Imagine a student who scores brilliantly on one exam, then drops back to an average grade on the next. A coach scolds a player after a terrible game, and the player improves. A new supplement makes you feel amazing the first week, then the effect seems to fade. In each case, we instinctively reach for a cause—complacency, the power of tough love, tolerance to the supplement.

But what if none of those explanations are correct? What if the most likely reason extreme outcomes are followed by ordinary ones is simply… mathematics? This is the principle called regression to the mean, and understanding it may be one of the most practical upgrades you can make to your everyday reasoning.

Most things we measure—test scores, athletic performance, daily mood, business results—are influenced by a combination of stable factors and random fluctuation. Your true ability on an exam might correspond to a score of 75, but on any given day, luck plays a role. Maybe the questions happened to align with what you studied, or you slept unusually well. The result? You score 92. That's a genuine score, but it sits far from your average because chance pushed it to an extreme.

Here's the key insight: the more extreme an outcome, the more likely it is that random variation contributed to making it extreme. This doesn't mean skill or effort don't matter. It means that on any single occasion, the gap between an observed result and someone's typical performance is partly signal and partly noise. The noise doesn't repeat reliably.

So what happens next time? The random factors shuffle again. They're just as likely to push your score down as up. Without that lucky tailwind, your next result drifts back toward your actual average. Nothing changed about your ability. Nothing went wrong. The mathematics of probability simply corrected itself. This is regression to the mean, and it happens whenever outcomes involve any degree of randomness—which is nearly always.

Takeaway

When you see an extreme result, ask yourself how much of it might be noise. The more extraordinary the outcome, the more you should expect the next one to look ordinary.

This is where regression to the mean becomes genuinely dangerous to clear thinking. Consider a famous example from the psychologist Daniel Kahneman. Israeli flight instructors noticed that when they praised cadets after an exceptionally smooth landing, the next landing was usually worse. When they harshly criticized a rough landing, the next one was usually better. The instructors concluded that punishment works and praise backfires. It seemed obvious from the evidence.

But it wasn't true. Exceptionally good landings were partly lucky—so the next attempt naturally regressed downward regardless of praise. Exceptionally bad landings were partly unlucky—so the next attempt naturally improved regardless of criticism. The instructors' feedback was irrelevant to the pattern, yet it felt deeply relevant because it always preceded the change. This is illusory causation: we insert a narrative of cause and effect into what is really a statistical inevitability.

This trap is everywhere. A city installs speed cameras at its most dangerous intersections, and accidents decline—but accidents at those spots were already statistically likely to decrease. A patient takes a remedy when symptoms peak, then feels better—but symptoms were already at their worst and due to ease. Regression to the mean provides the real explanation, but our brains are wired to prefer stories over statistics. Recognizing this bias doesn't make you cynical. It makes you careful.

Takeaway

Whenever an intervention follows an extreme event and things 'return to normal,' ask whether regression alone could explain the improvement before you credit the intervention.

Once you understand regression to the mean, you gain a practical tool for making better predictions. The principle is straightforward: when forecasting a future outcome, adjust any extreme observation back toward the average. If a rookie baseball player hits .400 in April, don't expect .400 for the season—expect something closer to the league average. If a business has a record-breaking quarter, budget for a more typical one next.

How much should you adjust? That depends on how much randomness is involved. In activities with high variability—like single exam scores, individual games, or monthly sales figures—regress heavily toward the mean. In activities with low variability—like a student's GPA over four years or a company's decade-long revenue trend—the observed result is more trustworthy and needs less adjustment. The noisier the measurement, the less you should trust any single extreme reading.

This doesn't mean excellence is an illusion or that standout performances are meaningless. It means that a single data point tells you less than you think. The remedy is patience: gather more observations before drawing conclusions. If a teacher seems exceptional after one semester, wait for three. If a new policy seems to fail in its first month, give it a year. Regression to the mean is not an argument against action—it's an argument against premature judgment.

Takeaway

The noisier the process, the more any single extreme result will mislead you. Before reacting to one exceptional data point, ask how many observations you'd actually need to draw a reliable conclusion.

Regression to the mean is not an exotic statistical curiosity. It's a quiet, constant force shaping outcomes all around us—in schools, hospitals, sports, business, and our personal lives. When we ignore it, we build false stories about what works and what doesn't.

The practical takeaway is a habit of thought: before explaining why something changed, consider whether it simply regressed. This single question can save you from superstition, bad policy, and wasted effort. It won't make you a pessimist about excellence—it will make you a realist about evidence.