Everyone says a recovery score tells you whether to train hard today, and WHOOP says it too, so for a while I believed it. I train MMA every day and add strength work on the days when I'm not too fatigued.

Since I also code and travel, all of that has to fit around training. The only decision I really need help with is a small one, which is whether I have enough left for the lifting session or whether I'll regret adding it tomorrow.

WHOOP hasn't been answering that for me because it shows green on mornings when I feel massively unrecovered.

On multiple mornings leading up to a fight I'd feel fatigued and WHOOP would still put me at around 70% recovery, which didn't make sense to me. On days when I didn't feel recovered I'd see green recovery too. It isn't constant but it happens quite often.

I can ignore it of course but I bought the strap and wore it every night and let it learn my baseline, so I want that number to mean something. If it doesn't I'm back to going by how I feel, which is what I did before I bought it.

That's why I pre-ordered Temple, and yes, it's another device that I bought partly because I'm fed up with the ones I already have. I got in through the founder email and early access rather than the public pre-order. I haven't worn it yet and I have no idea whether it will fit my training any better so I've ordered it to find out.

Why a green score doesn't convince me

Is WHOOP useless? Not at all. It's the tracker I've trusted most and that's exactly why the disagreements bother me, because if it were obviously wrong I would have stopped looking at it a long time ago.

It also isn't a single bad number since WHOOP says Recovery uses several inputs such as HRV, resting heart rate, sleep and respiratory rate. I'd love to blame one of them but I can't.

HRV is the one I end up arguing with and in my experience it fluctuates a lot. It's the variation between heartbeats and it does have a place in training research, but what I want from it is much bigger.

I want an answer for how MMA and lifting and sleep and everything outside the gym have landed in one body. That's a lot to ask of a heart signal.

I should be fair to the tracker here because I don't have a lab measurement of my own recovery to put next to it. Feeling tired doesn't prove the sensor is wrong and feeling fine doesn't prove I'm recovered.

But if I keep looking at a score and wondering what to do with it then it isn't doing its job in my life, even if the underlying readings are perfectly good.

Some of it may also be placebo. I tested it on myself and on the days when I looked at WHOOP's score first thing in the morning I felt considerably worse or better afterwards even when the underlying data was no different. That was only my own informal test and not a study, so I'm not claiming more than that.

The heart rate data is a separate problem from the recovery score. When I run WHOOP's heart rate zones are totally wrong because it has me comically in zone 5 the moment I start getting a decent pace. That one has nothing to do with HRV.

Fitbit hasn't given me enough confidence to make this call either. The Apple Watch has been better, but I can't wear a watch during contact MMA training, so each one gets me part of the way and I fill in the rest myself. That's why Temple caught my eye, since it doesn't sit on the wrist at all.

What the early studies say

Temple sits at the side of the head and uses optical sensing, which means light goes into the tissue and a detector reads the light that comes back as the blood volume changes with each heartbeat. That technique is called photoplethysmography and it's well established, so moving it to the temple doesn't create new physics.

What it does change is where you measure. The pulse has a shape as well as a rate. The site and the hardware and the tissue around the sensor all affect what you pick up, and so does movement.

Temple's BrainFlow index is meant to follow relative changes in blood circulation in the brain and that's a much bigger claim than "we can read a pulse from your head". Being close to the brain isn't enough because an optical sensor on the head also picks up tissue outside the skull.

So before I get excited, I need a reason to believe the signal is what Temple says it is. Do the studies give me one? Partly, since there are three of them and all three are preprints.

In the first, researchers compared BrainFlow with transcranial Doppler ultrasound, which measures blood-flow velocity in a major artery in the brain. They recruited 23 healthy adults and analysed 20 in each protocol using cycling and a standing-to-lying change.

BrainFlow moved with the reference and the median within-person correlations were around 0.80. The part I liked was the posture change because BrainFlow followed the reference while heart rate moved the other way, which suggests it might be more than a heart-rate trace in disguise.

Do note that this is about relative changes, so it doesn't mean 80% accuracy and it doesn't measure absolute blood flow in the brain. The study also lists Temple funding. And nothing in it answers my actual question, which is whether I should add lifting to an MMA day.

The second preprint gets closer to what the sensor is reading. Across 60 sessions from 44 healthy young men, it compared BrainFlow with a time-domain near-infrared method that models brain and scalp haemoglobin separately.

The head-down tilt result looks very good, with a median lag-adjusted correlation of 0.912 against the brain signal and 0.494 against the scalp one. But the separation wasn't consistent because for standing to lying the difference wasn't statistically significant.

Temple designed, funded and ran this study, and the authors disclose company and research-support relationships. The paper itself says more work is needed to establish that the signal is specific to the brain, so I can't take the best result and ignore that part.

The third preprint looks at age-related pulse features and fourteen of 89 features had robust age associations in the temple data compared with three at the wrist. Those were different datasets, including a public wrist dataset, and not the same people measured with the same hardware at both sites. The authors disclose company employment.

So where does that leave me? I'm interested and I'll keep watching but these are preprints and not peer-reviewed proof of better recovery advice. A second useful signal is a real possibility but I can't claim a finished answer from these papers.

Where I fill in the blanks

The current Temple site leads with Entropy, which it describes as the body's effort to shed heat. It talks about a low resting Baseline, a high Threshold under activity and something it calls Momentum.

I can see why that appeals to me because I'm always trying to fit more work around the work I'm already doing. Something that shows my resting state and my capacity under effort would be useful well beyond the gym.

But that usefulness is something I'm imagining since the blood-flow studies don't establish the consumer Entropy score as an accurate measure of whole-body heat loss and they don't show it improves training decisions. I also haven't found enough public detail to work out the scoring algorithm myself.

There's a gap between finding a signal, explaining a score and knowing what to do differently because of it. My problem with WHOOP sits in that last part, so Temple could collect different information and still leave me with the same problem.

I also have a bias, which is that I want Indian founders to build things that change what a category can do. Every tracker I've worn has felt like a different version of the same thing and I'd like to see someone try something else.

A company taking a hard measurement problem seriously matters to me even if the first version is incomplete, and it helped me decide to buy.

It can also make me too forgiving because being happy that an Indian company is attempting this won't make a reading more useful on a morning when I feel wrecked. It doesn't make Temple the only company working on head-worn sensing either. I can back the attempt and still expect the device I receive to earn its place.

Update, 1 November 2026: Temple's heart-rate study

Temple has published a heart-rate white paper along with the raw data. Temple funded, ran and analysed it, and it's a company white paper rather than a preprint or a peer-reviewed study.

They had 43 healthy adults, 34 men and 9 women aged 20 to 35, walk and run on a treadmill through zones 1 to 5 in a lab in Gurugram. Everyone wore Temple, Apple Watch Series 12, WHOOP MG, Garmin Forerunner 970, Fitbit Air and Oura Ring 5 at once, checked against a Polar H10 chest strap taped to the skin.

Per Temple, its error in zones 4 and 5 was 1.16 bpm. In zone 5 alone it was 0.95 bpm, against 1.84 for the Apple Watch and 3.54 for WHOOP. The Apple Watch came closest overall, which matches my experience. They also put the same Polar sensor on the temple, chest, thumb and wrist, and the temple had the lowest error.

So the study supports two things I care about. Heart rate at high effort can be read well, and where the sensor sits matters.

WHOOP was tested, which is the part closest to my complaint about being comically in zone 5. But WHOOP's big misses in the study were mostly low readings rather than high ones, so it doesn't explain mine directly. My problem could also be how the zones are set, which this doesn't test.

What it doesn't settle is everything else. It's Temple's own study and it isn't independent yet, though Temple says an extended study by independent researchers is underway. It was treadmill running only, with one session per person. It says nothing about recovery scores or HRV, which is my main complaint. And it doesn't show whether the adhesive holds up in contact MMA, where something on my head could get knocked.

So it's a good sign for heart rate and not much more than that for now.

What I'll check when it arrives

The main question is whether I'll actually wear it and whether it changes anything. I haven't tested how the adhesive holds up with sweat, whether it irritates my skin or what happens in sessions where something on my head could get knocked. So I'll check Temple's intended use and my gym's rules before I assume it belongs in contact training.

I've also understood that Temple Care includes a replacement for a nominal fee if I lose the device but I haven't verified that against a public policy. The published warranty covers manufacturing defects for a year and that isn't loss cover, so I want the actual Care terms before I treat replacement as sorted.

I'm expecting not to have to pay the full device price again if I need a replacement, but that's my understanding of Care rather than a confirmed public promise. How much exactly is the fee? I have got no idea.

Then there's my data because Temple says it doesn't sell personal or health data for money.

It also says some AI features may share de-identified device metrics with providers under a restriction on model training and I'll read that again once I know which features I'll actually use. These details are boring compared to measuring something near the brain but they decide whether a wearable is still on my body after the novelty is gone.

When mine arrives, I'll write down how I feel and what I plan to train before I open the app because otherwise it's far too easy to see the score first and then find a reason for it. I'll keep track of the days it disagrees with me, the sessions it misses and the times I can't wear it.

That won't turn my training log into a validation study, and Temple's own terms describe a wellness product and not a medical device, so I won't use its score as permission to train through symptoms. But a few weeks of notes should tell me whether it helps with my own decisions.

For now, I want it to work and I haven't had the chance to find out. I know what frustrated me enough to order it, but I don't know whether it will make a green-score morning any less confusing.

If you've used Temple or tried a head-worn tracker, let me know through the Contact page because I would love to hear how it went for you.