Revista Ciclismo

Openwater Swim GPS Test Extravaganza Results

Por Triatlonsindrafting @trisindrafting

A few years back I took a week while warm-water swimming to test a slate of GPS watches in openwater. Like a race, each watch got essentially one go at it, competing against another watch on the other wrist - and then a reference GPS track floating on a swim buoy behind me. That was popular, and with a new slate of GPS devices over the past year or two, I figured it was time to do it again. This time expanding the test group to a dozen GPS watches, over three weeks of openwater swims.

Here was the lineup this year, focusing on the most popular watches for triathletes and openwater swimmers:

Certainly, the case could be made for trying to squeeze in a few others (e.g. Garmin FR745, COROS Pace 2, or Polar Vantage M2), but ultimately, my little watch travel case only had so much space, and I only had so many days to swim on vacation. I suppose the alternative solution is just to go back on vacation. Again, in general I was focused on the most recent versions - with including both the VERTIX 1/2 being a case where I thought I'd have this published before the Vertix 2 came out. Though as a bonus, you actually get data for both (and, they're dramatically different)!

Now as a quick primer, openwater swim GPS accuracy is significantly different than regular running/cycling/whatever accuracy. That's because each time your wrist goes underwater for the pull portion of the stroke, it loses GPS signal. And then as your wrist/arm raises above the water, it regains it - for about 1 second. And then it loses it again. Rinse, repeat (literally). Each time it sees satellites it likely doesn't have a great lock on those satellites or its location. Thus, it's often plotting imperfect points, perhaps 10 meters incorrect, or even 50 meters incorrect. Thus, the 'secret sauce' for companies is an openwater swim algorithm to make sense of all that messy data and produce a 'clean' track with the correct distance.

Back when then Garmin introduced an openwater swim mode over a decade ago to the FR310XT, you'd hope to be +/- 15-20%. But these days, many watches can get down to the low single-digit levels, such as +/- 1 to 3%. Oftentimes even correctly showing a perfect corning around a swim buoy. It's super impressive.

Ok, with that quick overview out of the way, let's dive into the test setup.

The Testing Details:

For each test round I placed a single watch per wrist. While doing multiple watches per wrist would mean more data, it'd significantly reduce the accuracy of said data. Having two watches per wrist would invariably impact the satellite view, and given the watches have a difficult enough time as it is getting satellites signals in openwater swimming, there's no reason to make it worse. Plus, it wouldn't be realistic to what a normal person does.

Next, floating behind me I've got a swim buoy (which also holds my phone/keys/flip-flops/clothes). Atop it is a GPS watch that's set to running mode with 1-second recording. This records a 'clean' reference track, without any interference of an openwater swim algorithm. It's our reference. For all these tests I used the same watch, a Forerunner 945. I and many others, including virtually every wearable company I know of, have done it this way for many many years. This without question this produces the best reference tracks available. In one test, I even put two watches on there:

At the start of each swim, I'd let the watches get GPS signal for 10 minutes while I walked to the beach. Then, I'd also let them continue getting/confirming GPS signal another 5-8 minutes while I got all my junk sorted beachside. Finally, before pressing start on the watch once in the water I'd ensure the watch said it was fully green and good (GPS-signal wise). Then, I'd press start and *WAIT* another 15 seconds before actually swimming (ensuring they could record an accurate start point).

I then set out for a swim. In the main bay I was testing, there were channel/boat buoys, that I used as reference points and simply connected the dots. These allowed me to have relatively consistent testing each day as I was swimming the same route.

For each swim I'd make 1-2 'stops' along the way. For each of these stops, I'd simply tread water about 15 seconds or so. The reason was to replicate simply taking a break, but also swimming in groups (where you might pause to wait for someone), as well as in many mass-start races where the first swim buoy can often be crowded and result in a traffic jam (which may put watches underwater for short periods of time as people get unstuck).

Finally, at the end of swims, I'd take the watches out of the water and give them another 10-15 seconds to ensure they could plot their last GPS point 'above water', before ending the activity and saving it. For reference, when doing a race (or any swim for that matter), I recommend the 10-15 second above-water waiting period pre/post starting to actually swim. I find it produces the best results with negligible impact to your overall swim pace metrics.

As far as heart rate accuracy goes, I didn't care. Most companies in this space say that heart rate accuracy from optical sensors underwater is [uneasy shrug] at best. And I still haven't found much use for tracking my heart rate real-time during a swim. So we're skipping that here.

Finally, I wouldn't focus on the exact pairings for each round. I roughly tried to match up competitors based on price or likelihood someone would be looking at those two options. But again, I only have two wrists, and thus, only two watches per round. Instead, focus on how they compare to the reference track. Almost all the swims were done at the same beach/bay, and in any case, all the swims were done in similar sunny conditions with no nearby trees/buildings/obstructions. And almost all these watches are using the same/similar Sony GPS chipsets anyway (except the Apple Watch and Vertix 2), so it really is more of an exhibition of the algorithms each company uses combined with antenna/case designs.

Round 1: Wahoo RIVAL vs COROS Vertix 1

Ok, first up is the Wahoo RIVAL watch versus the COROS Vertix 1. Again, I've got the Vertix 2 down below, but this initial set was done before the Vertix 2 was announced. Here's the high-level view - I've kept the reference swim buy data as red for all of these.

You can see that at a high level, the two tracks look very close. Distance-wise, here's what each reported:

If we zoom in, we can see that track-wise, the Wahoo RIVAL was very close much of the time, but kept making these little errors, which seemed aligned to the moments I did my tread-water tests. Meaning, it wasn't properly recovering from those. Here's one near the first channel buoy (Wahoo RIVAL in teal):

And then later on, another:

It's each of these misfires that contributed to it overshooting the distance by some 170 meters, as it added these little out and back phantom trips to the totals.

Meanwhile, the VERITX 1 was very close, with only minor GPS differences, as we often see in GPS tracks, such as this (VERTIX 1 in purple):

Ultimately, the VERTIX 1 won this battle easily, due to the RIVAL getting tripped up on satellite re-acquisition and adding phantom distances. Had it not done that, both GPS tracks would have scored fairly well.

Overall, the COROS VERTIX 1 put itself into the top tier category with this swim. Note that placing itself in this category has nothing to do with how it ranked against the other watch on the other wrist, but how it compared to the reference track and where I actually swam.

Round 2: Garmin FR945 LTE vs Suunto 9 Peak

On this second day of swim testing I went to a different beach. Simply because that's where the family went, and thus, where I went. But water be water, and the testing was to be done. For this round I'd have the Suunto 9 Peak compared to the Garmin Forerunner 945 LTE. Interestingly, this was a combination I tested twice before, once in my FR945 LTE review, and once in my Suunto 9 Peak review, earlier this summer.

You can see that at a high level, You can see the Garmin FR945LTE and reference tracks are vey similar. However, you may also notice that the Suunto 9 Peak seemed to drown a few hundred meters into the swim. Here's the summary details:

Now, outside the obvious that the Suunto 9 Peak did poorly here, one might assume the FR945 LTE was inversely "perfect", and yes, the distances exactly matched, but keep in mind there was still some slight differences between the reference track. The way averaging works, it can undercut one section and overdo another. In fact we see that. But first, the Suunto.

I started off my swim, did about 250m and then paused for my first treading-water test. And it's here that the Suunto apparently got distracted with the fish and never came back. You can see the teal line just ends there, despite the timer continuing.

Interestingly, I saw this exact same behavior on the Suunto 9 Peak with some near-final beta firmware in early June. Yet later on with final firmware in July, I thought the issue went away. In reality, I know realize it was likely that particular swim. In my June swim I paused to tread water a few times, and it lost GPS then. Whereas in my July swim, I just swam non-stop, and it tracked OK. Ironically, Suunto has a long history with this sort of issue, one that I've encountered multiple times over the years on different Suunto swim products.

In any event, with the Suunto 9 Peak out of the race, let's look at the FR945 LTE's track. You can see on this 180* turnaround (which I did without stopping), it was a bit confused, and didn't quite nail it. In some ways it's a great example of how tricky openwater swim algorithms can be to nail.

Similarly, at the the next turnaround, it gets a bit off-track too. Though, do keep in mind how zoomed in I am (see scale at lower right). This shows an offset of only about 10-15 meters.

Still, through the magic of averaging (undercutting slightly in one spot, and overshooting slightly in another), the actual distance ends up being exactly the same.

As such, given that the FR945 LTE gets placed in the top tier of units (and there's more swims on Sets 7 & 8 down below to match that), and the Suunto 9 Peak unfortunately falls to the lowest tier (especially given my history with this unit having this problem once before).

Round 3: Apple Watch SE vs Garmin Swim 2

Next up we're going to compare two units that are geared more towards everyday people than straight endurance athletes. We've got the Apple Watch SE competing with the Garmin Swim 2. Historically speaking, the Garmin Swim 2 has held onto the crown as the most accurate openwater swim watch I've tested (potentially ever). Scary good accuracy, producing openwater swim tracks that are almost indistinguishable from the reference track. Similarly, Apple also consistently scores among the best openwater swim watches I've tested, even if their tracks are usually slightly oversmoothed. It's historically always beaten Suunto & Polar with ease. So in theory, this would be the Battle Royale of openwater swim watches.

Note that for this and the Apple Watch Series 6 test, I completely shut off my phone, which was in my swim bag (so it wouldn't use the phone's GPS signal).

And in fact, looking at the data, these two units laid down some incredibly good tracks.

The Apple Watch SE & Garmin Swim 2 were so good in fact that the reference unit itself said 'Nah, I'll pass', and drowned itself out of fear of losing. No, really, it somehow got flipped over on the swim buoy, and spent much of its time face-down in the water, producing these squiggly lines - rendering that track good useless for total distance, but good enough to validate our general position. Obviously, that's my fault for not catching it's faceplant.

So setting that aside, you'll see there are zero abnormalities from either the Garmin Swim 2 or Apple Watch SE on these tracks. They are both smooth, though spaced about 5-10 meters apart the entire time. Before the reference watch flipped over, you can see how close these are, with the Garmin Swim 2 matching slightly more closely to the reference track than the Apple Watch SE (a trend that continues even when the reference track is upside down).

Ultimately, there was no obvious meaningful between either track when it came to the GPS plot itself, with only the actual distance being different (but no clear-cut way to decide a true winner). But I think anyone would be pretty happy with either of those GPS tracks, thus both units get placed in the top tier.

Now I feel I'd be remiss to point out that it's important to fuel properly post-swim. And the Corsican beaches take nutrition seriously. Thus, a man (ostensibly, but probably not named Sunny) roams the beach with his beignet cart. As luck would have it, said cart was on my walk back from this swim. Certainly, this seemed like a good time to ensure I had some recovery nutrition:

And yes, that's most certainly Nutella in there.

Round 4: Polar Vantage V2 vs Fenix 6 Pro

Now for this set I've got a special treat for you today. No, not more donuts, but rather *two* reference tracks. This somewhat happened by accident. I had walked to the beach with the FR945LTE on my wrist (which wasn't participating in today's test), and rather than just stuff it inside the swim bag, I figured it could go for a ride atop it. And the results are pretty interesting.

Otherwise, I've got the Polar Vantage V2 comparing against a Garmin Fenix 6 Pro. Technically the Fenix 6 Pro Solar, though I've seen no difference in accuracy between Solar and non-Solar variants of the same model sizes. Here's the plot:

As you can see, one of these is not like the others. In that case, it's the green one, the Polar Vantage V2 - which wobbled all over the place. Here's the distance summaries:

Garmin Fenix 6 Pro: 1,260 meters
Polar Vantage V2: 1,550 meters
Reference Track FR945: 1,350m
Reference Track FR945LTE: 1,350m
Data set linked here in DCR Analyzer

Yes folks, the two reference units managed to record *exactly* the same distance. It's really kinda impressive.

Anyway, starting with the Vantage V2 in green below, you can see it was doing swirly drunk dance moves most of the swim. I promise, I wasn't swimming like this:

Meanwhile, the Fenix 6 GPS track was virtually identical to the reference tracks. So much so that I'm mildly curious why the distance is 90 meters off, as it's virtually identical, save this single point where it slightly juts out about 5-7 meters:

In case you're wondering, the track lengths/distances listed above are what each unit reports as the final distance. So the DCR Analyzer pulls that in (we don't recalculate it). Thus, it's what you see on the watch itself. Units can report one value as the final distance, while if you were to add-up the GPS track, it might show another. Which is how we can get scenarios like the previous Apple Watch SE vs Garmin Swim 2 sets where both tracks look near-identical, but yet have different distances. And, in this case, it's also how despite the Polar Vantage V2 clearly visually looking horrific, 'only' added about 200 extra meters.

In any event, after this swim, the Garmin Fenix 6 got put in the top-tier bucket, and the Polar Vantage V2 got put into the middle-tier bucket. One could argue it goes in the bottom bucket, but given that's where the Suunto 9 Peak is hanging out after its performance, it's like comparing a mob boss to someone who stole some tacos from a taco truck. In any event, you can decide for yourself how they should align.

Round 5: Garmin Instinct vs Apple Watch Series 6

This round was a shorter round, since I had my wife swimming with me, and thus we had pawned off our children to friends to try and contain for a short period of time. So we didn't have a ton of time. But it's interesting because we then had two sets of data (mine and hers) from the same swim, which I've shown back to back. Also, it means two different people swimming (with different swim styles).

The Apple Watch Series 6 and Garmin Instinct Solar are probably different markets for most people, but as I've said a few times now, I only had so many wrists, and thus, such is life. Here's the data set:

At first glance above and below, you might assume these were similar, but in reality, the Instinct Solar had some bobbles here, but again, through the magic of different distances than tracks, seemed to figure things out. More on that in a second.

Now you can see here that despite starting the Instinct well ahead of the swim start, with full GPS, it didn't seem to get a lock. So it started it's GPS track some time later. And then again at the mid-point of the swim it seemed to just sorts give-up, and connect the dots.

What's ironic here is that in the middle section where it missed it, I actually had the Instinct Solar above the water taking video of my wife swimming (while the Apple Watch was below the surface), and yet it still didn't figure it out. Yet somehow, inexplicably, the Garmin Instinct Solar got a roughly correct distance. Interestingly, if you look at the cadence vs speed data (in teal below), you can see it was tracking right from the beginning - yet wasn't producing any speed data (and thus, no distance data) until the 3-minute marker. Then it woke up and was like "Woah baby, how'd I get way out here?!?"

So while the track at a high level looks almost correct, it was mostly dumb-luck that it got it right (aside from the late start). Though, once it finally got itself on-track and fully locked, it nailed things. Also, frankly I don't understand how it got distance right, except I presume they must look at the stroke (cadence) data and do some guesstimating.

As for the Apple Watch Series 6 track, it was just like the Apple Watch SE - virtually perfect:

Yet distance-wise it overshot a bit. Again, slightly perplexing things, but, that's the data hand I've been dealt.

So, wrapping up this one, the Apple Watch Series 6 easily gets into the top-tier bucket. Whereas the Instinct Solar is sorta a perplexing one. Once it decided to play along with my swim game, you can see it was spot-on. And technically it had good distance too. But obviously, it started late and also skipped chunks of the track, thus, it gets relegated to the mid-tier.

Round 6: Garmin FR945 (Regular) vs Suunto 9 Peak

'Wait a second' I hear you say, didn't we already do the Suunto 9 Peak? Why yes, we did. But I had a spare swim, and a spare body with wrists (my wife) one day, and wanted to see if the results would be any different. Plus, while I had been using the FR945 attached to the swim buoy for all the tests, I figured it was time for the non-LTE FR945 to take a dip too. Thus, my wife got both of those affixed to her.

And off we went!

Now somewhat interestingly, both the Suunto 9 Peak and Forerunner 945 reported substantially different distances to the reference track (which was my swim buoy, swimming 1 meter to her side). This was interesting, because my two watch numbers (listed as Round 5 above) matched the swim buoy, but not hers. Yet, her FR945 GPS track visually matched virtually perfectly with the reference track. I have no logical explanation as to why this is. Seriously, I have no idea. But, I do know that the Suunto 9 Peak still produced oddities (note: The swim buoy track switched colors here to purple):

As you can see when you zoom in, the Suunto 9 Peak (in teal) seemed to have issues at points we paused, though, did manage to get back on track most of the time:

Meanwhile, the FR945 on her wrist seemed to basically perfectly match the reference track that was a meter or two to her side (attached to me):

Interestingly, despite the Suunto 9 Peak's random track bits, and inversely that the Garmin FR945 was near perfect, both different fairly substantially to the reference track in distance (higher). And both of those were offset substantially to the watches on my wrist (which matched the swim buoy). Frankly, that makes no sense - and makes me wonder how much stroke style might play a part in it.

My Wife's Garmin FR945: 980 meters
My Wife's Suunto 9 Peak: 990 Meters
Reference Track: 690 meters
DCR's Apple Watch Series 6: 760 meters
DCR's Garmin Instinct Solar: 730 meters
Data set linked here in DCR Analyzer

As a smaller female she has a higher cadence than I typically do (stroke rate) and thus, I wonder if the algorithms are overcompensating for that. I suppose that's something that could be dug into in the future, though, I doubt I'm going to convince her to do a bunch of openwater swims in the chillier waters of Amsterdam. Corsica was an easier sell.

Rounds 7 & 8: COROS VERTIX 2 vs FR945LTE/Fenix 6 Solar:

(Quick Update/Preface: Just a couple hours before hitting publish, COROS sent over a new firmware update for the VERTIX 2 that they believe may solve the openwater issues I saw. They said they found a bug that prevented the unit from loading the correct satellite information when in openwater swim mode. I'll attempt to test that in the following days and update accordingly.)

Now Round 7 & 8 weren't technically on the planned menu. And as you can see, the water doesn't at all look the same. Gone are the crystal clear waters of Corsica, and enter the murky waters of Newfoundland, Canada. You see, I had figured this entire post/video would be published before I the COROS Vertix 2 was announced. And thus, it wouldn't have been included. But alas, sitting around drinking rosé-all-day in Corsica meant that I never got my stuff together enough to edit everything. And before I knew it, the COROS Vertix 2 was announced.

However, as part of that review I had started while in Newfoundland to do openwater swim tests (which were also posted in that review). This is notable because this is a different GPS chipset (not Sony), and also dual-frequency. In theory, that's the holy grail of GPS accuracy potential. Unfortunately, at this stage and this firmware, I'm just not seeing that. Here's those results:

You can see that the COROS Vertix 2 struggled, but did so especially once I paused to tread water towards the bottom right (I swam clockwise here). Somewhat similar to the Suunto 9 Peak's issues. At that point, things went poorly. While I was keeping an eye out for moose, I did not go off into the woods on this swim, as the Vertix 2 assumes:

And you can see, it never really recovered:

The FR945 LTE was pretty much within a meter or two of the reference track the entire way. Here's the final stats for this first swim:

Now, the above set was arguably the worst of the two COROS sets, but there was also a slightly shorter swim at the same location, where it did better.

Here's a second set with the Vertix 2, and the FR945LTE now swapped out for a Fenix 6 Pro Solar:

Note, the swim buoy reference track got paused or something in the last couple mins, I'm not really sure what I did there - that's the first time that's ever happened to me, though it was pretty much a straight line in from there to the 'beach'.

While at first glance things look similar on the straight-aways, there are some differences here from the VERTIX 2, especially when I made turns, you can see how the Vertix 2 got a bit wobbly. Also, you can see on the lower right side how it actually had lost GPS entirely, and just connected two dots between two points I treaded water.

The Fenix 6 Solar meanwhile was basically glued to the swim buoy.

Now the notable part about the COROS VERTIX 2 is less it's wobbly performance, but rather, the GPS chipset. This is the first sports-focused wearable to have dual-frequency GPS in it. And while COROS won't confirm the GPS vendor, they will confirm it's not Sony. Point being, that while dual-frequency GPS is hailed as the holy grail, I suspect that COROS probably has a year or so working with this vendor to really flush out all the quirks. In the same way that when Garmin/Suunto/Polar all switched over to Sony (from MediaTek mostly), it took about a year or more before things really stabilized. One can remember my incredible frustration with Garmin's MARQ Athlete watch and openwater swims when it first came out on the new Sony chipset.

Figuring out regular land GPS accuracy on new chipset and new chip manufacturers is tough, doing it for openwater swims is even more challenging. Point being, while I wouldn't pick a Vertix 2 for an openwater swim today, dual-frequency chipsets are the future, and ultimately companies will sort this out.

Perhaps the most interesting thing about this entire test, aside from the donuts, is that these results almost perfectly matched what I saw in my in-depth reviews for each unit - even reviews for products 2+ years ago. Certainly, it's plausible to have a bad openwater swim track. I have no doubt the comments section below will be filled with plenty of 'One time at band camp my GPS watch was ate by an Orca and had a bad GPS track' comments. And that's not to dismiss those, but rather acknowledge that one-off testing has it's limitations (both good and bad).

That said, almost every watch here I've swam with multiple times beyond this mega-test, numerous times in many cases. And these one-off tests surprisingly matched my wider tests in virtually every case.

So, in summary, I'd basically put things into three buckets, almost entirely by brand. Which again, doesn't surprise me - at this point it's more about the vendor's openwater swim algorithm and chip groupings, than it is about any given specific watch. Of course, one could quibble that this only applies to the exact models above, but I've found that as long as we're talking generationally the same sets of models, it carries through. For example, the FR745 and FR945 are incredibly similar under the covers, whereas the original Fenix 5 to the Fenix 6, slightly less so. I noted the Pace 2 below (same chipset as Vertix 1), because that did very well in my previous openwater tests. I haven't tested the APEX/APEX PRO anytime recently in openwater swimming (though I'd expect them to be the same as the Vertix 1/Pace 2)

Best Swim Accuracy: Apple, Garmin, COROS (Vertix 1 & Pace 2)
Middle of the Road Swim Accuracy: Wahoo, Polar, Garmin Instinct
Not Great Swim Accuracy: Suunto, COROS Vertix 2 (currently)

Note that it's worthwhile mentioning that two summers ago (2019) after my MARQ failures (and subsequent frustration post about swimming accuracy), Garmin spent the entire summer and much of the fall trying to fix issues related to a more recent GPS chipset firmware that tanked many peoples openwater swim accuracy. The company said they put engineers and test groups on "hundreds of swims" over the summer trying to ensure they had fixed the underlying issue. It took a number of months, and we saw the details of those fix attempts in various public beta Forerunner & Fenix updates during that timeframe, before finally being rolled out to everyone. I rarely hear of swim issues anymore on those impacted devices.

I only mention that because there are countless stories of many bad swims from that spring/summer 2019 timeframe (or, people since that still haven't updated watches).

In any event, there's all the swimming I've got for now. I'm sure as the fall timeframe brings more wearables, I'll find myself out in more puddles of water poking at things, but will likely reserve doing that for watches that have a specific openwater swim mode.

With that - thanks for reading!

Found This Post Useful? Support The Site!

Hopefully you found this review useful. At the end of the day, I'm an athlete just like you looking for the most detail possible on a new purchase - so my review is written from the standpoint of how I used the device. The reviews generally take a lot of hours to put together, so it's a fair bit of work (and labor of love). As you probably noticed by looking below, I also take time to answer all the questions posted in the comments - and there's quite a bit of detail in there as well.

If you're shopping for the Openwater Swim Buoy or any other accessory items, please consider using the affiliate links below! As an Amazon Associate I earn from qualifying purchases. It doesn't cost you anything extra, but your purchases help support this website a lot. Even more, if you use or Competitive Cyclist with coupon code DCRAINMAKER, first time users save 15% on applicable products!

And of course - you can always sign-up to be a DCR Supporter! That gets you an ad-free DCR, access to the DCR Quarantine Corner video series packed with behind the scenes tidbits...and it also makes you awesome. And being awesome is what it's all about!

Thanks for reading! And as always, feel free to post comments or questions in the comments section below, I'll be happy to try and answer them as quickly as possible. And lastly, if you felt this review was useful - I always appreciate feedback in the comments below. Thanks!

Volver a la Portada de Logo Paperblog