Deload Criteria

What a Deload Is

A deload is an adjustment method "to clear accumulated fatigue, through a temporary reduction in load, and re-drive progress." As training continues, fatigue accumulates in the nervous system, the joints, the connective tissue, and the central nervous system. When this fatigue outpaces recovery capacity, rep count, weight, and subjective margin (RIR) all deteriorate, and progress stops. A deload functions as the measure that resolves this state.

A deload comes in two broad types: an intensity deload (lowering the weight while holding rep count) and a volume deload (lowering set count or frequency while holding weight and reps). The one DELT proposes is the former, the intensity deload, which carries the practical advantage of being completed as a set-level adjustment without requiring a redesign of the whole routine.

Why It Is Needed

Progressive Overload is, at its core, a fatigue-generating activity. Strengthening the recruitment patterns of the nervous system, the damage and repair of muscle fibers, and the rebuilding of connective tissue are all adaptive processes that demand recovery. Progress appears in the periods when recovery outpaces the stimulus; fatigue accumulates in the periods when recovery cannot keep up with the stimulus.

The scoping review by Bell et al. (2020) laid out the distinction between short-term overreaching and chronic overtraining in resistance training. Short-term overreaching, if managed appropriately, draws out meaningful adaptation within the fitness-fatigue model; but once it progresses to chronic overtraining, neural fatigue, disruption of the hormonal axis, and a decline in subjective recovery become prolonged, and restoration takes anywhere from several weeks to several months. A deload plays the role of intervening at the former stage to prevent the transition to the latter.

In the volume landmarks framework of Israetel et al. (2021), the basic structure presented is an accumulation cycle in which volume climbs from around MEV up to MAV, and upon reaching MRV is reset by a deload. The deload, at the "reset" stage of this cycle, sets up the starting point for the next accumulation. For the details of volume landmarks, see Volume Landmarks and MEV/MAV/MRV.

The Two Triggers

The trigger that makes DELT automatically propose a deload is consecutive failures (a short-term signal). Separately, medium-term stagnation is caught by a stagnation-detection mechanism, but stagnation routes to a multi-option discussion flow rather than an automatic deload.

Consecutive failures are read as a signal that "the recent intensity setting is too high for current fitness," and stagnation detection as a signal that "the current program is not producing the stimulus it needs, or insufficient recovery has become chronic." The former fires on a per-session scale; the latter fires on a scale of several weeks.

Choosing the Deload Depth

DELT designs the deload depth within the 5-10% range. Too narrow and no effect appears; too wide and the loss of progress is too large.

-5%: the design value for methods that run the rep range as a "span," such as Double Progression. Because it takes only 1-2 cycles to restore the original weight, the loss of progress is small. On the premise that the buffer of rep count keeps progress judgments forgiving, a deep deload is unnecessary.

-10%: the design value for linear progression with a fixed rep target (such as 5x5). Because low-rep, high-load neural training takes time to recover from fatigue, you need to drop deeper to secure recovery headroom. It takes 3-4 cycles to restore the original weight.

The selection rule: if the strategy is Double Progression, -5%; if it is linear progression, -10% is the starting point. A deload deeper than this (-15% or more) is treated as an exceptional measure, for chronic overtraining or for returning from injury.

DELT's Staged Design

DELT's deload judgment is deliberately staged. In Double Progression, "a single failure is treated as a one-off, and a -5% deload is proposed only after two consecutive failures." In linear progression too, "a -10% deload is proposed after two consecutive failures." On the first failure it does not deload, adopting a design that re-attempts at the same weight.

There are two reasons for this staging.

First, a single failure can occur from day-to-day condition swings. With sleep deprivation, undernutrition, stress, or a hot environment, the number of reps you can lift at the same weight swings by ±1-2. A logic that proposes a deload immediately on the first failure overreacts to this noise.

Second, excessive deloading impedes progress. If you lower the weight after one failure, the next time becomes a "success," and a cycle of gradually restoring the weight again sets in. When this cycle occurs more often than progress, net progress falls to zero or negative. In the short term it looks "well managed," but viewed by the month, you fall into a state where the weight has not climbed.

A staged threshold (two consecutive) functions as the judgment headroom that distinguishes noise from a true signal. When two consecutive failures are observed, the confidence is high enough that it is a signal that "the load really is too high."

Stagnation Detection and the Discussion Flow

Consecutive failures are a short-term signal, but longer-term stagnation is detected by a separate judgment mechanism. Here DELT sets no universal threshold. There is no academic source for the N in "no update for N sessions means stagnation." What defines stagnation as a specific count is the convention of an individual programme, not a constant derived from observation or experiment.

Instead, the baseline is taken from the lifter's own record. Stagnation is the state in which the number of sessions since the last improvement exceeds that lifter's own typical interval between improvements. The typical interval is taken as the median of the past intervals between improvements.

This definition needs no source. "You normally improve once every M sessions, and you have now gone past M without improving" is a verifiable fact about the lifter's own record. The form also adapts automatically to two populations that a single constant cannot serve at once. Someone improving every session is flagged after a few flat sessions; someone improving once every ten sessions is not. Three flat sessions are abnormal for the former and routine for the latter.

Until at least three improvements have been observed, no judgment is made and the state is treated as insufficientData. Taking a median of intervals requires at least two intervals, which is a statistical floor rather than a claim from training science. This is why the judgment does not run right after starting a new routine or right after returning from a long layoff.

"Improvement" is counted in one of two ways: setting a new all-time maximum weight, or exceeding the previous rep count at a weight that has been handled before. The estimated 1RM (e1RM) is not used to define improvement. Double progression raises the weight and resets reps to the bottom of the range once the top of the range is reached, so the e1RM necessarily drops once immediately after a weight increase. The several sessions it takes to climb back above the previous value are proceeding exactly as planned, yet an e1RM criterion would count them as "no update" and misjudge normal progression as stagnation.

Restricting rep-count improvement to "a weight that has been handled before" also has a reason. A lighter weight touched for the first time has no record, so any rep count becomes a personal best at that weight. Counting that as improvement would reset the stagnation judgment every time the load is dropped for a deload or a back-off set, leaving the periods of no actual progress unmeasured.

The same rule applies unchanged to bodyweight exercises. Since the weight does not move, only rep-count improvements are counted. There is no need to split the judgment from weighted exercises.

The maintenance purpose is outside the StagnationDetector's scope. Because running a stagnation judgment against a routine whose "purpose is to maintain the status quo" is meaningless by design, it is always treated as "progressing."

Even when stagnation is detected, DELT does not deload automatically. Unlike consecutive failures — a decisive overload signal — stagnation does not pin down a single cause (insufficient stimulus versus insufficient recovery). The engine still proposes its default progression (reps +1, etc.) while surfacing alternative directions as a discussion flow. The options are "Push reps," "Add a set," "Lighter, more reps" and "Hone previous" — four in all, reduced to three for bodyweight exercises, where the load cannot be lowered, by dropping "Lighter, more reps". Suggestions in this situation are marked with an experimental confidence, and the user chooses which direction to take.

Operational Limits and Caveats

• A deload is not a cure-all: when the cause of stalled progress lies in undernutrition, sleep deprivation, chronic stress, or injury, a deload alone will not resolve it. Even if you lower the weight and recover, if the root cause remains as it is, you re-stall at the same place on the re-accumulation after the deload
• The consecutive-failure judgment compares only against the previous session: variability in single-session conditions — such as the previous session falling on the day after a rest day and going abnormally high, or the current session going abnormally low due to poor condition — is not built into the judgment. Because the consecutive-failure judgment looks only at "the relative comparison over the past two sessions," long-term trends need to be monitored separately through autoregulation metrics such as RIR
• The stagnation judgment needs a history of progress: the StagnationDetector makes no judgment until it has observed at least three improvements on that exercise. Since the criterion is the lifter's own interval between improvements, nothing can be claimed until that interval can be derived. Right after starting a new routine, or right after returning from a long layoff, it is treated as insufficientData. Understand this period as the time the record spends learning that lifter's rhythm
• When re-accumulation after a deload does not advance: if you cannot return to the original weight even after 2-3 cycles, it is a signal that what is needed is not a deload but a review of the whole program. Consider redesigning the rep range, adjusting the volume landmarks, or changing exercise selection
• Distinction from a scheduled deload: what this article addresses is the reactive deload (a deload triggered by failure or stagnation). There is also the scheduled deload (a deload inserted preventively at regular intervals), which is effective as a method of managing the accumulation of fatigue preventively. The two are not exclusive; a reactive deload can be inserted within the interval of scheduled deloads

How DELT Handles It

DELT's deload is integrated within the suggestion algorithm for each set. There is no dedicated "deload week" mode; when the deload judgment (consecutive failures) is triggered, the next suggestion comes out at -5% (or -10%). The user has the options of accepting the suggestion, ignoring it, or logging a different weight.

For transparency, a suggestion comes with a reason. With a consecutive-failure deload it shows an explanation like "a -5% deload from two consecutive misses of the bottom." With stagnation detection, instead of a deload, the discussion-flow options are presented together with the reason "the estimated 1RM has stagnated over N sessions." It is a design in which the user can decide whether to accept or reject after understanding the basis for the suggestion.

The deload depth is fixed at -5% (Double Progression) or -10% (linear progression), and floors to the unit of the weight increment. For example, at 80 kg (176 lb) with a 5 kg (11 lb) increment, a -5% deload gives 80 × 0.95 = 76 kg, which floors at the weight increment to a suggested 75 kg (165 lb). Increments or decrements smaller than the weight step are not handled in the suggestion.

Putting It Into Practice

Operating a deload starts with first accepting the reduction in load that the suggestion algorithm puts forward.

1. Accept the suggestion straightforwardly (the first 2-3 times): when a deload suggestion comes up, run the suggested weight for the first few times. Even if you feel "by my own sense I still have margin," there is high confidence that the fact of two consecutive failures is not noise. Accept it straightforwardly, and observe whether the weight steadily returns on the re-attempt after the deload.
2. Record the recovery pace after a deload (2-3 cycles): observe how many sessions it takes to return to the original weight after a deload. If you can return in 1-2 cycles, the deload depth is appropriate. If it takes 3 cycles or more, the deload depth may be too deep, or the program may fundamentally not match your fitness.
3. Recognize the pattern and probe the root cause: when deloads repeat over and over on the same exercise, it is a signal that reactive deloading alone will not resolve. Sleep, nutrition, overall volume, and exercise selection are the candidate factors. Verify in turn the excess weekly volume covered in the chapter on volume landmarks, the mismatched range setting covered in the chapter on the rep range, and the chronic decline in subjective margin covered in the chapter on autoregulation.

A deload is a cost of long-term progress. By asking, each time you receive a suggestion, which of fatigue, volume, sleep, or exercise selection lies behind it, you raise the design precision of the next cycle.

Frequently Asked Questions

What is a deload?

It is an adjustment method that temporarily lowers the load to clear accumulated fatigue and re-drive progress. There are two types — an intensity deload (lowering the weight while holding rep count) and a volume deload (lowering set count or frequency while holding weight and reps) — and the one DELT proposes is the former, the intensity deload.

What triggers a deload?

There are two systems. As a short-term signal, consecutive failures (dropping below the bottom of the rep range on consecutive sessions, judged over 2 sessions) and, as a medium-term signal, stagnation detection (the number of sessions since the last improvement exceeding that lifter's own typical interval between improvements). The former is read as "the recent intensity setting is too high," the latter as "the program is not producing the stimulus it needs, or insufficient recovery has become chronic."

What percentage should a deload be?

DELT designs the deload depth within the 5-10% range. For methods that run the rep range as a "span," such as Double Progression, -5% is the starting point; for linear progression with a fixed rep target (such as 5x5), -10% is the starting point. Deeper than this (-15% or more) is treated as an exceptional measure, for chronic overtraining or for returning from injury.

Should I deload just because I missed my bench press target once?

No. There are two reasons. First, a single failure can occur from day-to-day condition swings (±1-2 reps), so reacting to it would overreact to noise. Second, excessive deloading impedes progress: if you lower the weight after one failure, the next time becomes a "success" and the cycle of restoring the weight occurs more often than progress, leaving net progress at zero or negative.

Why is the estimated 1RM (e1RM) not used to judge stagnation?

Because double progression raises the weight and resets reps to the bottom of the range once the top of the range is reached, the e1RM necessarily drops once immediately after a weight increase. The several sessions it takes to climb back above the previous value are proceeding exactly as planned, yet an e1RM criterion would count them as "no update" and misjudge normal progression as stagnation. DELT instead counts improvement as "a new all-time maximum weight" or "exceeding the previous rep count at a weight handled before." The e1RM itself is still used for personal-best displays in the Stats screen.

What should I suspect when deloads keep repeating?

When deloads repeat over and over on the same exercise, it is a signal that reactive deloading alone will not resolve. When the cause of stalled progress lies in undernutrition, sleep deprivation, chronic stress, or injury, a deload does not resolve it. Sleep, nutrition, overall volume, and exercise selection are the candidate factors, and you verify them in turn — redesigning the rep range, adjusting volume landmarks, changing exercise selection.

References

• Bell, L., Ruddock, A., Maden-Wilkinson, T., & Rogerson, D. (2020). Overreaching and overtraining in strength sports and resistance training: A scoping review. Journal of Sports Sciences, 38(16), 1897-1912. https://doi.org/10.1080/02640414.2020.1763077
• Israetel, M., Hoffmann, J., Smith, M., & Feather, J. (2021). Scientific Principles of Hypertrophy Training. Renaissance Periodization.
• Helms, E. R., Cronin, J., Storey, A., & Zourdos, M. C. (2016). Application of the repetitions in reserve-based rating of perceived exertion scale for resistance training. Strength and Conditioning Journal, 38(4), 42-49. https://doi.org/10.1519/SSC.0000000000000218
• Kreher, J. B., & Schwartz, J. B. (2012). Overtraining syndrome: A practical guide. Sports Health, 4(2), 128-138. https://doi.org/10.1177/1941738111434406
• Pritchard, H., Keogh, J., Barnes, M., & McGuigan, M. (2015). Effects and mechanisms of tapering in maximizing muscular strength. Strength and Conditioning Journal, 37(2), 72-83. https://doi.org/10.1519/SSC.0000000000000125