Stride length question

[WilliamHamilton]

I know that we can customize our stride length in settings.  This question is with regards to the automatic stride length.

Does the fitbit automatic stride length calculation calibrate according to pace, or is it a single number averaged across a range of paces?  When I run, my stride varies and I'm keeping the step rate mostly constant, so a single averaged number is going to be incorrect depending on my pace (whether I enter one or it is calculated). 

If it's not calibrated to pace, may I request that feature for new updates?

[JohnnyRow]

From Can I change my stride length in my Fitbit account? 

"After you track a run with GPS, your stride length updates automatically. For best results, run at a comfortable pace for 10 minutes or more."

To me, that means, it divides distance covered by number of steps to get stride length.

I don't know if that answers your question because I don't quite understand the 2 options you describe.

[WilliamHamilton]

First, thank you for the response JohnnyRow.  Unfortunately no, it does not answer completely.  That  my question doesn't really help me much probably makes it look a little nonsensical.

What piqued my curiosity was seeing pace and distance in the app after a treadmill run.  Surely it estimated that from the number of steps I took and what it thought my stride length was. It was about 25% off - no big issue since the treadmill tells me the pace and distance, but looking into it brought this question to mind.  It's because I'm a high-step-rate runner who varies his stride, and my step-rate varies just a little with different paces. Low step-rate runners vary their step rate and keep the same stride so it likely works pretty well for them.

If the Fitbit app is sophisticated enough, it could relate both step length and cadence (from the accelerometers) to a give pace for an individual runner, given enough runs using GPS. Even though only one stride-length number is displayed, with the basic explanation, that's what I'd expect even if the app had more sophisticated calculations.  So I'm asking if something more is going on behind the scenes, and if not please look into it.  

If not, I'll simply manually set the stride length corresponding to my treadmill pace setting (I'm usually at the same speed anyway on a treadmill). But if it is collecting data, that calculation would improve over time and I don't want to manually set it.

[JohnnyRow]

I would expect everybody's stride length varies, with speed and especially going up or down hill.  I doubt Fitbit tries to adjust your stride length depending on speed, but I don't have any inside info on that, and doubt anyone here does, but if really important to you, it does seem like something you could somehow test yourself.

Another thing to consider on treadmill, which might not affect you: if you are holding hand rail, steps might not get detected.

[WilliamHamilton]

Everyone does vary their stride length - that's my basic assumption for the question.  Low cadence runners, particularly recreational runners, tend to vary it much less than do high cadence runners.  Because they vary the turnover rate more.

I don't know why you doubt that anyone knows - it's not all that esoteric.  I'll know myself after 8 or 10 outside runs with some speed work mixed in, by looking at the calculated number and relating that to the estimated pace on the treadmill. I just thought someone might know the answer here and chime in.

[WilliamHamilton]

Fitbit already does set a different stride length for a walking pace and a running pace, and that's how it calculates distance and pace when you walk or run without GPS.  Pretty sure it uses step-rate to decide which activity it is.  Maybe some other measurements as well, but almost certainly at least step-rate.

Can someone who does know something about it - FitBit help perhaps - please answer?  Simple yes/no: does the app only determine two step-lengths (walking vs running) or are there more pace levels not exposed to the user?

I only need to know in order to decide how I want to set my stride length (not asking for proprietary algorithms).

[LZeeW · Accepted solution]

@WilliamHamilton   Like @JohnnyRow, I am a Community Council member and not a Fitbit employee.  Let me share stuff I know (some of this you already know):

• Fitbit sets the walking and running stride length based on average measurements when you enter your sex and height in your profile. 
• There is no walking automatic stride length adjustment
• Automatic running stride length measurement is done from measured GPS distance divided by number of steps
• The app is not as sophisticated as you want it to be
• No feature suggestion is likely to get this changed
• How the running stride length adjusts seems to be proprietary.  I do not know if it replaces the value, merges the data from multiple runs, etc.
• When you walk or run with SmartTrack (automatic exercise recognition), you do not get a distance

I do not use automatic stride length adjustment.  This is my reasoning.  GPS is not as pinpoint as we think it is.  I can do the identical route five times, under similar circumstances, and get five different GPS distances.  When GPS decides where you are, your position is actually somewhere within a circle of about 860 sq ft.

I am fortunate enough to live close to trails with mile markers.  If you have access to something like this, you can do the following.  You can use the Run setting in the Exercise app with the GPS off.  Run known distances that go up and down hills and calculate your own average stride length for outdoor running.

Treadmill distance tends to be pretty darn accurate, unless your treadmill is not well maintained.  If this is really important to you, you can calculate your own treadmill stride length and your outdoor stride length and change the value each time before you start.  Or be like me and just put one foot in front of the other and enjoy the view.

[WilliamHamilton]

I get the feeling that respondents aren't grasping gist of the question. It's not that I want to know how far I went on the treadmill (just look at the treadmill display, it's dead on of course). It's also not that I want to know what my stride length is for a run (I already know what it is, for different paces).

The point of it all is that you have "5 miles" on the treadmill and "4.2 miles" on the Fitbit activity. Yes, I know that I ran 5 miles. I know that I can check the treadmill, that I can count steps etc. But the point is, the FitBit calculations are based off of this, and when those numbers are wrong, FitBit is wrong everywhere it utilizes that variable. 

FitBit apparently uses an unsophisticated idea of "stride length": one number for any "run", regardless of pace or step rate. It's a shame, because it wouldn't be all that hard given enough data for an individual, to more accurately estimate this.

[JohnnyRow]

I think you are looking for a lot more running-specific sophistication than what Fitbit offers, regardless of how difficult it would be to add.

I know there are specific running-oriented trackers by other companies selling for hundreds of dollars more, that might have such capabilities.

[t.parker]

@WilliamHamilton calculating distance with the stride is just an approximation. Fitbit devices don't use any sophisticated algorithm. When you change your stride length while running, there is no chance you will get an accurate result. How Fitbit computes the stride length when you use GPS? Nobody knows and I wouldn't expect Fitbit to reveal that. Fitbit devices are not really made for running if one wants accuracy with lots of details.

Actually, I don't think any watch can be accurate using stride length and it will always be poorly approximated. However, Devices from Garmin, Polar and Suunto can connect to foot pods (these are additional sensors made for runners). Foot pods are little devices attached to your shoe which can provide a lot more accurate metrics (not only distance, pace but also running power). Some of them require calibration. The one I use, Stryd, is not the cheapest but can auto-calibrate and can provide very accurate metrics even when running on a treadmill.

There is no answer to your question as a) nobody here knows and b) Fitbit doesn't ever reveal any technical details. If you are more serious about running, Fitbit devices are not for you. I would invest in more dedicated gear.

[WilliamHamilton]

@t.parker Fitbit divides GPS distance by number of strides to get stride length.  That's probably all that they do, and if so they're missing out on a fairly easy, fairly important functionality. It's about more accurately report treadmill/non-gps paces, not about analysis or knowing your stride length.

FitBit works so well in getting accurate data to other platforms and applications for analysis, that it's just a shame that the treadmill data is essentially useless.  And honestly it's not that hard a problem.  A simple parametric fit might even work.

[t.parker]

@WilliamHamilton treadmill is in fact quite hard to track accurately. The stride length will change with speed of the device. The best solution is to use foot POD sensor like Stryd, Garmin or Wahoo which you attach to the shoe and it can figure out your speed, pace and cadence from motion. Fitbit unfortunately is not able to pair with such device (or in fact any other external device) so treadmill won't be track with good accuracy. Actully, foot POD is great solution if you run also outdoor without GPS - lot more accurate than Fitbit's steps x stride.

[WilliamHamilton]

Both step rate and stride length will generally change for a runner at different paces - how much of which depends on the runner.

It may seem hard to estimate but you're only thinking about directly measuring it.

Have you considered correlating the measured parameters to pace, individualized to a runner's own data over consistent longer distances (when he uses GPS)?  I'd be willing to bet that a function of step rate, heart rate and fatigue (based on how long the exertion has been) correlates strongly with a specific runner's pace on a steady run.  Those three measurements are provided by a fitbit.  That derived individual function could be applied to the treadmill for a more accurate pace.

Rather than randomly argue about it, I think I'll work from my own data and see how well that works.