of years, drawdowns of 20 to 30 percent, and a win rate of just 30 to 40 percent. Before you commit to this style, weigh it consciously against a shorter horizon — the comparison of day trading versus position trading helps, because trend following sits firmly on the patient side. Then do not start with money; start with three steps, each of which you can take without risking a single dollar.

1. Write the complete rule set down on a single page. Define the entry signal, the stop-loss level, the exit signal, and the position size precisely enough that any trade can be executed without a single discretionary call. If you cannot state a rule in one sentence, it is not yet ready to be traded with real money.
2. Run a backtest on at least ten years of data. Go back through a full market cycle and measure not just the final result but, above all, the deepest drawdown and the longest losing streak in a row. Those two numbers, more than any return figure, tell you whether you can psychologically survive running this system live.
3. Run the system on one tenth of the target capital for six months. Trade it mechanically, skipping no signal whatsoever, and keep a journal of every trade. Most traders abandon trend following after the first long losing streak — if you sit through it and wait for the first big trend, everything after that becomes far easier.