You hit the number. The one you'd been circling for two months. The bar moved, the rep locked out, and for about an hour you felt like a different person.
Then Thursday came. The same lift, the same weight you'd handled so easily, and it crawled. You were sure something was wrong. Maybe you'd peaked and it was all downhill, or you'd gotten lucky once and the truth was catching up with you.
Nothing was wrong. You were looking at one of the most reliable and least intuitive effects in statistics, and almost nobody warns lifters about it.
The story we tell about a bad day after a good one
Humans are causal-storytelling machines. When a number drops, we look for a reason: the sleep, the carbs, the program, a character flaw. When a number rises, we credit the new cue, the new supplement, the new us.
Sometimes those stories are right. Often there's no story at all, only arithmetic.
The arithmetic is called regression to the mean. Francis Galton noticed it in the 1880s when he saw that unusually tall parents tended to have children who were tall, but not quite as tall. He first called it "regression towards mediocrity." Extreme results tend to be followed by less extreme ones, with no force pulling them back.
Where it comes from: ability plus noise
Every performance you log is two things added together:
- Your underlying ability, meaning what your body can reliably do right now.
- Noise, meaning everything that wobbles around that ability on a given day. That includes sleep, hydration, how the warm-up went, a bar that felt oddly balanced, a good song, a bad commute, and the coffee.
Over months, ability changes slowly. Noise changes every session, and it goes both ways.
Now think about what a personal record is. A PR is the single highest number in your history. For that to happen, you almost certainly had two things at once: real ability and a favorable roll of the noise. Everything lined up.
The next session, ability is still there, but the noise gets rolled again. And the chance of rolling a second extreme in the same direction is small. Most of the time you'll land closer to your typical level, which is below the peak. That isn't decline. It's the lucky part of the PR failing to repeat itself.
The pilots who were scolded into mediocrity
Daniel Kahneman tells a story in Thinking, Fast and Slow about teaching flight instructors in the Israeli Air Force. The instructors believed that praising a cadet after an excellent maneuver made the next attempt worse, while yelling after a poor one made the next attempt better. So they concluded that criticism works and praise backfires.
Kahneman's point was that the pattern needed no cause. An unusually good landing is usually followed by a more ordinary one. An unusually bad one is usually followed by a more ordinary one too. The instructors saw regression and mistook it for the effect of their own words.
Lifters do the same thing to themselves. After a great session, we see the lower number that follows and assume we did something wrong, whether we celebrated too hard, changed our routine, or lost our edge. After a terrible session, we change something, the next one is better, and we credit the change.
The two expensive mistakes
Once you see this, you can see the damage it does.
Mistake one: overreacting to a bad session. You miss a rep you expected to hit, so you rewrite the program, drop the weight, add a deload nobody needed, or decide the plan has stopped working. You've reacted to a single noisy data point as if it were a trend.
Mistake two: overreacting to a great session. You hit a PR and decide that's your new baseline. Next week you load it as your working weight, grind, stall, and feel like you've failed. You were never at that level reliably. You'd visited it once.
The cruel part is that both mistakes feel like diligence. Reacting feels like taking your training seriously.
What actually carries information
A single session is a weak measurement. A pattern is a much stronger one. The question to ask is not "was today better or worse than last time?" It's "which direction are the last several sessions heading?"
It helps to think in terms of a range rather than a point. If your top set on a lift has recently landed between, say, 5 and 8 reps at a given weight, then a 5 and an 8 are both normal. The 8 isn't proof you've leveled up, and the 5 isn't proof you've broken something. What matters is whether the whole band is creeping upward over a month or two.
Two things would tell you something is genuinely wrong:
- The entire band shifts downward and stays there across multiple weeks, not just across one session.
- The dip comes with other signals, like persistent pain, poor sleep that doesn't resolve, or illness.
One soft Thursday after a Monday PR is neither of those.
Why this is really good news
There's a quiet kindness in this idea. It means your worst days are not as informative as they feel, and neither are your best ones. You can have a session that goes badly and let it be only that. You aren't a fraud because you couldn't repeat the PR. You aren't "back" at some imaginary baseline either.
It also explains why progress feels jagged even when it's real. A lifter improving steadily will still produce a scatter of numbers, and the line only shows once you stop staring at individual points.
Your next moves
- Write down the range, not just the best. For each main lift, look back over your last four to six sessions and note the lowest and highest top-set result. That band is your real current level. Treat anything inside it as normal.
- Set a "two-sessions" rule before you change anything. Decide now that you won't alter weights, volume, or exercises after a single disappointing session. Wait for the same signal across at least two or three sessions in a row.
- Don't load your PR as the new working weight. After a PR, start the next block of work a notch below it and earn your way back up. Think of the PR as proof of a ceiling, not a floor.
- Compare monthly bests, not weekly ones. On the first of the month, look at your best top set on each lift for the previous four weeks against the four weeks before that. That comparison smooths out the noise and shows the actual direction.
- Log a one-line note on strange days. When a session feels unusually good or bad, jot the reason you suspect, such as short sleep, a skipped meal, or travel. Over time you'll see which dips had a cause and which were just noise, and you'll stop inventing explanations for the second kind.
Letting the line speak
The reason this is hard in practice is memory. We remember the PR vividly and the average sessions barely at all, so the high point becomes the yardstick against which everything else looks like loss. You can't see a trend through memory. You need the record.
That's the part where a tracker earns its place. Rep is a fast, clean strength log where your sets, PRs, and history sit in one place, so when a Thursday disappoints you, you can scroll back and see the real shape of the last two months instead of arguing with a feeling. It's a one-time purchase, with no subscription. If the idea in this article was useful, a record you can trust is what lets you act on it.