Trend Following Explained: Dow Theory, the Turtles, and Why Korean Momentum Is Weird

A few months back I wrote a long piece on mean reversion, the idea that prices snap back toward an average. Trend following is its opposite, and the two strategies have spent the last hundred years fighting over the same markets. One says extremes get corrected. The other says extremes keep going. Both are right, just not at the same time, and not in the same place.

I find trend following more interesting than mean reversion for one reason. It violates the thing most people think they know about investing, which is “buy low, sell high.” Trend following does the opposite. It buys high and sells higher, or sells low and covers lower. It feels wrong. And yet the historical record behind it is one of the longest and most stubborn in all of quantitative finance.

This piece walks through where the idea came from, the people who turned it into money, the academic papers that finally explained why it works, and then the part I actually care about: why Korean equities behave so strangely when you try to apply it. A heads up before we start, I’m not a licensed advisor and none of this is investment advice. I write these because the history is genuinely good and the English-language coverage of the Korean angle is thin to nonexistent.

Buying high and selling higher

Here’s the core claim, stripped down. If an asset has been going up, it is more likely than not to keep going up over the near-to-medium term. If it has been going down, it tends to keep falling. Past returns carry information about future returns, at least over horizons of roughly one month to twelve months.

That sentence would have gotten you laughed out of an economics department for most of the twentieth century. The reigning theory was the random walk, the idea that price changes are independent and unpredictable. Trend following says no, there’s autocorrelation, and you can trade it.

The mechanism, as far as anyone can tell, is human. Investors underreact to new information at first, so prices drift toward their fair value slowly instead of jumping there instantly. Then, later in the move, other investors pile in chasing the rise, which overshoots fair value. Underreaction starts the trend, herding extends it. Eventually it breaks, which is why trend following also needs an exit rule, not just an entry.

So the irony is that mean reversion and trend following both come from the same source: people don’t process information rationally. Reversion profits from the overshoot getting corrected. Trend following profits from the slow drift and the herd. Different parts of the same broken human response.

The newspaperman who started it all

Trend following as a written idea begins with Charles Dow, who co-founded Dow Jones & Company and the Wall Street Journal. Between roughly 1899 and his death in 1902, Dow wrote a series of editorials about market behavior. He never called it a theory or a system. He was just observing that markets moved in identifiable trends, primary and secondary, and that volume and the confirmation of one average by another told you something.

After Dow died, two writers turned his scattered editorials into a structured framework. William Peter Hamilton, his successor at the Journal, wrote “The Stock Market Barometer” in 1922. Then Robert Rhea published “The Dow Theory” in 1932, which laid out the tenets most people still cite: the market has three movements, trends have three phases, the averages must confirm each other, volume goes with the trend, and a trend stays in force until clear signals say otherwise.

That last tenet is the whole philosophy in one line. A trend is assumed to continue until proven otherwise. You don’t predict the top. You wait for the trend to break and then you’re out. Every trend following system built since is, in some sense, a quantified version of Rhea’s sentence.

Donchian and the birth of managed futures

Dow Theory was qualitative. You read the charts and used judgment. The man who made trend following mechanical was Richard Donchian.

Donchian started trading in the 1930s and, in 1949, launched Futures, Inc., widely described as the first publicly held managed futures fund. His central tool was beautifully simple. The Donchian channel: if today’s price breaks above the highest high of the past four weeks, you buy. If it breaks below the lowest low of the past four weeks, you sell. He later worked with moving average crossover systems, like the 5-day and 20-day, where you go long when the fast average crosses above the slow one.

What matters about Donchian isn’t the specific rule. It’s that he proved a purely mechanical, emotion-free system could be written down, followed by anyone, and still make money over time. He removed the analyst’s gut feeling from the loop. That idea, that you could systematize discretion, is the foundation of the entire managed futures industry, which today runs hundreds of billions of dollars.

The Turtles: can trading be taught?

This is my favorite story in finance, so forgive the detail.

In 1983, two Chicago traders, Richard Dennis and William Eckhardt, had an argument. Dennis believed trading could be taught to almost anyone using a clear set of rules. Eckhardt thought great traders were born, not made. To settle it, they ran an experiment. They placed ads in the Wall Street Journal and other papers, interviewed hundreds of applicants, and selected a group of complete beginners. A former blackjack dealer, a fantasy game designer, an accountant. People with no trading background at all.

Dennis called them his “turtles,” supposedly after visiting a turtle farm in Singapore and saying he wanted to grow traders the way Singaporeans grew turtles. He gave them a two-week course in a trend following system, then handed each of them real money, in some accounts up to a million dollars or more, to trade.

The rules were textbook trend following. They used Donchian-style breakouts. Enter on a 20-day breakout, add to winners as the trend extended, cut losses fast at a fixed multiple of recent volatility, ride winners until a shorter-term breakout in the opposite direction signaled the trend was done. Position sizing was tied to volatility, measured by something called N, which was essentially average true range.

The result settled the argument. Over about four years, the turtles reportedly earned well over a hundred million dollars collectively. Dennis won the bet. Trading, or at least this kind of trading, could be taught. The full rule set later leaked and was eventually published, and you can read it today. It still broadly works, though crowding has thinned the returns.

When the academics finally caught up

For decades, trend following was a practitioner’s game that the academy dismissed as folklore. Then the evidence got too large to ignore.

The breakthrough on the stock side came from Narasimhan Jegadeesh and Sheridan Titman in 1993. Their paper, “Returns to Buying Winners and Selling Losers,” published in the Journal of Finance, did something simple and damning to the efficient market hypothesis. They ranked US stocks by their returns over the past 3 to 12 months, bought the top performers, shorted the bottom ones, and held for several months. The strategy earned roughly 1% per month over 1965 to 1989. This is cross-sectional momentum, picking relative winners against relative losers, and it became one of the most replicated findings in the field.

The other landmark came from Tobias Moskowitz, Yao Hua Ooi, and Lasse Heje Pedersen in 2012, working out of AQR. Their paper “Time Series Momentum” in the Journal of Financial Economics studied 58 futures and forward contracts across equities, bonds, currencies, and commodities. They found that an asset’s own past 12-month return predicted its next-month return, consistently, across nearly every market they tested. This is closer to what CTAs actually do: judge each asset against its own history, not against other assets.

Then in 2017, Brian Hurst, Ooi, and Pedersen published “A Century of Evidence on Trend-Following Investing,” which back-tested a simple trend system back to 1880. It delivered positive returns in most decades, and notably it tended to do well precisely when stock markets did badly. That property has a name in the industry: crisis alpha. Trend following frequently makes money during prolonged market declines, because it can go short and ride the move down.

So the lineage runs Dow to Donchian to Dennis to AQR. Editorial intuition, then a mechanical rule, then a teachable system, then a peer-reviewed century of data. Not a bad arc for an idea everyone called nonsense.

Why momentum gets weird in Korea

Now the part I actually wanted to write about.

If trend following and momentum are this universal, you’d expect them to work cleanly in Korea. They don’t. Korea is one of the markets where the textbook momentum effect is unusually weak, and in some studies it basically disappears or even flips.

The most cited explanation comes from Andy Chui, Sheridan Titman, and K.C. John Wei in their 2010 Journal of Finance paper, “Individualism and Momentum around the World.” They linked momentum profits to Hofstede’s individualism scores across countries. Their finding: momentum is strong in individualistic Western markets like the US and weak in collectivist East Asian markets, with Korea, Japan, and Taiwan showing little to no momentum profit. The behavioral story is that momentum is driven partly by overconfidence and self-attribution bias, traits more pronounced in individualistic cultures. Where investors herd toward consensus rather than their own conviction, the slow-drift underreaction that fuels momentum is muted.

There’s a structural layer too. Korean equity flow is dominated by retail investors to a degree unusual among developed markets, and retail flow tends to be contrarian at short horizons. Korean retail investors are famous for buying the dip aggressively, which is reversion behavior, not trend behavior. That same dynamic I mentioned in the mean reversion piece, the one that makes short-horizon reversion clean in Korea, works directly against short-horizon momentum.

So the practical picture in Korea is roughly this. At very short horizons, days to a few weeks, reversion dominates, because retail buys dips and sells rips. At the classic momentum horizon of 3 to 12 months in individual stocks, the effect is weak and unreliable. But trend following on the KOSPI 200 index and on KRW futures at the index and macro level still works reasonably, because index-level and currency trends are driven by foreign institutional flows and macro forces, not by the contrarian retail crowd picking single names.

That split is the actionable insight. Single-stock momentum in Korea is a trap that looks like it should work and mostly doesn’t. Index and currency trend following is a different animal and has held up better.

A short list of ways it goes wrong

Trend following has failure modes, and they’re brutal precisely because they’re the mirror image of mean reversion’s.

The big one is the whipsaw. In a sideways, choppy market with no real trend, a trend system gets chopped to pieces. It buys every breakout that immediately fails and sells every breakdown that immediately recovers. Death by a thousand small losses. Mean reversion loves choppy markets. Trend following hates them, and most of the time the market is choppier than it is trending.

The second is the long drought. Trend following makes most of its money in a handful of big trends, and between those it bleeds slowly. Practitioners talk about trend systems being flat or down for years at a stretch. Many CTAs had a famously hard stretch through much of the 2010s, when central bank intervention smothered the kind of sustained trends the strategy needs. You have to survive the drought to be there for the payoff, and most people emotionally can’t.

The third is the size of individual losing trades. Because you wait for a trend to break before exiting, you always give back a chunk of profit at the top, and you take a full loss on every breakout that fails. The win rate is often below 50%. Trend followers win infrequently but big, and lose frequently but small. If you can’t stomach being wrong most of the time, the strategy will break you psychologically long before the math does.

And the fourth, same as every quant strategy: crowding. Once a thousand funds run the same breakout rules, the breakouts get front-run and the edge thins. The returns to published momentum strategies have shrunk since the 1990s. The effect is real but it is not as fat as the back-tests promise.

What I’d take away from all this

A few things I keep in mind when I think about trend following.

Know which regime you’re in before you pick a strategy. Trend following and mean reversion are tools for opposite environments. Running a trend system in a choppy market, or a reversion system in a strong trend, is how you lose money with a strategy that “works.” The regime question comes before the indicator question, always.

Respect the psychology. Trend following is mechanically simple and emotionally savage. Low win rate, long droughts, giving back profits at every top. The Turtles succeeded not because the rules were secret but because they followed them when it hurt. Most people won’t, which is arguably why the edge survives at all.

For Korea specifically, drop single-stock momentum and look at the index and macro level. The cultural and structural reasons Korean single names resist momentum are well documented, and fighting them with a US-style winners-minus-losers stock screen is a slow way to lose. The KOSPI 200 and the won trend like other macro assets. Individual Korean stocks, less so.

And the honest closer: nobody knows how much of trend following’s century-long record will survive the next decade of crowding and central bank distortion. The data is strong. The future is not the past. I keep a small allocation to the idea, mostly for the crisis-alpha property, and I try not to confuse a good back-test with a promise. Where it goes from here, I’ll be watching, same as everyone else.

References

• Chui, A. C. W., Titman, S., and Wei, K. C. J. (2010). Individualism and momentum around the world. Journal of Finance, 65(1), 361–392.
• Hurst, B., Ooi, Y. H., and Pedersen, L. H. (2017). A century of evidence on trend-following investing. Journal of Portfolio Management, 44(1), 15–29.
• Jegadeesh, N., and Titman, S. (1993). Returns to buying winners and selling losers: Implications for stock market efficiency. Journal of Finance, 48(1), 65–91.
• Moskowitz, T. J., Ooi, Y. H., and Pedersen, L. H. (2012). Time series momentum. Journal of Financial Economics, 104(2), 228–250.
• Rhea, R. (1932). The Dow Theory. Barron’s.
• Hamilton, W. P. (1922). The Stock Market Barometer. Harper & Brothers.

Disclaimer: This article is for educational purposes only and does not constitute investment advice. The author is not a licensed financial advisor. Trend following strategies carry real risks including extended drawdowns, low win rates, whipsaw losses in non-trending markets, and the erosion of historical edges through crowding. Past performance and back-tested results do not guarantee future returns. Consult a qualified financial advisor before making investment decisions.