A deload is a planned lighter stretch that lets fatigue clear. Kaltro never schedules one by the calendar; it reads patterns in the sets you log. It watches three signals in your log: reps sliding at the same weight, a long time away from a lift, and logged effort staying high across consecutive sessions. Any one of them, sustained, can bring a lighter suggestion — or, when there is nothing lighter to load, a hold at your last weight. If two signals fire at once, the lighter of the two suggestions is the one you see.

Autoregulating from what actually happened — rather than a fixed weekly schedule — is a well-supported principle (Graham & Cleather, 2021). The specific thresholds below are our design choices: informed by the evidence, chosen by us, and tunable. Read the plain-English term in the glossary: what a deload is.

A deload always comes first

Before any weight-increase logic runs, Kaltro checks the three deload signals. If one fires, the deload wins and pre-empts progression — you will never be told to add weight in the same breath as a stall.

A deload surfaces both as an in-workout suggestion and as a short advisory when you open the app. It is a suggestion, not a command; you decide whether to take it.

Signal 1 — reps sliding at the same weight

This is a like-for-like comparison. Kaltro compares your reps at the heaviest weight you worked, session against session, and only against sessions that topped out at that same weight — not your effort, and not a rolling average across different weights.

  • It compares last session’s average reps at your heaviest working weight against the median of your most recent sessions that topped out at that same weight (within about 2.5%, and never lighter than last session) — up to three of them — not against a rolling average across different weights. When you add weight, your reps dropping back to the bottom of your range is the progression working, so a session at a lighter weight is never the yardstick. Only the sets at that weight count on either side, so lighter back-off or pyramid sets never water the comparison down — they still count toward your volume.
  • It needs enough data to be sure: at least 2 prior sessions that topped out at that same weight, and at least 3 sessions of the lift overall (4 while you are in a calorie deficit).
  • The drop has to be real: average reps at that weight down 25% for compound lifts, or 20% for isolation, before this triggers.

What the lighter weight is

When rep regression or sustained high effort calls a deload, the cut is modest and rounded to what you can load.

  • Typically around 10% lighter — the weight drops to roughly 90% of where it was. That 90% is taken from your first working set of the lift last session, because that is the set the suggestion applies to: if you build up across your sets, it is your opener, not your top set.
  • Rounded down to a real increment — and the suggested weight never drops below a single loadable increment. If rounding would erase the reduction entirely, Kaltro holds off rather than pretend.

Signal 2 — time away

Come back after a real layoff and Kaltro eases you in with a lighter restart — or, when there is nothing lighter to load, holds you where you were — instead of pushing on from your old working weight. Research on detraining and retraining supports lowering the load after a break (Bosquet et al., 2013); the exact day counts and reductions are ours.

  • Around two weeks away triggers it. The exact entry point varies by experience level — about 10, 14, or 18 days for advanced, intermediate, and beginner lifters, once Kaltro knows which you are. The same entry point also decides whether Kaltro will speak at all about a lift it has seen only once or twice.
  • The longer you were gone, the lighter the restart. It scales from typically around 10% lighter after a couple of weeks toward much larger reductions — down to roughly 65% lighter after a year away.
  • Bodyweight lifts with nothing added aren’t made easier — Kaltro holds you at last session’s reps for your first session back instead of asking for more. If you log added load, such as a plate on a dip belt, that added load is cut on the same time-away scale and rounded down to whole 1.25 kg or 2.5 lb plates; if the cut would come to less than one plate, Kaltro keeps your last load.
  • On bar-loaded lifts (barbell, EZ bar, trap bar, Smith machine) the restart never goes below the empty bar. If the cut would land under the bar, Kaltro suggests the bar itself — using the bar weight in your equipment settings. If you were already lifting the empty bar, Kaltro keeps you there.
  • If rounding would wipe out the reduction, Kaltro skips it and keeps your last weight — no token change dressed up as a deload, and no push to add weight on your first session back.
  • While a lift carries an injury flag you set in the app, Kaltro offers no restart weight for it — you choose the load. Once you clear the flag, a restart weight can be offered again. If you had flagged the lift as Injured, Kaltro’s in-workout suggestion may first pace your return from the injury, with your time away counting toward that pace.
  • After a shorter break, Kaltro helps you climb back. For breaks up to about 12 weeks, once the lighter restart is behind you, Kaltro steps back toward a weight your log shows you lifting in two sessions in the six weeks before the break — the same weight both times, or within about 2.5% of it. It moves one capped step per session that meets its rep-and-effort target. A session that falls short holds the weight. Two sessions in a row that fall short at the same step, or any drop in weight — including taking a deload Kaltro calls — ends the climb, and the normal rules take over. It never goes above that weight, or above the heaviest load you have told Kaltro you own for that equipment. It does not apply to beginners, to lifters whose training age is set to Returning, while your nutrition phase is a deficit, to a lift coming back from an injury flag, when your recent logs already show more than one sign of a stall across your training at once (such as estimated strength slipping from its recent peak, logged effort rising at the same weights, or sets cut short), or to bodyweight lifts with nothing added. It also does not run at any weight where even the smallest step your equipment allows is larger than the cap on a single jump; the normal rules take over there.

Signal 3 — effort staying high

If you log RPE, Kaltro watches for the effort you log staying at or above a trigger line across consecutive sessions (Zourdos et al., 2016). This is read entirely from your logged numbers — Kaltro does not sense fatigue or detect overtraining.

  • Sustained high effort triggers a deload: an average RPE at or above your trigger line across 2 consecutive sessions for beginners, 3 for intermediate lifters, or 4 for advanced lifters — one more session while you are in a calorie deficit, since elevated effort during a cut is often diet-driven. That trigger line varies by lifter and by experience: never lower than an average RPE of 9, never higher than 9.5.
  • Every one of those sessions has to have an RPE logged — Kaltro never fills in a missing effort score, so this only fires on data you actually recorded.
  • Advanced lifters have one extra guard: if effort rises steadily while median reps fall across 4 sessions, Kaltro can call a deload even before effort reaches the trigger line.

Weekly set counts are not a fourth signal

Kaltro also counts the resistance-training working sets each muscle group gets in a week, and compares that count against a published volume-guidance range. That comparison never calls a deload — it never drops the weight on the bar.

  • The three signals above are the only deload triggers. A weekly set count — whatever the number — records no deload, opens no gate, and changes no suggested weight. Most of what Kaltro shows next to the count is an observation about what you did. Sometimes it is a suggestion: when a group’s weekly sets have stayed above the productive range for weeks, and your own logged response agrees — reps coming down at the same working weight, or the effort you log holding high or climbing — Kaltro can suggest easing those weekly sets back into the range for about a week. Kaltro makes that suggestion only once it has enough of your own logged weeks behind the range.
  • The range it points back to is partly yours. Kaltro starts from published volume guidance and moves the range only where your own logged weeks give it evidence; until then, the published numbers stand.
  • Counting sets never talks over measuring response. A set count says how much work you logged. The three signals say how you responded to it. Where the two could disagree, the response decides — a high count on its own stays an observation. When several muscle groups are in that position at once, across different parts of your training, Kaltro makes the point once about the week instead of naming a group: a deload week looks worth taking — a week of easing the working sets back, not a cut to the weight on the bar.
  • Anything Kaltro says about weekly set volume stays about sets — never about the weight on the bar, and never the reason a lighter weight was suggested.

One hard session never triggers a deload. The rep and effort signals look for a sustained pattern; time away is measured in weeks, never one workout. A bad night’s sleep or a single grinder is noise, and Kaltro treats it that way.

Let Kaltro watch the patterns for you

Log your sets and Kaltro tracks these signals in the background, then tells you when a lighter stretch is the smarter call — and shows you why.

Keep reading: how Kaltro decides when to add weight, how much history it needs first, and how it reads your e1RM trend. Terms in plain English: deload, RPE, and recovery. To see all four outcomes in the app, read how Kaltro works.

The research behind this

  • Graham & Cleather (2021). Journal of Strength and Conditioning Research.
  • Bosquet et al. (2013). Scandinavian Journal of Medicine & Science in Sports.
  • Zourdos et al. (2016). Journal of Strength and Conditioning Research.