YR Sports Lab
Your watch hands you a green "readiness 82" in the morning, and by the second round your grips are gone. That is not necessarily a broken watch, it is that you do not know how the number was built. There is a more basic problem too: you cannot wear a watch while rolling. Anything hard on the wrist can cut a training partner or catch in a gi, and most gyms and rulesets do not allow it, so the watch on your wrist has never actually been on the mat. It records the version of you before and after training, and while you sleep, and an algorithm infers the rest. That sounds like a fatal flaw, but after tracing every metric back to its original literature the conclusion is more optimistic than expected: what you can still collect off the mat is precisely the kind of measurement the research uses to make decisions, and the gap on the mat has a tool built for it. This article works through sRPE, HRV, the acute:chronic workload ratio, fitness and fatigue curves, sleep and the accuracy of the watch itself, and says which deserve a place in your judgement and which are only background.
1. Separate two questions: does the metric have evidence, and does it have BJJ evidence?
The easiest confusion in this field is treating "this metric is scientifically sound" and "this metric has been validated in BJJ" as one question. The first is about the model, the second is about extrapolation. A metric can have a dozen randomised controlled trials in runners and never once have been studied in a jiu-jitsu gym, so the two belong in separate columns.
The grading used here is a reading aid, not an official classification from any of the papers: Grade A means a meta-analysis or several randomised trials pointing the same way; Grade B means observational or validity studies exist but predictive power and causation remain disputed; Grade C means a proprietary algorithm or coaching convention with no peer-reviewed validation available. "Direct BJJ evidence" is its own column and only asks whether anyone has studied it in BJJ or combat-sport populations.
| Metric | Underlying model | Overall evidence | Direct BJJ evidence | Collectable without a watch? | Easiest mistake |
|---|---|---|---|---|---|
| sRPE internal load | Foster's RPE × minutes | A | Yes (combat-sport review) | Yes, logged after class | Perceived effort tracks lactate loosely in grappling |
| HRV trend | 7-day average of nightly RMSSD vs personal baseline | A | Descriptive studies only, no trials | Yes, measured during sleep | Insensitive to chronic fatigue; normal is not proof |
| ACWR | Gabbett's 7-day ÷ 28-day | B | None | Partly, BJJ sessions need an sRPE entry | Used as an on/off switch for today's session |
| Fitness / fatigue / form | Banister impulse-response model | B | None | Partly, BJJ sessions need an sRPE entry | No predictive power without regular performance tests |
| Sleep duration | Expert consensus on under 7 hours | A | None | Yes, measured during sleep | "Sleep debt" is bookkeeping, not a physiological quantity |
| Readiness, Body Battery | Proprietary algorithms | C | None | Yes, but BJJ is missing from the input | Formula undisclosed, cannot be independently verified |
| Pace at fixed heart rate, aerobic decoupling | Cardiovascular drift | C | Not applicable | No, needs steady-state exercise heart rate | Sparring has no steady-state pace to speak of |
The grading is a framework used in this article for readability, not an official classification from the literature. The basis and limits of each metric follow below.
2. Why off-mat data is still worth using: the comparison is "nothing at all", not a perfect experiment
It is easy to reach a pessimistic conclusion here: you cannot wear a watch, the watch has error, so what is the data even for? The flaw in that conclusion is the wrong comparison. The choice a practitioner actually faces is never "watch data" against "laboratory-grade complete measurement", it is "off-mat data plus sRPE" against "recording nothing and going on feel". By that standard the data you can collect is worth a good deal, for three reasons.
Reason 1: what you can collect is what the research actually uses
Randomised trials of HRV-guided training base their decisions almost entirely on measurements taken at rest, not on heart rate during exercise. Kiviniemi and colleagues and Vesterinen and colleagues both measured HRV every morning; Carrasco-Poyatos and colleagues had professional runners record 60 seconds of RMSSD each morning; Nuuttila and colleagues used a wrist-worn device to capture sleep and nightly recovery. In other words, the segment that "no watch while rolling" removes is not the segment these studies depend on.
Reason 2: the part you cannot collect is where the watch is least accurate anyway
Gielen and colleagues tested 10 wearables in a climate chamber in 2026 and found that what degraded accuracy across several devices was intermittent activity. Sparring is intermittent activity in its most extreme form: explosive bursts, isometric stalemates, scrambles, and a wrist that is repeatedly gripped and loaded. Even if the rules allowed a watch on the mat, the optical heart rate from those minutes would likely be the least reliable data of your day. Losing it costs less than it appears.
Reason 3: the gap on the mat has a tool designed for it
Foster and colleagues introduced sRPE in 2001 precisely to quantify the non-steady-state work that heart-rate methods handle badly. Bok and colleagues found in 2022, in eight elite karate kata athletes, that Edwards' heart-rate load and Banister's training impulse could not distinguish between three different training sessions, while sRPE could. For BJJ, sRPE is not the fallback you accept when there is no watch; it is the methodologically better fit.
So the position taken here is that off-mat data is worth using, but the degree has to be graded metric by metric. Resting HRV trends, sleep duration, resting heart rate and sRPE can be brought into your judgement with reasonable confidence; ACWR and proprietary composite scores, whether for want of evidence or want of verifiability, belong in the background. The sections below set out the basis for each.
3. sRPE: the best-supported metric here, and the best fit for BJJ
The method Foster and colleagues published in 2001 is almost crude in its simplicity: after a session, rate the whole thing for effort on a 0 to 10 scale and multiply by the number of minutes. They validated it against steady-state and interval cycling plus basketball practice, and found the score tracked heart-rate-derived load closely, though the absolute value ran higher.
Haddad and colleagues took stock of the method in 2017: Foster's paper had by then been cited by 950 studies, 36 of which examined its validity and reliability across ages, sexes and competitive levels in a range of sports. The review also notes that RPE is influenced by sleep, psychological state and muscle soreness, which is exactly what makes it a measure of internal load rather than a flaw.
Combat-sport data, and one detail that works against BJJ
When Slimani and colleagues reviewed RPE in combat sports in 2017, they found sRPE correlated moderately to strongly with heart-rate methods: 0.58 to 0.95 against Edwards' training load and 0.52 to 0.86 against Banister's training impulse. One finding matters especially for BJJ: the correlation between RPE and blood lactate was r = 0.81 in official striking matches but only r = 0.53 in grappling ones.
Put differently, in a sport built on isometric effort and grip endurance, perceived exertion maps onto metabolic markers more loosely to begin with. That is not an argument against sRPE in BJJ. It is an argument that sRPE and heart-rate methods are not measuring quite the same thing, which is all the more reason not to expect heart rate to replace it.
Judo data supports sRPE as a daily monitoring tool. Ouergui and colleagues followed 61 judo athletes aged 14 to 17 in 2020 through four weeks of intensified training and 12 days of tapering, and found sRPE during the intensified phase correlated positively with sleep, fatigue, delayed-onset muscle soreness and the Hooper index, and negatively with a recovery scale.
When you log it changes the number
Tibana and colleagues had 13 high-intensity functional training practitioners rate RPE at 0, 10, 20 and 30 minutes after exercise in 2018. Load calculated from all four time points correlated strongly with the heart-rate method, but the rating given at 30 minutes was significantly lower than the earlier three.
The practical reading is blunt: fill it in the changing room, not the next morning. A late entry is not merely vague, it is systematically low, which drags your whole weekly total down with it.
4. HRV: strong evidence in runners, no intervention study in BJJ
The basic approach to prescribing training from heart rate variability comes from Plews, Laursen and colleagues: ignore single days, watch the 7-day rolling average relative to a personal baseline. Their 2013 review is careful to note that most of this work was done in recreational and well-trained athletes, and that the few studies in elite athletes gave contradictory results, with both rises and falls in HRV linked to negative adaptation, and cases where fitness improved alongside an unusual drop in HRV. Absolute values mean little; an individual long-term baseline is what counts.
What the meta-analyses say
Bellenger and colleagues pooled 24 studies in 2016 and found two layers. Where training produced improved performance, resting RMSSD rose slightly (standardised mean difference 0.58), as did post-exercise HRV and heart rate recovery. But where overreaching produced worse performance, resting HRV was largely unaffected, while post-exercise HRV and heart rate recovery rose as well.
This point is worth keeping on its own
Two opposite states, "adapting well" and "training too much", can move post-exercise HRV and heart rate recovery in the same direction. The authors therefore state that these measures alone cannot tell you which is happening, and that additional measures of training tolerance are needed.
Translated to the mat: a healthy HRV number is not proof that fatigue is not accumulating. It has to be read next to perceived fatigue, sleep and sRPE.
The intervention trials are more cautious than the reputation suggests
Letting HRV decide whether today is hard or easy does have a run of randomised trials behind it. The most cited is Vesterinen and colleagues in 2016: 40 recreational runners split into an HRV-guided group and a predefined-programme group over 8 weeks. The HRV group did significantly fewer moderate and high-intensity sessions (13.2 against 17.7, about a quarter fewer), and improved 3000 m performance by 2.1 percent, which was significant, against 1.1 percent in the traditional group, which was not. In fairness, though, the between-group difference was only a small effect, not a decisive win.
The earlier study by Kiviniemi and colleagues in 2007 (26 participants, 4 weeks) found the HRV group improved maximal running velocity significantly more than the traditional group, while the difference in peak oxygen uptake between groups was not significant.
The two meta-analyses put the overall picture as "small but real". Granero-Gallegos and colleagues calculated an effect size of 0.187 for HRV-guided training on maximal oxygen uptake in 2020, significant but small, with amateur and female subgroups benefiting more. Manresa-Rocamora and colleagues were more conservative in 2021: HRV guidance was significantly better for vagal-related HRV indices (standardised mean difference 0.50), but not significantly better for maximal aerobic capacity (0.20) or endurance performance (0.20).
BJJ does have HRV research, but not this kind
HRV studies in BJJ populations exist. They are a different type: they describe what a match or a session does to the autonomic nervous system, rather than testing whether daily HRV should shape your week. No intervention trial of the latter kind can be found in BJJ.
- Henríquez and colleagues compared heart rate recovery and HRV in the first minute after maximal exercise in 18 BJJ practitioners in 2013 (10 highly trained, 8 moderately trained), and found the better-trained group recovered faster, supporting post-exercise HRV as an index of autonomic control.
- Andreato and colleagues put 10 BJJ athletes through four simulated 10-minute matches with repeated blood sampling in 2015. Lactate, adrenaline, noradrenaline and insulin all fell across the later matches, while creatine kinase, aspartate aminotransferase, alanine aminotransferase and creatinine kept climbing, pointing to accumulating cellular damage, with mean RR interval shortening throughout.
- Souza and colleagues compared pre-competition and pre-training states in 54 athletes including 18 BJJ competitors in 2019, and found HRV, salivary cortisol and somatic anxiety were all significantly higher before competition, with no main effect of sport type.
5. ACWR: most often used as a switch, least able to bear the weight
Gabbett proposed the "training-injury prevention paradox" in 2016: athletes accustomed to high loads get injured less, and what is dangerous is a sudden increase. Operationally that became load over the last 7 days divided by load over the last 28, with roughly 0.8 to 1.3 treated as the safer zone and risk rising above 1.5. The framework spread widely, and most sports watches now ship with some version of it.
Subsequent systematic reviews have steadily marked it down. Maupin and colleagues reviewed 27 studies in 2020 and found the load variables, ratio thresholds and reference groups differed enormously, with methodological quality scores between 48.2 and 64.3 percent. Andrade and colleagues reviewed 20 studies covering 1234 professional team-sport athletes and 2375 time-loss injuries in 2020, and noted that almost no two studies binned the ratio the same way; most suggested higher ACWR carried greater risk, but methodological heterogeneity limited the strength of any recommendation.
The most direct verdict is the 2026 multilevel meta-analysis by Ding and colleagues. Across 16 studies and 797 athletes, elevated ACWR showed a small-to-moderate association with injury risk (Hedges' g 0.35, 95% CI 0.16 to 0.54), with heterogeneity at I² = 95.8 percent and no significant effect for time-loss injuries. The conclusion is unambiguous: current evidence does not support using ACWR as a stand-alone causal or predictive model, and it should be read as one contextual indicator inside an individualised, multi-marker system.
BJJ adds a problem: the denominator is not yours to control
Every ACWR study was done in team sports, where load is set by a coach, quantified and controlled. In BJJ, sparring intensity is substantially decided by your partner: swap in a training partner one weight class up who happens to be peaking for a competition, and the same six minutes is a different physiological event.
So load in BJJ is harder to quantify than in team sports to begin with. Taking a metric that meta-analysis has already downgraded in team sports and extrapolating it to BJJ multiplies the uncertainty, not the precision.
6. Fitness and fatigue curves: the Banister model and what it assumes
The "fitness", "fatigue" and "form" lines on your watch descend from the impulse-response model Banister proposed in 1975 (published in the Australian Journal of Sports Medicine, which is not indexed in PubMed; the references here are to the later validation work). The idea is that each session produces a positive effect that decays slowly (fitness) and a negative one that decays quickly (fatigue), with performance as the difference. Busso made it non-linear in 2003, letting the weight of fatigue vary with recent training volume.
The model itself fits well in several sports. Busso and Chalencon compared six impulse-response models in 2023 using data from 11 swimmers across two competitive seasons, 61 weeks in total, with performance measured by 50 m trials twice a week. Banister's original and Busso's variable-dose version carried the greatest model weights and could be used to derive how performance evolves after a session and what an optimal taper looks like. Sanchez and colleagues modelled 5 elite female gymnasts in 2013 and reached a correlation of R² = 0.81 between modelled and actual performance; simulation suggested a taper preceded by an overload period produced slightly higher competition-day performance (106.3 percent) than one without (105.1 percent).
BJJ adds a further difficulty: there is no ready-made performance test. Running has pace, lifting has a 1RM, and "getting better at jiu-jitsu" resists a number. The pilot study by Rufino and colleagues in 2024 (7 male athletes) points one way forward: six sets of an all-out guard-passing test produced heart rate, blood lactate and RPE responses that did not differ significantly from simulated combat, and the authors suggest it works as a BJJ-specific protocol, while stating explicitly that its reliability and validity as a performance assessment still need further study.
7. Sleep: the consensus is clear, but "sleep debt" is bookkeeping
Sleep is the one metric here with an expert consensus on how much you need. Walsh and colleagues noted in their 2021 consensus statement that elite athletes commonly experience habitual short sleep (under 7 hours a night) and poor sleep quality; that a night or more without sleep clearly harms performance, but the effect of partial sleep restriction over 1 to 3 nights, the far more realistic scenario, remains unclear. The statement also cites general-population data linking habitual sleep under 7 hours to greater susceptibility to respiratory infection. Notably, the consensus argues against one-size-fits-all advice such as "7 to 9 hours" and recommends an individualised approach based on perceived sleep need.
The crossover study by Roberts and colleagues offers a useful angle. Nine athletes completed endurance time trials on four consecutive days under three conditions, with time in bed for three nights either reduced by 30 percent, left normal, or extended by 30 percent. Time trials were faster after sleep extension, and the ratio of perceived exertion to heart rate fell; after restriction, the same ratio rose. The earliest signal of insufficient sleep, in other words, is not resting heart rate or HRV itself but feeling harder at the same heart rate.
The effect sizes of sleep deprivation on performance, and the link between sleep and injury risk, are covered separately on this site: sleep and BJJ performance.
8. How accurate is the watch itself: the first question, and the most often skipped
Everything above rests on a hidden assumption: that the number the watch records is correct. It deserves its own check, because error at the input is inherited by every derived metric downstream.
Everyday heart rate: trustworthy
Tan and colleagues had 10 healthy adults wear a Holter electrocardiograph and a Garmin vivosmart 5 simultaneously for 72 to 96 hours under entirely free-living conditions in 2026, comparing 2-minute averages. The overall intraclass correlation averaged 0.902 (range 0.819 to 0.937) with a mean bias of 1.16 bpm, though limits of agreement were ±12.4 bpm and regression indicated the watch tended to underestimate heart rate. The conclusion was that the device maintained reasonable accuracy amid fluctuating heart rates in daily life.
Gielen and colleagues tested 10 consumer devices in a climate chamber in 2026 (including the Garmin vivosmart 5 and vivoactive 5), with 45 participants completing rest, a cognitive stress task, steady walking and intermittent walking at neutral (23°C), hot (36°C) and cold (10°C) temperatures. Two findings stand out, one of them counterintuitive: climate conditions did not significantly affect accuracy, while intermittent activity was what degraded several devices. The spread between devices was wide, with the Fitbit Charge 6 and Google Pixel Watch 2 performing best and the weakest group reaching mean absolute errors of 9 to 14 bpm.
Nightly HRV: a 2025 validation shows how much the model matters
Nightly HRV is a primary input to readiness and Body Battery scores, and the usual claim is that the body is still during sleep, so an optical estimate of HRV should be reliable. That reasoning is plausible, but it is reasoning, not validation, and direct validation data now exists.
Dial and colleagues had 13 healthy adults (6 women) wear an electrocardiograph alongside five consumer devices during sleep in 2025, accumulating 536 nights of paired data. Accuracy differed significantly between devices: Oura was most accurate for nightly HRV, WHOOP moderate, while the Garmin Fenix 6 and Polar Grit X Pro showed lower agreement.
| Device | Nightly HRV concordance correlation | Mean absolute percentage error | Authors' assessment |
|---|---|---|---|
| Oura Gen 4 | 0.99 | 5.96 ± 5.12% | Highest accuracy |
| Oura Gen 3 | 0.97 | 7.15 ± 5.48% | Highest accuracy |
| WHOOP 4.0 | 0.94 | 8.17 ± 10.49% | Moderate agreement |
| Garmin Fenix 6 | 0.87 | 10.52 ± 8.63% | Poor agreement |
| Polar Grit X Pro | 0.82 | 16.32 ± 24.39% | Poor agreement |
Source: Dial and colleagues, Physiological Reports 2025, 13 healthy adults, 536 nights, against electrocardiography. Garmin was excluded from the resting heart rate analysis for methodological inconsistencies, so only HRV results are listed. Model and firmware affect these figures, and the table should not be extrapolated to all devices.
How to use that table
Not as "Garmin is unusable". A mean absolute percentage error around 10 percent is still workable for watching a long-term trend, since you are comparing yourself with yourself and reading a 7-day average rather than a single day.
What it does rule out is treating one night's HRV value as a precise physiological reading. When the measurement itself may carry around ten percent of error, changing today's session over a small daily wobble is making decisions about noise.
Readiness, Body Battery and stress scores
What these scores share is an undisclosed algorithm, which makes them impossible to verify independently and leaves no peer-reviewed study to tell you their margin of error. By nature they are weighted summaries of heart rate, HRV, sleep and activity. This article grades them C not because they must be wrong, but because they cannot be examined.
Nuuttila and colleagues are worth noting here: across a 3-week baseline, a 2-week overload block and a 1-week recovery period in 24 participants, perceived strain and muscle soreness rose significantly during overload, yet sleep quality and nightly recovery metrics showed no consistent change at group level. Between-individual differences were substantial, however, and each person's change in sleep heart rate and HRV related to their change in 3000 m performance. The same principle again: watch your own trend, not a group threshold.
A note on why running metrics do not transfer
Pace at a fixed heart rate and aerobic decoupling (the drop in the pace-to-heart-rate ratio over the second half) rest on cardiovascular drift, the slow climb in heart rate during prolonged steady work as body temperature rises and plasma volume falls. The familiar "decoupling above 5 percent means an inadequate aerobic base" is a coaching threshold, and this search found no study validating that number. More fundamentally, both metrics assume there is a steady-state pace to speak of, and in sparring, highly intermittent and jointly determined by your partner, that assumption does not hold.
9. What the injury statistics say that load numbers cannot
The ultimate justification for load monitoring is usually injury prevention, which means asking where BJJ injuries actually come from.
Stegerhoek and colleagues surveyed 881 BJJ practitioners worldwide in 2025 (90 percent male, mean age 30.8), covering 888 injuries over 12 months. Self-reported injury incidence was 5.5 per 1000 hours of training (95% CI 4.9 to 6.1) and 55.9 per 1000 matches (38.8 to 73.0), with higher rates at higher belt levels. By distribution, 89 percent of injuries happened in training, 79 percent of those during sparring, and the knee (25 percent) and shoulder (13 percent) were the most affected regions.
Piekarski and colleagues quantified the risk of a single technique in 2026, comparing brown and black belt competition data from an IBJJF year where heel hooks were permitted (2021) with one where they were banned (2009). In 2021, competitors exposed to heel hooks sustained knee injuries at 26.5 per 1000 matches against 2.2 for the unexposed (relative risk 12.0). To be precise: ankle injuries did not differ significantly between exposed and unexposed (19.8 against 8.8), and the overall knee and ankle rates did not differ significantly between 2009 and 2021 either.
What those numbers mean for monitoring
The overwhelming majority of BJJ injuries happen during sparring in training, not at the point of physical exhaustion. Competition carries a per-match rate an order of magnitude higher, but because exposure is rare, cumulative risk still sits in training.
So the rule "if today feels rough, swap sparring for technique" may prevent more than any workload ratio. It changes the exposure itself rather than merely describing it.
10. Reading the metrics properly: five common misreadings
| Common reading | The problem | A more defensible reading |
|---|---|---|
| HRV dropped today, so rest | A single day carries measurement error and nightly variation | Check whether the 7-day average left your baseline range |
| HRV is normal, so I am fine | Resting HRV is insensitive to overreaching | Read it with perceived fatigue, sleep and sRPE |
| ACWR is above 1.5, so no training | Meta-analysis does not support stand-alone predictive use | Treat it as a hint that you ramped up quickly |
| Form score turned positive, time to compete | The model was never calibrated to your performance tests | Read it only as a description of training volume |
| Body Battery is low, recovery is incomplete | Undisclosed algorithm, cannot be verified | Take it as a summary signal, not a physiological value |
One principle runs through all five: these metrics are good at raising questions and poor at issuing verdicts. An HRV average that has slipped below baseline, or an unusually high ACWR, is a prompt to ask how you have been sleeping, how hard the sparring has been and whether anything hurts, not a reason to rewrite the session on the spot.
11. When to stop and see a doctor
The following are not problems that adjusting your training will fix, and warrant assessment by a sports medicine or relevant specialist:
Resting heart rate clearly above your norm for several consecutive days alongside declining performance; weeks of fatigue, insomnia or deteriorating sleep quality combined with recurrent upper respiratory infections; palpitations, chest tightness, dizziness or fainting during exercise; a joint that swells, locks, feels clearly unstable or will not bear weight; pain at an injured site that keeps worsening despite rest.
Recognising overtraining syndrome and structuring a deload are covered separately on this site: BJJ overtraining.
Read across the whole body of literature, the conclusion comes in two layers. The first is about where the evidence comes from, and BJJ monitoring is plainly in a borrowing phase. The combat-sport data behind sRPE is mostly judo and karate; every randomised trial of HRV-guided training was run in endurance athletes, most with fewer than 40 participants; ACWR comes from team sports and has already been downgraded by meta-analysis; and the convenient composite scores on your watch cannot be verified at all.
The second layer is what this means in practice, and it is considerably more optimistic. Not wearing a watch on the mat removes the segment the watch measures worst and the research does not rely on. What remains, nightly HRV, sleep and resting heart rate, is exactly the type of measurement HRV-guided training studies use, and an sRPE entry after class covers the gap on the mat. Together that is enough for a monitoring practice with a basis in the literature. Its proper use is as a long-term trend, not a morning verdict; and in a sport where most injuries occur during sparring, the question those trends are best at answering is usually whether to roll today, with whom, and for how many rounds.
FAQ
My HRV dropped a lot the morning after BJJ. Should I take the day off?
A single low day is not enough reason to stop training, because the research works from a 7-day rolling average relative to your own baseline range. The decision rules in HRV-guided training trials typically trigger when the rolling average falls below that range or keeps declining, and a dip the morning after hard sparring is an expected response. Note too that Bellenger and colleagues found resting HRV is largely unaffected by overreaching, so a normal reading does not prove you are fresh.
If I have no watch, is logging RPE alone enough?
Yes, and sRPE has the strongest evidence of any metric in this article. You rate the whole session for effort and multiply by the minutes, and in combat sports it correlates moderately to strongly with heart-rate-based load. What matters is logging every session and doing it soon after class: a validation study in high-intensity functional training found ratings given 30 minutes afterwards were significantly lower than those given immediately, at 10 minutes or at 20 minutes.
Does an ACWR above 1.5 mean I will get injured?
No. The acute:chronic workload ratio only compares recent load with the past four weeks, and a 2026 multilevel meta-analysis found a small-to-moderate association with injury risk, very high heterogeneity between studies and no significant effect for time-loss injuries. That paper states plainly that current evidence does not support using it as a stand-alone predictive model. Treat it as background context on how fast you have ramped up, especially as sparring intensity in BJJ is largely set by your partner.
You cannot wear a watch for BJJ, so is the data it collects still useful?
Yes, because HRV-guided training research makes its decisions from resting morning or overnight measurements, not from heart rate during exercise. The optical heart rate you cannot capture while rolling is exactly where wearables are least accurate, during intermittent activity, and the load on the mat can be covered by an sRPE entry after class. The value does need grading, though: nightly HRV trends, sleep and sRPE deserve a place in your judgement, while proprietary composite scores such as Body Battery are only background.
References
1. Foster C, et al. (2001). A new approach to monitoring exercise training. J Strength Cond Res;15(1):109-115. PubMed PMID: 11708692 (the original sRPE method, validated against heart-rate load in cycling and basketball)
2. Haddad M, et al. (2017). Session-RPE method for training load monitoring: validity, ecological usefulness, and influencing factors. Front Neurosci;11:612. PubMed PMID: 29163016 (source of the 950 citations and 36 validity studies cited above)
3. Slimani M, et al. (2017). Rating of perceived exertion for quantification of training and combat loads during combat sport-specific activities: a short review. J Strength Cond Res;31(10):2889-2902. PubMed PMID: 28933715 (source of r = 0.81 for striking versus r = 0.53 for grappling, and the Edwards and Banister correlation ranges)
4. Ouergui I, et al. (2020). Relationship between perceived training load, well-being indices, recovery state and physical enjoyment during judo-specific training. Int J Environ Res Public Health;17(20):7400. PubMed PMID: 33050671 (61 judo athletes aged 14 to 17, four weeks of intensified training plus 12 days of taper)
5. Tibana RA, et al. (2018). Validity of session rating perceived exertion method for quantifying internal training load during high-intensity functional training. Sports (Basel);6(3):68. PubMed PMID: 30041435 (RPE recall timing; the 30-minute rating was significantly lower)
6. Plews DJ, et al. (2013). Training adaptation and heart rate variability in elite endurance athletes: opening the door to effective monitoring. Sports Med;43(9):773-781. PubMed PMID: 23852425 (basis for individual baselines and 7-day averaging, and for the contradictory findings in elite athletes)
7. Bellenger CR, et al. (2016). Monitoring athletic training status through autonomic heart rate regulation: a systematic review and meta-analysis. Sports Med;46(10):1461-1486. PubMed PMID: 26888648 (24 studies; RMSSD standardised mean difference 0.58; resting HRV largely unaffected by overreaching)
8. Kiviniemi AM, et al. (2007). Endurance training guided individually by daily heart rate variability measurements. Eur J Appl Physiol;101(6):743-751. PubMed PMID: 17849143 (26 participants, 4 weeks, morning measurement; greater gain in maximal running velocity)
9. Vesterinen V, et al. (2016). Individual endurance training prescription with heart rate variability. Med Sci Sports Exerc;48(7):1347-1354. PubMed PMID: 26909534 (40 recreational runners, 8 weeks; 13.2 versus 17.7 hard sessions; small between-group effect)
10. Figueiredo DH, et al. (2022). Individually guided training prescription by heart rate variability and self-reported measure of stress tolerance in recreational runners. J Sports Sci;40(24):2732-2740. PubMed PMID: 36940300 (36 runners; the self-report group improved more than the HRV group)
11. Ranieri LE, et al. (2025). Performance and physiological effects of race pace-based versus heart rate variability-guided training prescription in runners. Med Sci Sports Exerc;57(7):1510-1522. PubMed PMID: 39935030 (28 participants, 6 weeks; no group interaction)
12. Granero-Gallegos A, et al. (2020). HRV-based training for improving VO2max in endurance athletes: a systematic review with meta-analysis. Int J Environ Res Public Health;17(21):7999. PubMed PMID: 33143175 (source of the 0.187 effect size for maximal oxygen uptake)
13. Manresa-Rocamora A, et al. (2021). Heart rate variability-guided training for enhancing cardiac-vagal modulation, aerobic fitness, and endurance performance: a methodological systematic review with meta-analysis. Int J Environ Res Public Health;18(19):10299. PubMed PMID: 34639599 (vagal indices significant; aerobic capacity and endurance performance not)
14. Nuuttila OP, et al. (2025). Monitoring sleep and nightly recovery with wrist-worn wearables: links to training load and performance adaptations. Sensors (Basel);25(2):533. PubMed PMID: 39860902 (24 participants; no consistent group-level change in nightly recovery during overload)
15. Gabbett TJ (2016). The training-injury prevention paradox: should athletes be training smarter and harder? Br J Sports Med;50(5):273-280. PubMed PMID: 26758673 (the original acute:chronic workload framework)
16. Maupin D, et al. (2020). The relationship between acute:chronic workload ratios and injury risk in sports: a systematic review. Open Access J Sports Med;11:51-75. PubMed PMID: 32158285 (27 studies; methodological quality 48.2 to 64.3 percent)
17. Andrade R, et al. (2020). Is the acute:chronic workload ratio associated with risk of time-loss injury in professional team sports? Sports Med;50(9):1613-1635. PubMed PMID: 32572824 (20 studies, 1234 athletes, 2375 time-loss injuries)
18. Ding L, et al. (2026). Acute:chronic workload ratio and load management for team sports: a multilevel meta-analysis. Front Public Health;14:1896651. PubMed PMID: 42662491 (Hedges' g 0.35, I² 95.8 percent, no significant effect for time-loss injuries, not supported as a stand-alone model)
19. Busso T (2003). Variable dose-response relationship between exercise training and performance. Med Sci Sports Exerc;35(7):1188-1195. PubMed PMID: 12840641 (the non-linear version of the impulse-response model)
20. Busso T, Chalencon S (2023). Validity and accuracy of impulse-response models for modeling and predicting training effects on performance of swimmers. Med Sci Sports Exerc;55(7):1274-1285. PubMed PMID: 36791017 (11 swimmers, 61 weeks, 50 m trials twice weekly, six models compared)
21. Sanchez AM, et al. (2013). Modelling training response in elite female gymnasts and optimal strategies of overload training and taper. J Sports Sci;31(14):1510-1519. PubMed PMID: 23656356 (5 gymnasts, R² 0.81, 106.3 versus 105.1 percent in the taper simulation)
22. Walsh NP, et al. (2021). Sleep and the athlete: narrative review and 2021 expert consensus recommendations. Br J Sports Med;55(7):356-368. PubMed PMID: 33144349 (under 7 hours a night, unclear effect of 1 to 3 nights of partial restriction, respiratory infection risk)
23. Roberts SSH, et al. (2022). Monitoring effects of sleep extension and restriction on endurance performance using heart rate indices. J Strength Cond Res;36(12):3381-3389. PubMed PMID: 34711770 (9 athletes, time in bed varied by 30 percent, change in the RPE-to-heart-rate ratio)
24. Grandou C, et al. (2019). The effects of sleep loss on military physical performance. Sports Med;49(8):1159-1172. PubMed PMID: 31102110 (aerobic and muscular endurance most affected, lower-body maximal strength relatively resilient; military population)
25. Tan C, et al. (2026). Reliability and validity of a smartwatch with a photoplethysmograph: comparison with a Holter electrocardiograph. Chronobiol Int;43(9):1350-1358. PubMed PMID: 42099266 (intraclass correlation 0.902, bias 1.16 bpm, limits of agreement ±12.4 bpm)
26. Gielen J, et al. (2026). Accuracy of optical heart rate measurements for 10 commercial wearables in different climate conditions and activities. JMIR Form Res;10:e85186. PubMed PMID: 41701929 (45 participants; climate did not significantly affect accuracy, intermittent activity did)
27. Dial MB, et al. (2025). Validation of nocturnal resting heart rate and heart rate variability in consumer wearables. Physiol Rep;13(16):e70527. PubMed PMID: 40834291 (13 adults, 536 nights, five devices against electrocardiography; source of the device comparison table)
28. Andreato LV, et al. (2017). Physical and physiological profiles of Brazilian jiu-jitsu athletes: a systematic review. Sports Med Open;3(1):9. PubMed PMID: 28194734 (58 studies, 1496 athletes; aerobic fitness does not discriminate competitive level)
29. Andreato LV, et al. (2015). Brazilian jiu-jitsu simulated competition part I: metabolic, hormonal, cellular damage, and heart rate responses. J Strength Cond Res;29(9):2538-2549. PubMed PMID: 26308831 (10 athletes, four 10-minute matches; lactate and catecholamines fell while damage markers rose)
30. Henríquez OC, et al. (2013). Autonomic control of heart rate after exercise in trained wrestlers. Biol Sport;30(2):111-115. PubMed PMID: 24744476 (18 BJJ practitioners; the journal title says wrestlers, the population is Brazilian jiu-jitsu)
31. Souza RA, et al. (2019). Heart rate variability, salivary cortisol and competitive state anxiety responses during pre-competition and pre-training moments. Biol Sport;36(1):39-46. PubMed PMID: 30899138 (54 athletes including 18 BJJ competitors; all three markers higher before competition)
32. Rufino HVO, et al. (2024). Physiological and perceptual responses of a guard passing test and a simulated Brazilian jiu-jitsu combat: a pilot study. J Strength Cond Res;38(10):e574-e578. PubMed PMID: 38900221 (7 athletes; no significant difference in heart rate, lactate or RPE against simulated combat)
33. Stegerhoek PM, et al. (2025). Injury prevalence among Brazilian Jiu-Jitsu practitioners globally: a cross-sectional study in 881 participants. BMJ Open Sport Exerc Med;11(1):e002322. PubMed PMID: 40092168 (5.5 per 1000 training hours, 55.9 per 1000 matches, 89 percent in training and 79 percent during sparring)
34. Piekarski M, et al. (2026). Knee injury in competitive Brazilian jiu jitsu athletes: implications for training. Sports Health;18(4):890-897. PubMed PMID: 41549501 (26.5 versus 2.2 knee injuries per 1000 matches, relative risk 12.0; ankle difference not significant)
35. Schmitt L, et al. (2018). Live high-train low guided by daily heart rate variability in elite Nordic-skiers. Eur J Appl Physiol;118(2):419-428. PubMed PMID: 29247273 (24 elite Nordic skiers; one of the sources for the sample-size chart)
36. Carrasco-Poyatos M, et al. (2022). Heart rate variability-guided training in professional runners: effects on performance and vagal modulation. Physiol Behav;244:113654. PubMed PMID: 34813821 (12 professional runners, 60 seconds of RMSSD each morning; one of the sources for the sample-size chart)
37. Bok D, et al. (2022). Validation of session ratings of perceived exertion for quantifying training load in karate kata sessions. Biol Sport;39(4):849-855. PubMed PMID: 36247939 (8 karate kata athletes; heart-rate methods could not distinguish sessions while sRPE could)
Banister's original 1975 impulse-response paper appeared in the Australian Journal of Sports Medicine, which is not indexed in PubMed, so the validation and refinement studies are cited instead (references 19 to 21). All PMIDs were checked against NCBI E-utilities for bibliographic details and abstracts on 16 September 2026.