Short-Term Momentum in Stocks - Why Winners Keep Winning
Every trader has said something like this at some point: “The stock has good fundamentals, it’s cheap, but the price action looks terrible. I’m going to wait.”
That sounds irrational, right? If a stock is cheap, you buy it. Who cares what the price did last month? But here’s the thing. There’s actual research showing that recent winners tend to keep winning. And recent losers tend to keep losing. At least in the short term.
Chapter 16 of Burton and Shah’s book is about this exact phenomenon. It’s called momentum. And it comes in two flavors.
Price Momentum: Winners Keep Winning
In 1993, two researchers named Jegadeesh and Titman published a paper that made Wall Street traders feel very smart. They showed that stocks which performed well over the past 3 to 12 months tended to continue performing well going forward.
They tested a simple strategy. Take the last six months of data. Buy the winners. Sell the losers. Hold for six months. The result? This strategy beat the market by about 12 percent per year.
That’s a huge number. And it confirmed what traders had been saying for years. Stocks with recent uptrends tend to keep going up. Stocks with recent downtrends tend to keep going down.
But wait. Didn’t we just learn in the previous chapter about mean reversion, where winners eventually become losers and vice versa? How can both be true?
The answer is time horizon. Mean reversion happens over years. Three to five years, typically. Momentum works over months. Three to twelve months. So a stock can ride momentum upward for several months, and then over the next few years, revert back down. Both effects are real, but they operate on different timescales.
So Why Does Momentum Happen?
Jegadeesh and Titman had a theory. They suspected it had something to do with earnings announcements. When a company reports good earnings, the stock goes up. But maybe it doesn’t go up enough. The market underreacts to good news. So the stock keeps drifting upward for weeks or months after the announcement as the full impact slowly gets priced in.
They tested this and found some confirmation. Good earnings announcements did predict continued good performance. But only for a few months. After that, the loser stocks actually started catching up and eventually outperformed the winners.
This is where things get interesting. Because it points to a second type of momentum.
Earnings Momentum: The Ball and Brown Discovery
Way back in 1968, two accounting researchers named Ray Ball and Phillip Brown found something curious. Stock prices start anticipating earnings surprises about 12 months before the announcement. The market slowly figures it out. But here’s the key part: the price keeps drifting for about a month after the announcement too.
That post-announcement drift is earnings momentum. The market gets the news, reacts, but doesn’t fully react. Prices keep moving in the same direction for weeks after the surprise.
Ball and Brown needed a way to define what counts as “surprising” earnings. They built a statistical model. The best predictor of how much a company’s earnings will change is how much earnings changed on average across the whole industry. If actual earnings come in higher than what the model predicts, that’s a positive surprise. If lower, negative.
And the drift worked both ways. Positive surprises led to continued price increases. Negative surprises led to continued decreases.
How Do You Measure an Earnings Surprise?
Ball and Brown used a statistical model because in 1968, there weren’t many Wall Street analysts publishing earnings forecasts for individual companies.
By the mid-1990s, that changed. Plenty of analyst forecasts were available. So researchers started asking: should we use statistical models, or what the actual analysts are predicting?
The most popular statistical measure is called SUE, Standardized Unexpected Earnings. It compares this quarter’s earnings to earnings four quarters ago and then adjusts for how volatile those earnings typically are.
Here’s the funny thing. SUE, the statistical measure, actually performs better at predicting future price movements than the real analyst forecasts. You would think that actual human analysts who study these companies full time would produce better predictions. But the simple statistical formula wins.
The authors point out something important here. SUE doesn’t match what real market participants mean by “expected earnings.” On Wall Street, expected earnings means the consensus estimate from analysts. Not some formula comparing this quarter to last year. Nobody has really connected these two things convincingly. That’s a gap in the research.
Are Price Momentum and Earnings Momentum the Same Thing?
This is the big question. And researchers disagree.
In 1996, Chan, Narasimhan, and Lakonishok studied both types. They concluded that price momentum and earnings momentum are distinct phenomena. And price momentum was the bigger effect. Sorting stocks by their prior six-month returns produced return spreads of 8.8 percent over the next six months. Sorting by earnings revisions produced 7.7 percent spreads. Both are significant, but price momentum was larger and lasted longer.
Then a decade later, Chordia and Shivakumar looked at the same question and got the opposite result. They found that earnings momentum actually swamps price momentum. Their argument was that in countries where earnings momentum doesn’t exist in the data, price momentum doesn’t exist either. So earnings momentum might be the root cause, and price momentum is just a side effect.
Why Does This Distinction Matter?
It matters because of the efficient market hypothesis.
If price momentum is real, then the weak form of market efficiency is false. Period. The weak form says you can’t predict future prices from past prices. But price momentum is literally that. Past prices going up predicts future prices going up. You can’t explain that away by saying it’s some hidden risk factor. Past prices are past prices.
Earnings momentum is different. It could be capturing some real economic risk factor that the market is pricing. Maybe companies with earnings surprises share certain characteristics related to GDP growth or inflation. If so, the higher returns could be compensation for risk, not a market inefficiency.
Chordia and Shivakumar tried to make this case. They argued that stocks most affected by earnings momentum have characteristics linked to future GDP growth. So the extra returns might be a risk premium, not free money. The argument isn’t totally convincing, but it gives the efficient market theory some breathing room.
Hedge Funds and Momentum
Here’s something the academic papers can’t tell you about. Hedge funds.
Quant funds, the ones that use computers and algorithms to trade, have spent enormous resources studying price momentum. But they don’t publish their findings. For obvious reasons.
What we do know is they look at very short-term data. Holding periods of minutes or even seconds. They’re not thinking about six-month trends. They’re looking for micro-patterns in price data.
Burton and Shah raise a fair concern here. When you run enough statistical tests on historical data, eventually you’ll find patterns that look significant but are actually just noise. This is called data mining. And hedge funds are very susceptible to it. If a strategy worked on past data, there’s strong incentive to believe it will work in the future. The compensation structure practically demands it.
And there’s another problem. Many quant strategies produce steady small gains most of the time but have a tiny probability of catastrophic losses. This is tail risk. The strategy looks great on paper. It makes money month after month. Until one day it blows up spectacularly. We’ve seen this happen multiple times.
Does Any of This Actually Matter?
Here’s what we know. Price momentum exists in the data. That’s pretty well established. And it’s a problem for anyone who believes markets are efficient. You shouldn’t be able to beat the market by simply buying last month’s winners.
Earnings momentum also exists, but the research is messier. The statistical measures used by academics don’t match what traders actually mean by “expected earnings.” That disconnect needs to be resolved.
But even if you set aside the academic debate, there’s one observation that’s hard to ignore. Real traders, the people with actual money on the line, obsess over price charts. Every trading desk has them. Every financial news channel shows them. If past prices truly didn’t matter, why would the entire industry spend so much time staring at them?
Maybe those traders are all irrational. Or maybe they’re onto something that the efficient market theory can’t explain.
Either way, it’s another crack in the foundation.
Previous: Mean Reversion - Value vs Growth Stocks