Swim analysis with the Garmin 910XT

So I did a 2000m TT with the Garmin 910XT on my wrist. What did it tell me?

garmin swim graph

Sweet FA if I look at the graphs on the Garmin Connect website. The scale on their graph is rather unhelpful – I either swim at 0 mins/100m or 2mins/100m. The Garmin website has a way to go on this.

They do however provide useful data tables which show you your “Average Pace” and “Average Strokes” (per length), although they also do their best to hide this useful data with a whole lot of useless data like “SWOLF”, “Efficiency” and “Total Strokes”. But if you scroll across, you’ll find the very useful “Average Strokes” tucked on the end.

useful garmin data

You can also see a breakdown of times for each length: 55.4, 1:02.6, 58.4, 56.8 etc. Which doesn’t mean much, until you export it into Excel and display it:

Raw 910XT swim data

The red line is my time in seconds per lap (the dotted red line is the average). The green bars are the number of strokes per lap (again, the dotted green is the average).

Looking at the dotted lines, you can see that my pace increases throughout the time trial, and the faster I go, the more strokes I take. That makes sense. I was swimming steadily, building speed as got closer to the finish. I was definitely increasing my stroke rate towards the end.

What doesn’t make sense however, is the solid red line, showing how long each lap takes. According to this, I swim every second 50m 10 seconds faster? What the?!?

OK, so now I’m wondering if it has a little trouble telling when I’ve touched the wall. Maybe I turn to a different side at each end and that throws the readings out? Maybe there’s a downstream current in the pool? Maybe the pool is on a slope? (HAHA)

I was swimming using the pool clock – I’m pretty sure I was swimming the first 400m on 1:55s. By averaging each score based on the scores next to it, I get a better graph, but I’m not going to do this for every swim!

Smoothed 910XT swim data

I’m curious, how wonky is the auto-lap feature? I thought I was doing the same touch technique (open turn) at each end, but maybe not? I did a swim with a second watch (one watch on each wrist – clumsy!) and MANUALLY clicked each lap. Unfortunately, on the 2nd lap I hit the “Stop” button instead of the “Lap” button. 40 laps later I realised I’d been wasting my time! That certainly made me appreciate the auto-lap feature of the 910XT!

I guess I’m a bit disappointed with the swim features of the Garmin 910XT. I was expecting these new swim measurements would tell me something about my swimming; give me something tangible to work on. Like perhaps my strokes per length are too low/high, or my stroke rate declines as I get tired, that I’m over-gliding or under-gliding.

When I got the speedo for my bike, I could look at my speed and know that faster was better. When I got the power meter I knew that more watts were better. I’m not sure what the 910XT gives me. I don’t know what a 22.7 stroke average means. Is it good or bad? When I swim with a rashie on my average changes to 24.0. Can I learn something from that?

Perhaps it’s just that the Garmin Connect website doesn’t offer any analytical features. It doesn’t for run or bike either – I export the data for these activities into WKO+ and Strava to analyse them. But neither WKO+ nor Strava help with swimming data right now. Apparently Training Peaks does.

I think there’s a real opportunity for someone to position themselves as the ‘Swim Data Expert’, much like Joe Friel and Hunter Allen have done with bike power. People are starting to capture this swim data but don’t know what to do with it. Someone can advise us that Lance swims at 20 strokes per 50m so we all should (much like the 90 cadence mantra).

And with the 910XT accelerometer on my wrist, there must be a tonne of other data they can capture and report on:

  • the acceleration of my arm during the pull phase,
  • the length of time during a stroke spent in a glide position,
  • the speed of my arm during recovery,
  • the length of my stroke,
  • sideways forces during the catch, etc.

Surely someone with a good knowledge of stroke mechanics can pull this data apart and provide some good stroke analysis. Swimming is such a technical sport – everyone needs technical guidance!

And what else is coming? I imagine that heart rate via the watch strap, power via ‘power paddles’, and leg power via ‘power flippers’ are all in the works. Combine these features with auto-generated training programs like www.swimplan.com, add a waterproof iPad propped up at the end of your lane telling you your next set, and we’ll all be smiling!

And, for $20 extra, switch the front camera on and your online coach can watch your technique in real-time and give you some feedback on it. I don’t know if there’s an app for that yet, but you know it’s coming!

This entry was posted in time trial, training toys and tagged , , , , , , . Bookmark the permalink.

11 Responses to Swim analysis with the Garmin 910XT

  1. Russ says:

    Interesting to see data from the 910, similar to the Finis, though based on your evidence my Swimsense is a little more reliable in terms of reporting speed.

    The SWOLF and efficiency scores are supposed to tell us something about our swimming – it being a combination of stroke rate and stroke length that determines our speed, but in isolation I think you’re right you can’t say too much about the information. It might help to perform a Swim Smooth ramp test – http://www.swimsmooth.com/ramptest.html the watch could help there. Otherwise to swim faster you’re either looking to increase stroke rate and/or increase stroke length, at the very least if you use the watch you’ll see if it’s happening.

    Biggest problem with all this swim data, you see it after that fact, come back, do some drills and test again. Even with readier access to the numbers, actually changing swim technique to any significant order remains a slow process. With my Swimsense I’ve been watching the numbers coming out of training just to get a feel for where I am – what I notice at a lower level of swim fitness, stroke length has dropped slightly, but stroke rate has significantly fallen for now; if I try and increase the latter you can watch the length fall!

    So all that’s needed is multiple motion sensors, a new iPad with HD video and the data connection to have a remote coach. Throw in a small ear piece so instructions can be given while swimming and you’re set! Actually it would be interesting for someone to capture and plot the motions of a wide range of swimmers, just to see the variations that occur, both for the fast guys and the slower ones.


    • jontsnz says:

      Thanks for the tip Russ. I might try the Ramp Test this arvo. It’ll be entertaining to see if I can up my stroke rate to 20/min more than I do now!

      In theory I can look at this data as I go. The screen can show SWOLF, stroke rate etc. I sometimes modify my technique (try and push harder to my thighs, catch shallower, elbow higher, tuck my chin in tighter, etc) and can never really tell if things are better or worse. Perhaps I will do these in controlled sets of 400m and look at the data then and there. But the improvements are so small I’m not sure they’d show up.

      Swimming does strike me as the one sport where everyone can make just as significant improvements through technique as through endurance training. If you can come up with a scaleable way to teach technique, you’ll be onto a winner!

      • jontsnz says:

        Forgot my watch, so the ramp test didn’t happen on Friday.

        Got me thinking over the weekend though, and I’m also going to try a Ramp Test for cycling: maybe 5 min intervals starting from 105 cadence and working my way down to 70 with large rest between. Will be interesting to see what it reveals about 90 cadence for me!

      • Russ says:

        You may find that on the bike, 5 minute intervals won’t tell you enough. I think with the swim ramp test, the significant impact of technique on speed through the water makes the change in stroke rate and its impact on stroke length quite obvious. What you see is a definite deflection point, where raising stroke rate results in reduce speed due to shortened stroke length.

        On the bike you’ll not have that change, you’ll be effectively looking to find which cadence makes you the most efficient and over 5 minutes, you can probably make some judgement about RPE at constant watts (or HR if there’s nothing else), but it could be the case that you rarely train at RPE 100 so while it is most efficient, it feels wrong. Raising or lowering cadence isn’t going to show and obvious alteration in speed if you’re maintaining constant watts, it may show alterations in RPE, but I think you’ll need longer to really get a feel for this.

        Worth experimenting though and having an idea of your response to differing cadences, certainly can’t do any harm. I tend to let people opt for a self-selected cadence, maybe throw in some low cadence, big gear work from time to time, but don’t get too specific as to how fast you should turn the pedals.

        And of course I watch cadence on the run too. There the reason for higher cadence is as much reduced injury risk as efficiency.


  2. jontsnz says:

    I guess I was thinking that I’d hold the same power through a range of cadences, and see what effect it has on HR (or vice versa).

    I’ve certainly noticed that I self-select a much lower cadence when climbing hills (which I am much better at than riding flats) – low 80’s up a steady incline rather than the mid-low 90s I do on the flats. I’m wondering if me being told that you should ride at 90 cadence, while I was new on the bike, might have cost me speed? While I self-select a high cadence on the flats now, I’m not sure that’s my natural cadence. A test will be interesting, but not conclusive by any means.

    This line of thought was probably triggered by Brett Sutton’s recent blog on low cadence cycling. http://www.teamtbb.com/?option=com_content&task=view&id=1363

    • Russ says:

      I’d seen the discussion following that TBB post over on Slowtwitch, some good stuff in there (with the standard caveat that you have to sift through the dross to find it)

      I think the problem with a 5 minute test, especially if you’re trying to compare HR and power, is it’s so short, you may not find much. HR is a pretty variable metric of performance, why I always prefer to think about RPE in conjunction with it (and ideally pace and power too). Of course you need some way of assessing the impact of changes to cadence. I’d suggest running it as a series of tests, perhaps vary by 10rpm at first to narrow the field and do at least 20 minutes. Ideally average a few and ensure you’re rested to maintain roughly the same conditions. Then you might see some more consistency over HR and RPE.

      More broadly, playing with gearing can help a lot, my self-selection has tended to a slightly lower cadence, but it rises slightly when rested and racing. One of the things you’ll see on the Slowtwitch thread is one aspect of cadence is it may vary according to the power you are riding at – i.e. to sustain a much higher power most people will see an increase in cadence (when I did a RAMP test for power in a lab last year, I was hitting 100+ for most of the test, I never ride at that in any other circumstance.) That in itself definitely points to the potential relevance of lower cadence riding to long distance time trial events like an Ironman.

      If you have WKO+, and perhaps in Training Peaks now, you could have a look at the Quadrant Analysis from a number of rides. Set preferred cadence and FTP appropriately and see how your pedalling distributes. I think you can then select intervals of work and perhaps see how you achieved different blocks of work. Relate that in a long ride to feel, HR etc and perhaps see if there’s a trend in general riding or racing.

      Ultimately – something to experiment with for sure. I have a preference for lower cadence/higher force riding in training generally, seems to be very effective in developing my cycling, but I do throw in some higher cadence stuff and generally my Sweet spot and threshold work will be much closer to 90rpm….

  3. Russ says:

    One more thing…

    Out of curiosity, because I’ve had WKO+ opened for a week now looking at long term training data and because I’m procrastinating over writing up a blog post on it. Here’s a quick graph of the 4-5 year trend in my cycle cadence:

    I’ve also plotted normalised power and heart rate averages for all rides during that period; it gives something to compare against. This is all rides, the only selection I’ve made is removing 0 values which tended to indicate an issue with the data being recorded.

    What does it say? Well I do ride harder these days than I used to, power has improved, but worth noting I also tend to train at a slightly higher HR than I used to; this seems a fair reflection of how my training has changed, apart from being fitter now than then, my approach has trained and I certainly push more now. So cadence, other than a patch of little ring only work over the winter before last I’ve never consciously done anything to target specific cadences, it’s always been quite low and has actually trended up slightly; perhaps that month or two of spinning over a year ago did it, or perhaps it simply relates to the harder riding?

    Okay, so more procrastination, had to draw up another chart to compare power and cadence more directly:

    Clear trend upwards in power with cadence, not so much for heart rate though, but given these are averages over every ride of every type, perhaps other factors mask the degree or nature of the trends.

    Right, I’ve done far too much on this and not enough on my planned blog post!


    • jontsnz says:

      Hopefully all this thinking has given you some ammo for a blog post down the road!

      Really intrigued by your first chart. Your cadence creates a wandering trail over the years, but the trail is never much more than 15 rpm wide, even though it’s as high as 100 at some points, and as low as 55 at others.

      I’m tempted to try to plot mine – but I should also get back to work!


      • Russ says:

        I’d not really noticed that with my cadence, possibly it reflects the impact of fatigue at certain periods of time; though at least one period relates to the small ring only policy, leaving me know choice, but to spin. I tried pulling some simple metrics – duration of cycling and running along with TSS – to see if there was a correlation there, a crude attempt to measure that impact of fatigue, there wasn’t anything significant. Might be worth comparing with performance management chart metrics – ATL and TSB particularly – as a better indicator of fatigue.

        Charts will come in handy for the blog post, which probably won’t happen until this evening after a big day of training…

        In the mean time, I have produced one more chart, a weightier comparison of lots of data from the last 4 years:

        That one is definitely for the blog.


  4. Marcus says:

    I took my new 910xt for it’s first swim on the weekend (http://connect.garmin.com/activity/158652107 – ignore the erroneous spike in the middle!), and I agree completely with your data-capture wishlist, and your point about the Garmin Connect speed chart being completely useless (mine is all under 0 according to the scale?!).

    With regard to your query on where your stroke rate sits, take a read of http://www.swimwellblog.com/archives/1340, along with any articles on how Alex Popov trained – slowly, and enforcing perfect technique with a deliberately low stroke rate around 10 spl (both arms) less than what he’d race at.

    I’d say your stroke rate can and should be lower – you should be moving faster for how fast you’re spinning your arms 🙂 But of course, exactly the same applies for myself, so I’m not singling you out. If I’m concentrating on it in training, I try to aim for less than 30 strokes (both arms) per 50m – at least having this watch now will help me easily identify if I’m hitting that when I’m not focused on it!

    • jontsnz says:

      Hey Marcus, nice swim! 6km non-stop!! Will have to try that one day.

      Great link you sent me – timely too as I’ve always focused on an efficient glide/catch/pull, but have lately been considering upping my stroke rate to improve my speed. I’m rethinking that now.

      Sun Yang makes it look so easy at 26 SPL! I see he trains at my local pool – I’ll have to try to catch a glimpse of him in action. 1500m in less than 15mins – twice as fast as me with less strokes! Mind-boggling. Amazing how high keep can keep his hips/legs while barely kicking too. I’ve got a way to go yet!


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s