Training Stress Balance is useful because it compresses recent cycling load and longer-term load into one visible line. It is dangerous when that convenience gets mistaken for biological certainty.
The practical answer is simple: use TSB to check the direction of a planned build, recovery period or taper; do not use it to declare that a cyclist is ready, overtrained or guaranteed to perform.
If you need the definitions and maths for the whole Performance Management Chart, start with CTL, ATL and TSB explained for cyclists.
TSB in 30 seconds
| Question | Evidence-bounded answer |
|---|---|
| What is TSB? | A modelled difference between chronic and acute training load. |
| TrainingPeaks formula | Today's TSB = yesterday's CTL − yesterday's ATL. |
| Does positive mean race-ready? | No. It means recent modelled load is lower relative to chronic load. |
| Does negative mean overtrained? | No. It means acute modelled load exceeds chronic modelled load. |
| Best race-day number? | No universal value has been validated. |
| Useful job | Confirm that load is moving in the intended direction and compare the model with the rider's response. |
TrainingPeaks describes TSB as “Form” and calculates it from the previous day's Fitness (CTL) and Fatigue (ATL). Its own documentation also states that TSB is not a predictor of performance (TrainingPeaks Form/TSB documentation).
That product boundary should govern how Roadman talks about the number.
What positive, neutral and negative TSB really mean
Negative TSB: the modelled acute load is greater than the modelled chronic load. That is common during a training block. The size of the negative number does not diagnose fatigue, injury risk, illness or overtraining.
Near-zero TSB: the two modelled loads are similar. It does not mean “nothing is happening,” and it does not identify a physiological plateau.
Positive TSB: the modelled acute load has fallen below the chronic load. That often happens during reduced training. It does not prove freshness, preserve event-specific sharpness or show that the athlete has slept and fuelled well.
The chart knows the load values it receives. It does not know why the athlete feels different.
Why there is no magic race-day TSB
The old version of this guide prescribed +5 to +25 for most riders, said values above +30 meant lost sharpness and assigned masters cyclists to the upper end. Those claims were not tied to a validation study.
Three problems make a universal range unreliable:
- The input can differ. FTP setting, estimated versus power-based TSS, unrecorded sessions, strength training and platform calculations change the load history.
- The same number can come from different training. Event-specific intervals, endurance volume and unrelated stress may produce similar chart values but different preparation.
- Performance has more inputs. Sleep, fuelling, illness, travel, heat, motivation, pacing and competition context sit outside TSB.
Historical TSB from a rider's strong and weak events can still help. Compare like with like: similar event duration, training phase, data quality and taper. Treat the result as a clue. Three good races at the same TSB do not prove that the fourth requires it.
What taper research supports
A 2023 systematic review and meta-analysis included 14 endurance-athlete studies. The pooled analysis found improvements in time-trial and time-to-exhaustion performance after tapering, while VO2max and movement economy did not significantly improve (PMID 37163550).
Subgroup results supported tapers lasting up to 21 days that reduced volume by 41–60% while maintaining training intensity and frequency. An earlier 27-study athlete meta-analysis also favoured an approximately two-week taper with a 41–60% volume reduction and maintained intensity and frequency (PMID 17762369).
Those results provide a starting framework, not a compulsory cyclist prescription:
- the included athletes, sports and taper designs varied;
- a group mean does not choose one rider's volume, interval count or duration;
- retaining “intensity” does not mean reproducing the full pre-taper workload; and
- neither meta-analysis validated a destination on a TSB chart.
Plan the taper from the event demands and the rider's history. Use the chart to see whether recorded load fell as intended.
A practical TSB taper workflow
1. Verify the source data
Check FTP changes, missing rides, duplicate activities, estimated load, indoor/outdoor device differences and whether strength work appears. A precise-looking TSB built from inconsistent inputs is still inconsistent.
2. Define the event priority
A one-day hill climb, long sportive, criterium and multi-day event do not ask the same thing. Identify the work that should remain familiar and the fatigue you are trying to reduce.
3. Write the taper before reacting to the chart
Choose the volume reduction, retained event-relevant work and recovery days. TSB should report the effect of the plan, not provoke a new plan every morning.
4. Watch direction and pace, not a destination alone
Is ATL falling? Is CTL behaving roughly as expected? Is TSB moving positive as reduced load enters the model? A surprising curve should trigger a data and plan review, not a panic rest day.
5. Cross-check the athlete
Record sleep, energy, soreness, mood, symptoms and actual session response. Athlete-monitoring research supports structured subjective reports as useful trend information, while also showing that common measures have validation limits (PMID 26423706; PMID 32957081).
6. Review after the event
Save the model values, taper, sleep/fuelling context, warm-up response and performance. That makes the next comparison more specific than copying a generic internet range.
Common TSB mistakes
Chasing CTL before chasing form
CTL summarises modelled load. It does not prove fitness gained, session specificity or successful adaptation. Adding low-value work to protect a chart line can damage the taper's actual purpose.
Treating TSB as today's readiness score
TSB is based on recorded training load; a daily cycling readiness check organises sleep, soreness, energy, mood, stress and recent context. Neither is a medical diagnosis or standalone command. They answer different questions.
Using one negative threshold to diagnose overreaching
No −20, −30 or other TSB cut-off diagnoses functional overreaching, non-functional overreaching or overtraining syndrome. The ECSS/ACSM consensus describes overtraining syndrome as a difficult exclusion diagnosis without one generally accepted marker (PMID 23247672).
Assuming masters riders need one correction
The taper reviews did not establish an over-40 TSB range or a universal longer fatigue constant. Training age, event, health, recent load and individual response matter. Use history; do not add points because of a birthday.
Ignoring low energy availability or illness
A positive TSB cannot rule out illness, inadequate fuelling, iron deficiency, sleep problems or low energy availability. Persistent fatigue, performance decline or concerning symptoms need the appropriate investigation rather than a more favourable chart.
Which Roadman resource should you use?
| Need | Owner |
|---|---|
| Understand CTL, ATL, TSB and the PMC | Training load explained |
| Calculate a modelled load sequence | Training load calculator |
| Estimate one ride's TSS | TSS calculator |
| Review today's subjective context | Daily readiness guide and free check |
| Build a taper for a target event | Cycling taper guide |
| Fit strength and recovery around an existing ride week | Upcoming Roadman strength and recovery app |
| Have bike and gym load reviewed by a person | Roadman cycling coaching |
The upcoming app will use recent bike load as context when placing strength and recovery. It will not convert one TSB value into a new cycling plan or diagnose the rider.
Sources and editorial limits
- TrainingPeaks: Form (TSB) definition and formula
- Endurance taper systematic review and meta-analysis (PMID 37163550)
- Athlete taper meta-analysis (PMID 17762369)
- Athlete subjective monitoring systematic review (PMID 26423706)
- Athlete-reported measure validation review (PMID 32957081)
- ECSS/ACSM overtraining consensus (PMID 23247672)
This page was last reviewed on 1 September 2026. TSB terminology can differ by platform and settings. Roadman does not present a chart value as a diagnosis, performance forecast or guaranteed taper target.