Apple Watch 6 Accuracy
I’m planning a more detailed look at the sporty accuracy of the Apple Watch 6 later in the year as I put together lots more workouts stats, so please don’t take these results as a final view. That said, the Apple Watch 6 does look surprisingly accurate, to the point where it may well be later proved to be more accurate than sport-specific watches. *IF SO* that could be a game-changer.
What I have for you here are comparative GPS and oHR stats of the Apple Watch 6 vs. leading sports watches for a single swim, a single bike, a single run and, somewhat randomly for me, a walk around Oxford Town centre. Let’s have a detailed look at the charts or you can just skip ahead to the ‘take-out’ summary at the end.
Open Water Swimming
This is two laps of Shepperton Lake. I have LOTS of historical swims here and until relatively recently most watches weren’t too great. However, over the last year, Garmin seems to have nailed Open Water Swim (OWS) GPS stats at least to the point where it’s ‘accurate enough’, especially considering you can rarely look at your watch when swimming.
- Apple Watch 6 – Blue, oHR
- Garmin 745 – Red (oHR, Glonass)
- Garmin 945 – Green (HRM-TRI, Glonass)
You’ll have to trust me when I say that all these GPS tracks are very good compared to historical efforts. They all roughly agree but I would never know for certain which one is correct as I don’t always swim in a super-straight line.
The heart rate track is more interesting as the green line can be assumed to be correct as it is a chest strap. Yet the Apple Watch 6 matches it closely. It’s not perfect and seems to flat-line for a minute when I first get into the water. Other than that, I’d certainly be happy to use its HR data in my stats. Compare that to the red Garmin Forerunner 745 line which is from the just-released Garmin tri-watch that’s specifically designed to be used in open water and which uses Garmin’s latest HR sensor….not good.
Bike Ride – the Richmond Hill Octopus
This ride is similar in name to the Leith Hill Octopus albeit with 8 significantly easier ascents. Nevertheless, these much shorter suburban/rural hill-repeats give a decent elevation track and some interesting twists and sharp turns to check out the responsiveness of the GPS. The ride also lets me add in some other sections in a town which introduces some limited GPS challenges from nearby tall buildings. Earlier Apple Watches (or earlier version of WatchOS) fiddled the data with a smoothing algorithm which made it look better than it was in reality…fine for a pretty picture but not very confidence-inspiring when you are more concerned about true accuracy and real distances covered.
Also, note that I am using an ANT+ chest strap and I do not have my iPhone with me. There is no way the Apple Watch can be picking up a good signal or a good track from another device.
- Apple Watch – Yellow
- Wahoo Bolt + HRM-TRI – Blue
- Polar Vantage V Precision Prime oHR – Green
- Garmin 745 + Elevate oHR – Red
The overview shows nothing special, all devices perform similarly well. Once we start to drill down then the errors become apparent.
The triangular trip around Richmond town centre (image 2) has all devices very slightly struggling in places as the GPS signals are reflected off buildings, so changing the apparent position. The Garmin, Apple and Wahoo tracks each take it in turn at being the ‘best’ depending on which particular part of the triangle you look at. Effectively they are all ‘the same’ and all more than ‘good enough’.
Image 3 (roundabout/circle) zooms in further and shows that all devices have issues by showing me cycling through buildings and on the wrong side of the road.
Images 4 is further up the hill where there are more buildings on one side next to a right angle bend. This image gives an indication that each device is repeating the same kind of error at each pass of the corner – the passes were perhaps 5 minutes apart. Here the Wahoo and Polar perhaps have the best track.
Image 5 shows a switchback and then a climb up a hill which must be about a 20% gradient near the top – surprisingly steep for SW London. The switchback and climb with some tree cover fools all the devices…except the Apple Watch 6. Very nice job Mr Apple.
Image 6 shows a few descents (and one climb) down the Start & Garter hill. Even though the trees are fairly overhanging and dense, the tracks are good, aided by the speed of the descent which somewhat smooths out the tracks
Image 7 shows open-sky, benign conditions which give no issues to any of the devices.
Image 8 is on my way home as I replicate a similar test to the Richmond Town Centre triangle but this time there are a couple of right-angle turns in a suburban area. Here the rider’s proximity to buildings is lessened because of the gardens and you can see the this notably improves the track quality for all devices. If you were to zoom in further you would see the Apple Watch even having me on the correct side of the road. That doesn’t mean its inherently more accurate than the other…it probably just got especially lucky this time. All watches give fine tracks.
Finally, let’s look at the heart rate tracks and the elevation stats for the ride
Overall this wasn’t a super-hard effort for me, although 170bpm is probably not far from a maximal bike effort, they were short efforts with plenty of recoveries. Nevertheless, 170bpm would have been an HR level where most optical devices struggled a few years back. Most of these HR tracks we see here are probably acceptable to some degree. However, what is surprising is that the Apple Watch is mirroring the accuracy of the HRM-TRI chest strap on the 945. Again the brand new Garmin 745 is lacking at times…can it really be true that Apple has produced a better optical HRM for sports than Garmin? Please note that I am normally a BAD candidate for optical HR, it usually goes wrong with me for various physiological reasons – sweat, skin tone, whatever.
Road noise can also be a problem for optical HR, so it’s worth noting that some of these roads were a little bit bumpy and some were smooth.
Finally looking at the elevation track you can’t see the Apple Watch 5 line as it has changed colour to pink. Why can’t you see it? A: Because it is pretty much hidden by the green track of the ‘correct elevation’ from an SRTM data source. Scarily good. (Remember that the AW6 has an always-on altimeter, one advantage of this is that it will probably start all your workouts at a properly calibrated elevation and it might also do clever things with checking the true elevation of your GPS location on the internet if your phone is present)
Run Around the Block – Again & Again
I have added this kind of run into my suite of run tests as it represents the conditions of runs that very many of us living in built-up areas encounter. I don’t particularly like these kinds of runs but the GPS data is instructive especially when considered in the context of other run types (that I have planned for the future)
- Apple – Blue
- Polar Vantage V Precision Prime oHR – Yellow, Galileo
- Garmin 945 HRM-TRI, Glonass – Red
- Garmin 745 Elevate oHR – Green
Apart from one error by the Apple Watch 6, circled in white, the track it produced was the best. If I were to rank the rest of the test from the Apple Watch it would have scored VERY highly as the track was consistently within the +/-5m error bands that I would say is expected from GPS. I got quite excited about this as the track was unusually good, however, I then looked at my PACE figures and the Apple Watch was 15-20 secs/km different from STRYD at times. So, there’s something else going on here that I hope to get to the bottom of.
GPS Testing Can Drive you round the bend
On a different ride a few days ago, I replicated a test ride I remembered from my look at the AW4 (yellow, image 2). As you can see Apple has now decided to stop smoothing tracks to make them appear more realistic. However AT SLOWER SPEEDS, like when running, you can still see some of the corner-cutting.
I’m sure that Apple used to do this to hide errors in the track. Now they seem to have accepted some errors but, from what I’ve seen so far, they’ve nothing to hide.
A Walk In Oxford
Here I was really testing out on-wrist navigation from routes sync’d from Apple Maps. Whilst Apple Maps is not as good as Google Maps on the smartphone and in the car, the wrist-based implementation with haptic alerts signalling turning points worked well. Apple Maps is now available again as a WatchOS app but can’t get it to install 🙁
This was a bit of fun from a GPS test point of view. I thought I’d see how the AW6 handled a walk around Oxford compared to the Garmin 745. These tracks have moments of accuracy and the Apple is a bit better than the Garmin but both have some quite serious issues probably linked to the slow walking speed. then I realised that I had my iPhone with me and the Watch was almost certainly taking a GPS fix from that. I shall dwell on this no further, I just liked the image…and the city-walk navigation experience!
Summary | Take Out
These stats DO look very promising for Apple but I have had false dawns before.
Even if my detailed Apple Watch 6 accuracy testing throws up new errors not encountered here, I would still expect it to show a decent performance on all fronts. The realistic, best possible case for Apple is that it is better than Garmin for sports accuracy. I suspect that will be the case but I wouldn’t bet on it yet as some other tests I have performed show that there is rounding and corner-cutting in some circumstances.
Whatever pans out over the next few weeks for my tests, however, will confirm what many of you Apple Watch users already know…it’s a decent performer for all-things-sporty.