Your log already shows the bad month that repeats every year.

Three Februaries, three abandoned programs

February, year one. A lifter eight weeks into a 5/3/1 block starts missing reps on a squat top set that moved easily in November. Bar speed is gone, warm-ups read a full point higher on RPE than they should, and by the third bad week the block is dead. March brings a new program. Within a month, training feels good again.

February, year two. Different program, an upper/lower hypertrophy split this time. Same slide. Same verdict: the program stopped working. Another restart in March.

February, year three. You can guess.

Three different programs took the blame for three bad winters. Lay the three years of log side by side, though, and the slump lands in the same stretch of weeks every year, under three different sets of percentages. That's a calendar signature, and the lifter has been firing programs for it.

The 12-week blind spot

Most training gets planned and judged in blocks of 8 to 16 weeks. That's the right unit for programming. For diagnosis it's a bad one, because the things that drag performance down on an annual cycle run on a 52-week period, and any single block only ever sees a slice of it.

Call it the 12-week blind spot. Inside one block, a February slump looks exactly like a program that has run its course: loads stall, RPE creeps, the lifter feels flat. The block has no way to tell those two apart. Only the same weeks from earlier years can.

So the honest read on a bad month starts with a question most lifters skip. What did this exact stretch of the calendar look like last year, and the year before that?

A 12-week block can't see a 52-week pattern.

Why does my training performance vary so much?

The standard answers are all local. Sleep was bad, work was stressful, the volume is too high, you need a deload, you need a new program. Some of those are true some of the time. Every one of them gets diagnosed from the last three or four weeks of data.

The program hop is the expensive answer. Each March restart gets a fresh baseline, fresh training maxes, a fresh spreadsheet tab. The comparison that would expose the pattern, this February against last February, never happens, because last February lives in a different program's numbers and sometimes in a different app entirely.

And the restart usually seems to work, which teaches the wrong lesson. Performance comes back in March because the seasonal drag lifted. The new program collects the credit, and the cycle is set up to repeat.

What repeats on a 52-week clock

A lifter's year runs on cycles, and so do several of the inputs that move a top set. None of them is exotic. Most lifters could name all of them. What they rarely do is check which ones actually show up in their own numbers.

The map below is generic on purpose. Which rows matter for one lifter, and how hard they hit, is personal. A population study can't answer that. A few years of that lifter's own log can.

How to read your log for a seasonal dip

The method is boring, and it works just as well in a spreadsheet as in an app. It needs three or more years of mostly unbroken logs. With two years, a coincidence can pass for a pattern.

Pick two or three anchor lifts you've trained in some form every year: a squat, a press, a hinge. For each calendar month, take the average estimated 1RM from your top sets, or the average load at a fixed RPE if your RPE logging is consistent. Then normalize each month against that year's own average. Otherwise a stronger year three hides a dip that only shows up relative to its own year.

Now line the years up by month. A seasonal dip is a month that sits below the months on either side of it in most years, on most anchor lifts. One bad February is noise. Three Februaries in a row sitting below January and March is a signature.

Last step: read the notes. If the weak month also carries the comments about illness, travel or a brutal work quarter, you've found the cause along with the timing.

Stop peaking inside your weak months

Once the weak months have names, the program stops absorbing the blame and the calendar becomes a planning input.

The prescription is narrow. Don't put a peaking block, a meet, a test day or a heavy-singles week inside a month your log says is historically soft. Put accumulation, technique work or a planned deload there, and aim the hardest intensity at the months that have held up year after year. A lifter whose Februaries sag can run volume through February and test in late April, and get a fair read on the program instead of a seasonal one.

The second change is harder. Stop killing programs in the weak month. A block that stalls in the same weeks where every previous block also stalled hasn't told you anything about itself yet. Judge it on the weeks outside the dip.

The pattern only exists if the years stay together

Every step above assumes something most lifters don't have: three or four years of training in one place, still queryable. The usual history is scattered. Year one in a spreadsheet, year two in one tracker, year three in another after a subscription lapsed and the export never happened. Each fragment is fine on its own terms. None of them can line up a February against a February.

Platepusher is built for keeping those years in one record. Drop in a CSV from Strong or Hevy and every lift, every set and every date imports as native data, on the same timeline as what you log next week. Full history and CSV export are in the free tier, so the record outlives any change of plan. The plateau flag marks the week a top set stops moving, and with the old years sitting in the same log, you can check whether that stall is new or one you've hit in this same month before.

The broader argument is in [Your training history is the experiment](/articles/your-history-is-the-experiment). The seasonal read is its most concrete case. The evidence for your worst month already exists. It has to be in one place before anyone can read it.