Cross-market Game Theory | A Brief History of Stock Market Pullbacks

Wall Street is putting on a self-congratulatory play. War, a surge in oil prices, the re-ignition of inflation expectations, the ghost of a Federal Reserve rate cut pause, an AI money-burning competition, widening labor market cracks, and the software industry’s collective passage through a life-or-death trial—lay this list in front of any stock-market expert, and he’d probably guess that the S&P 500 will have dropped by at least 15%. But the reality is that the index has only slipped by less than 9% from its all-time high.

In fact, this doesn’t even count as a correction. Because it hasn’t even reached the threshold for a “correction” yet.

Why not set up a stock-market casualty grading system that isn’t scientific, but is extremely practical: a 5% drop is a pullback—normal market digestion; a 10% drop counts as a healthy correction—like a doctor furrowing his brow during an annual checkup; a 15% drop is worthy of being called a correction—investors start sleeping poorly; a 20% drop moves into bear-market territory—finance-show hosts’ tones begin to tremble; a 30% drop is a collapse—you start losing so badly you doubt your life; a 40% drop is a crash—your wife starts doubting you; a 50% drop is a crisis—everyone suspects that the end is nigh. With that, a decline of fewer than 9 percentage points is, at most, the market giving a rather loud sneeze.

But sometimes a sneeze is the prelude to pneumonia, and sometimes it’s just pollen allergies. The problem is: you can never tell in the moment.

The data provides a few sobering anchor points. Among the 37 complete calendar years since 1990, there have only been 3 years in which the year as a whole saw no pullback of at least 5%—namely 1993, 1995, and 2016. In other words, a 5% drawdown isn’t an accident; it’s the norm. It’s the “entry fee” you pay for holding stocks. Zoom the perspective out even more. From 1928 to today, the S&P 500 has gone through 56 declines of 10% or more—about once every 1.8 years, with a frequency rivaling U.S. presidential midterm elections. There have been 30 declines that broke below 15%, 22 that broke below 20%, 19 that broke below 25%, 13 that broke below 30%, and 7 that worsened past 40%—while truly breaking below 50%, putting the entire financial system near a stop of the heart—has occurred only 3 times. Reducing 56 instances to just 3 is, by itself, a profound statistical narrative: most fears ultimately die a very ordinary death.

What’s really worth pondering, however, isn’t the absolute number of occurrences at each level, but their “contagion rate.” History tells us that once the market falls below 10%, there’s a 54% chance it will continue worsening to 15%—basically flipping a coin. But once it slides to 15%, things start getting seriously off: a 73% probability that it will keep falling through the bear-market threshold of 20%. This is the most dangerous turning point in the entire chain—three-quarters of the “corrections” after reaching this depth failed to hold their ground. After that, the chance of 20% evolving into a 30% collapse is 59%; the probability of 30% sliding into a 40% crash falls back to 54%; and the chance of 40% worsening into a 50% crisis drops to 43%.

These figures sketch an asymmetric map of risk. Between 10% and 15% there’s a blurry dividing line, and the market has nearly a half chance of stopping the bleeding on its own. But 15% is a psychological and liquidity double threshold; once crossed, momentum, leverage liquidations, and panic selling reinforce each other in a self-strengthening spiral. In other words, today’s decline of less than 7% is still separated from the truly dangerous zone by an entire buffer band of a “healthy correction.” That’s good news. The bad news is that the width of the buffer band is meaningless in the face of panic—back in March 2020, it took the market only 23 trading days to go from the peak to the bottom. It was so fast that the phrase “healthy correction” didn’t even have time to escape analysts’ mouths before it was already thrown into the abyss of the bear market.

So the question has never been “whether it will fall deeper,” but “when.” Measured on a sufficiently long time horizon, every pullback ultimately runs into a real bear market, and every bear market ultimately collides with a collapse. This isn’t pessimism; it’s arithmetic. Since 1928, the S&P 500 has averaged a bear market of the 20% category about once every five years. The last one was in 2022—if you believe in mean reversion, the next statistical window likely opens around 2027. Of course, the distance between the statistical window and reality is sometimes farther than Earth is to Mars.

The narrative list the market faces right now is indeed frightening: geopolitical risk is driving up energy costs; higher energy costs are pushing up inflation expectations; inflation expectations are tying the Federal Reserve’s hands; and the astronomical scale of AI capital expenditures is quietly transforming from a “growth story” into doubts about a “cash-flow black hole.” A re-pricing of valuation in the software sector is another heavy blow—when the market begins to distinguish between companies “benefiting from AI” and those “being replaced by AI,” a large number of SaaS companies find themselves standing on the wrong side of the blade. Taken individually, each of these is enough to write a panic report. But stacked together, they’ve only delivered a decline of less than 9 percentage points—which in itself is an intriguing signal.

It may mean the market has already priced in most of the bad news, liquidity remains ample, and buyers stay active on dips. It may also mean the market hasn’t truly understood the tail-end effects of these risks, and that the current shallow pullback is just that eerie calm before the storm. Both interpretations make sense—and that’s exactly where the market is most ruthless: it never gives you certainty precisely when you need it.

Fundamental analysis at a time like this is almost meaningless. When the market enters a downtrend mode, what drives prices is no longer the corporate earnings model or discounted cash flows, but the two oldest forces in human nature: the tug-of-war between greed and fear. An earnings report that beats expectations can’t stop a wave of panic selling; a bad employment data point may not be enough to curb the impulse to buy the dip. In the short term, the only variable that matters is the direction of investors’ collective behavior, and no model can predict that variable.

This is the fundamental reason the stock market offers a long-term risk premium. What you get isn’t free money, but compensation paid for accepting short-term unpredictability. Every pullback is a bill—reminding you that the entry fee for this game was never cheap. Some years’ bills are small, like 2026 so far—no worse than a mild sting. Some years’ bills are large enough to make you question your life, like 2008 or 2020. But the bills always come.

That line will never keep extending to the upper right at a 45-degree angle. If it did, everyone would buy stocks, the risk premium would be compressed to zero, and the word “investment” would lose all meaning. It’s precisely because that line periodically bends downward, sometimes breaks, and only in rare cases plunges into the abyss, that long-term holders can receive the returns they deserve.

Don’t forget: when every crash begins, it always looks like a correction. The difference is that you only know after the fact. And what the market is best at is—at the moment when everyone thinks they’ve figured things out, it slaps you in the face hard.

Stay positioned, buckle up, and accept one unsettling fact: you’ll never know whether this time was just a sneeze or pneumonia. That’s the price of investing in the stock market.

Xu Liyan (This column is published every Monday)

https://www.facebook.com/hsulylab