Trend Following: How to Make a Fortune in Bull, Bear, and Black Swan Markets

ELI5 / TLDR

A small group of traders — Bill Dunn, John W. Henry, Bruce Kovner, Ed Seykota, the Turtles, the AHL crowd — have for forty years run the same boring rules across stocks, bonds, currencies, and commodities: when something’s going up, buy it; when it’s going down, short it; cut losers fast, let winners run, never have an opinion about where price should go. They’ve made fortunes doing this through every kind of market the modern era has produced — Black Monday, Asian crisis, dot-com bust, 2008. Academia largely ignored them and Wall Street worked around them, but the audited monthly statements are public and they don’t bend. This book stacks them on the table and asks why a strategy with a forty-year, multi-asset, public track record gets less ink than the latest stock-picking guru.

The Full Story

What trend following actually is

The strategy fits on an index card. You watch prices, and only prices. When a market moves a meaningful amount in one direction, you take a position in that direction. You set a stop. If price keeps moving with you, you stay in and add. If it reverses, you exit and you do not argue. You apply the same rules to corn, the yen, German bunds, crude oil, the FTSE, copper, sugar, and the S&P, simultaneously, mechanically, with risk per trade dialled to a small percentage of equity. You do not forecast. You do not have a view on whether the move makes sense. You let the move tell you.

That is it. Nothing about earnings, central banks, supply chains, narratives, geopolitics. The rule set is so simple it embarrasses the people running it, which is why most of them won’t disclose it. Bill Dunn used to say the firm “does not pretend to determine the value of what we are trading, nor what that value ought to be, but they do produce absolute returns fairly consistently.” Campbell put it slightly more carefully: “Our trend following methods do not pretend to determine the value of what we are trading, nor do they determine what that value ought to be, but they do produce absolute returns fairly consistently.”

The track records that won’t go away

The book is built around audited monthly numbers, not stories. The cumulative chart that anchors everything: Dunn Capital Management’s composite, October 1974 through September 2016. $1,000 invested with Dunn at the start grew to $1,052,398. The same dollar in the S&P 500 total return index grew to $118,939. Compound annual rates: Dunn 18.02%, S&P 12.05%. Correlation between the two over forty-two years: -0.06.

That single chart is the book’s argument in pictures. The catch is also in the chart: twelve drawdowns greater than 25%, average drawdown 37%. The biggest was -63%. Dunn’s own marketing material warned investors: “If the investor is not willing to live through this, they are not the right investor for the portfolio.” Trend following is not a smooth ride; it’s a ride that arrives somewhere very different than the smooth one.

The pattern repeats across every major trend follower with a long track record:

Mulvaney, who runs a long-vol commodity-heavy program, posted +108.87% in 2008, +67.36% in 2014, +43.12% in 2013. Also -33.72% in 2012 and -23.14% in 2007.
John W. Henry’s Financials and Metals program earned +251% in 1987 — the year the S&P lost 28% — and roughly tripled his Original Investment Program at +58.2%.
Campbell & Company: $1,000 in January 1988 grew to $21,819 by December 2016, against $17,320 for the S&P. Of the 348 months, the strategy was profitable in 199. Of the 348 12-month rolling windows, profitable in 271. Of the 60-month rolling windows, profitable in 286 of 289 — 98.96%.
2002, the year the dot-com unwind finished and the S&P lost 23%, the Nasdaq 31%, and the Dow 17%: Bill Dunn +54.23%, John W. Henry +45.06%, David Druz +33.17%, Salem Abraham +21.37%, Mulvaney +19.37%, William Eckhardt +14.05%, Jerry Parker +11.10%.

The book leans on a table called “Crisis Events” pulled from Sunrise Capital. Across the eighteen worst quarters for the S&P 500 since 1987 — Black Monday, the dot-com burst, 9/11, the credit crisis, the European debt crisis, the China devaluation — the Barclay CTA Index was positive in fourteen of the eighteen. Black Monday Q4 1987: S&P -22.53%, CTA Index +13.77%. WorldCom Q3 2002: S&P -17.28%, CTA +6.77%. Russian default Q3 1998: S&P -9.95%, CTA +8.95%. Dot-com peak burst Q4 2000: S&P -7.83%, CTA +9.86%.

Covel’s argument from this is blunt: this is what crisis alpha actually looks like, and it has shown up reliably every single time stocks have had a real problem.

Why it works (the structural argument)

The book’s theoretical claim is light but coherent. Markets trend because human beings under-react and then over-react. New information seeps in, prices drift, eventually a crowd notices, prices accelerate, eventually they overshoot, and reverse. The trend follower doesn’t need to know the shape of any particular trend, only that some of them will be large enough to pay for the many small losses on the rest. Most trades lose. The few that win, win huge. The mathematics is asymmetric on purpose.

This is why trend followers care obsessively about position sizing and stops. Craig Pauley spelled it out: “If you have a $100,000 account and you’re going to risk 5 percent, you’d have $5,000 to lose. If your examination of the charts shows that the price movement you’re willing to risk equals $1,000 per contract, then you can trade five contracts.” The point is not that the strategy is clever; the point is that it survives long enough to catch the next big trend. Bruce Kovner: “Novice traders trade 5 to 10 times too big. They are taking 5 to 10% risks on a trade they should be taking 1 to 2 percent risks.”

Diversification across markets is the second pillar. Man AHL’s portfolio over the period broke down roughly as: Currencies 24.3%, Bonds 19.8%, Energies 19.2%, Stocks 15.1%, Interest rates 8.5%, Metals 8.2%, Agriculturals 4.9%. The point isn’t that any one of these markets is special. The point is that on any given month, something will be trending strongly enough to pay for the chop in everything else.

Covel’s correlation tables make the same point. The major trend followers are highly correlated with each other (most pairs 0.65 to 0.85), modestly correlated with the Barclay CTA Index, and roughly uncorrelated with the S&P 500 — most pair correlations between -0.21 and 0.16. This is the statistical signature of a real, separate return stream. They’re all surfing the same wave; the wave isn’t equities.

The multi-centennial backtest

The book’s most ambitious chapter is Greyserman and Kaminski’s research on a representative trend-following portfolio extended back to 1223 AD using historical commodity, currency, and bond data. From 1223 to 2013, their constructed buy-and-hold portfolio returns 4.8% annualized at 10.3% volatility, Sharpe 0.47. The trend-following portfolio returns 13.0% at 11.2% volatility, Sharpe 1.16.

The CTA Smile is the chart that earns its name: when the table sorts equity returns into quintiles from worst to best, trend following’s average return is higher in the worst quintile and the best quintile than in the middle. It’s smile-shaped. This is the data answer to the question “what does this strategy do in a portfolio” — it pays you most when you most need to be paid, and again when things are great, and is mediocre when things are merely fine.

When they slice by inflation regime — under 5%, between 5% and 10%, above 10%, over the period 1720 to 2013 — trend following Sharpe ratios are 0.87, 1.02, 1.02. Trend following doesn’t care what regime it’s in. Buy-and-hold famously does.

The Lempérière, Deremble, Seager, Potters, Bouchaud paper (Capital Fund Management) extends the same point with formal statistical tests on futures from 1960. Trend Sharpe ratio of 0.8, t-statistic 5.9, debiased t-statistic 5.0. Sectors: Currencies 0.57, Commodities 0.80, Bonds 0.49, Indexes 0.41. By decade since 1960, the trend Sharpe stays in a tight band: 0.66, 1.15, 1.05, 1.12, 0.75. It does not look like a backtest artefact.

Combining trend with traditional portfolios

The pragmatic case is the strongest part of the book. Greyserman and Kaminski compute the Sharpe ratio of a 60/40 portfolio at roughly 1.0, of equities alone at 0.7, of bonds alone at 0.8, of trend alone at 0.85. Adding trend to the 60/40 pushes it to roughly 1.2. The three-asset combo dominates every two-asset combo on every risk-adjusted measure they run.

The Drury vs Berkshire table is a smaller version of the same point. Drury Capital alone: 11.3% return, 20% vol, 32.5% drawdown, return-over-drawdown 0.35. Berkshire Hathaway alone: 10.4% return, 20.6% vol, 44.5% drawdown, 0.23. Half Drury and half Berkshire: 10.9% return, 14.4% vol, 23.9% drawdown, 0.50. The combination’s drawdown is smaller than either component’s. The combination’s return is roughly the same. Berkshire is doing something good. Drury is doing something different that’s also good. The two together are doing something better than either.

The Niederhoffer counterexample

The book includes one extended cautionary tale to set off the trend-followers’ discipline by contrast. Victor Niederhoffer ran a fund that wasn’t trend following — option-selling, short-vol, contrarian — and posted gentle gains of -1% to +8% per month through early 1997. In August 1997 the fund lost 50.18%. In October 1997, it lost 99.99%. By December 1997, “amount managed” was zero. The Niederhoffer table sits in the middle of the book like a warning sign: this is what happens when you build a strategy that’s right 95% of the time and devastatingly wrong the other 5%, and trend following is its mathematical opposite.

What the book is and isn’t

The fifth edition is roughly 480 pages of prose plus this exhibits volume. The structure is loose: chapters circle the same arguments from different angles, layering in interview snippets with John Henry, Larry Hite, Ed Seykota, Bruce Kovner, William Eckhardt, Salem Abraham, Bill Dunn, and others. Pull-quotes from Feynman, Larry Bird, Charlie Munger, Howard Marks, Gary Snyder, George Carlin, Anaïs Nin, Lewis Carroll, and Sir Arthur Conan Doyle festoon the margins. The author admires his subjects and isn’t shy about it.

The argument is repeated more times than is necessary. Every chapter ends roughly where it began. The level of polemic against academic finance, buy-and-hold orthodoxy, and the financial media is high enough that a reader who already believes the case will enjoy it and a reader who doesn’t will roll their eyes. The empirical case is overwhelming; the rhetorical packaging is excessive.

The actual how of trend following is mostly absent. Specific entry rules, exit rules, position sizing formulas, parameter choices — these are gestured at, never specified. Covel’s defense is that the rules are simple and well-known; the difficulty is psychological. That is partly true and partly evasive. A reader who actually wants to build a system will need to look elsewhere — Andreas Clenow’s Following the Trend is the standard practical companion.

Claude’s Take

The score is a 6, not higher, because the book is a triumph of curation and a flop of discipline. The empirical case for trend following is genuinely strong and gets stronger the longer the time frame: the multi-centennial Sharpe of 1.16 is a number that asset pricing has to take seriously, and the audited monthly statements of the major CTAs are not vibes. If you accept that prices trend because of human under- and over-reaction, you accept that something like trend following should work, and the data say something like trend following has worked, repeatedly, across every asset class with a futures market, for a very long time.

The problem is that Covel keeps making the case past the point of diminishing returns. By chapter eight you have it. By chapter fifteen the same data is being recycled with new pull-quotes around it. The book’s missing virtue is editing. Almost every flaw — the repetition, the hagiographic tone, the unwillingness to engage seriously with the strategy’s bad decade (2010–2018, when most major CTAs underperformed buy-and-hold by a wide margin), the absence of practical implementation detail — would be fixed by cutting it in half.

The other thing missing is honesty about the costs. Trend following’s drawdowns are deep and long. Dunn’s worst was -63% and the average of his twelve big ones was -37%. Most retail investors who say they can stomach a 50% drawdown discover otherwise. The book mentions this, then moves on. It is in fact the central reason the strategy works for some people and is unavailable to most: it requires a tolerance for being wrong, in a row, that is rare. The discipline isn’t in the rules; it’s in the bag of cement you have to swallow every morning during a bad year.

A fair-minded reader who finishes the book will probably end up roughly here:

The case for putting some allocation to trend in a multi-asset portfolio is strong. The Sharpe-ratio and drawdown-reduction math is clean. 5%–20% of a portfolio in a managed-futures fund, or in one of the cheaper trend ETFs that have appeared since this edition, is defensible.
The case for trend as a primary strategy, the way Covel implicitly argues, is much weaker. You’d have to be psychologically built to ride three-year drawdowns and ignore the entire equity bull market sitting next to you.
The actual implementation — sizing, stops, signals, market list — is not in this book. Read Clenow, or Andrew Lo’s papers, or AQR’s research notes, for that.
The book is best read by skipping ahead to the chapters with the multi-asset and multi-centennial research (chapters 19, 20, 27 in this edition) and treating most of the rest as supporting evidence and atmosphere.

The fifth edition came out in 2017. The decade since has been instructive. Trend had a poor stretch through most of the 2010s — low volatility, central-bank-suppressed cross-asset trends, a relentless US equity bull market — and then a spectacular 2022 when inflation came back and trend funds posted 25–40% gains while every other liquid alternative struggled. That cycle is the book’s thesis playing out exactly as advertised, with the unhelpful complication that the dry decade preceded it. Anyone who couldn’t sit through 2010–2018 wasn’t there for 2022. Which is, again, the central point: this only works if you stay in your seat.