Your estimated 1RM is a trend line, not a number.

The number dropped, so you assumed you did

Five reps at 140 that usually move clean, and today the bar dragged off your chest like it was bolted to the floor. You rack it, open the app, and the estimated 1RM it stamps on the session is four kilos lower than last week. So now the question sits there: did you actually get weaker, or did you just sleep badly, skip a meal, and warm up cold?

Most lifters answer that question one of two wrong ways. Some start max-testing every couple of weeks to "check," which shreds their programming. The rest stop looking at e1RM entirely, because a number that swings four kilos on mood is a number you learn to ignore. Both reactions come from the same mistake: reading a single reading as if it meant something on its own.

The slope is the signal. The stamp is weather.

Here is the reframe the whole piece runs on: a lone estimated 1RM is a weather reading, and the slope of many of them is the climate. One reading tells you what the sky did on Tuesday. Six to eight weeks of readings, plotted, tell you whether the block worked.

That distinction is not pedantic. e1RM is a modeled quantity, not a measured one. You never lifted that number. The formula inferred it from a set you actually did, and every input feeding that set (sleep, food, hydration, caffeine timing, how cold the room was, how honest your RPE was) rides along in the estimate. Sampled once, all that noise is baked into the figure and you cannot separate it out. Sampled twenty times across a block, the noise averages toward zero and the real direction of travel surfaces as a line you can read at a glance.

Why the badge framing gets it backwards

Open almost any strength tracker (Strong, Hevy, Jefit, the spreadsheet you outgrew) and estimated 1RM shows up the same way: a single big figure, stamped on a date, sometimes with a little confetti when it ticks up. That presentation quietly tells you the number is an achievement, an event that happened on that day.

It is not an event. It is one sample from a noisy process, dressed up as a milestone. When the badge goes up you feel strong; when it goes down after a rough session you feel like you regressed, and neither feeling is warranted from a single point. The framing manufactures a feedback signal precise enough to react to and too noisy to trust. That is worse than no number, because it invites exactly the compulsive re-testing or total avoidance that breaks the feedback loop you wanted in the first place.

Epley or Brzycki matters less than you think

Lifters lose hours arguing which formula is "right." Epley estimates your max as the weight times one plus reps over thirty. Brzycki uses weight times thirty-six over thirty-seven minus reps. Run the same top set through both and they disagree, but look at how they disagree.

For a 100-unit top set the two formulas sit within roughly four units of each other across the normal 2-to-8 rep range, and they actually cross near ten reps. More to the point, they move together. If your logged sets trend up over a block on Epley, they trend up on Brzycki too, at almost the same slope. The formula sets the intercept; your training sets the slope. So pick one, Epley or Brzycki, and never switch mid-cycle. A consistent formula applied to a dense log gives you a clean trend. A precise formula applied to two data points three weeks apart gives you a coin flip.

What eight logged weeks actually looks like

Take a lifter on a 5/3/1 bench block, logging every working set with honest RPE. Week one his top-set e1RM lands around 138. Week two he slept four hours before the session and it reads 133. If that were his only data point since week one, he would swear he detrained. Week three he is fresh and it reads 141. Week four, a deload, dips by design. Weeks five through eight climb: 140, 143, 142, 145.

Stare at any single week and the story is noise: down, up, down, up. Fit a line through all eight and the story is a quiet, unmistakable upward slope of a couple units a week. The week-two crash you panicked over is a single dot sitting below a line that is clearly rising. You did not regress in week two. You had a bad night inside a good block, and only the density of the log let you tell the difference.

A single e1RM is a mood reading. Six weeks of them is a measurement.

Read the slope and your programming stops flinching

If e1RM is a trend, three things change about how you train. You stop max-testing to "confirm" progress, because the trend already confirms it from sets you were doing anyway. You stop overreacting to one heavy top set that felt bad, because you can see it sitting below a rising line. And you get a real deload trigger: when the slope goes flat across three or four sessions despite the same effort, that is a plateau you can act on, not a single stalled week you got spooked by.

That last one is the actual payoff. A plateau is not one bad session and it is not a number failing to tick up once. It is a slope that stops rising while your input stays constant. You can only see that if you have enough points to draw the line.

Buried history shows you a number. A dense log shows you a slope.

This is where log density stops being a technicality and becomes the whole game. You cannot fit a trustworthy trend line to sparse data. Apps that treat e1RM as a badge tend to surface the last reading and bury the history two or three taps deep, which is exactly backwards, because the history is the signal and the last reading is the least informative point in it.

Platepusher plots every working set's e1RM against time, on one formula, and shows you the slope of the block rather than the stamp of the day. Same math the calculators use, applied to a log dense enough to smooth. The chart runs while you rack the bar. When the last block actually worked, you see it as a line pointing up, not a number you have to talk yourself into trusting.

What we're watching next

The open question is how few sessions it takes before a slope is trustworthy for a given lifter, and how much honest RPE tightens that. A lifter who logs RPE straight will have a cleaner line than one who rounds every hard set down to an 8. The more the log knows about how a set actually felt, the sooner the trend becomes readable. That relationship, effort-honesty in, signal-latency out, is the thing worth measuring next.