It’s pretty pessimistic too. Average upper echelon male sprinter all age groups is not as fast as your reported speed. And you do not self-identify as upper echelon.
Yeah, but, again, I’m not doing a standing start, and not running as far.
From your link, top speed averaged over 100m for 50-59 male elite athletes was 18.8 mph. So their peak speed is probably around 22 mph.
These guys are specimens who undergo rigorous training, and they are running that one race. You are running around the field for a while so your legs are already a little fatigued. Yeah, it feels like you’re running top speed when that disc is floating just in front of you, but you’re not running as fast as you would fresh.
Like I said above, I think 17 mph would be pretty damn fast, but it’s possible. I really don’t see any way you’re hitting 20 mph.
Was the 20mph a sharp peak or several seconds long? If you are running full speed then lunge I could see your watch going pretty fast, but only for a split second. A good strong backhand huck with windup and follow through would also get your wrist moving fast for a yard or so.
I don’t think the GPS is going to pick up a huck, but in any case, I was running, no diving or lunging. It’s a spike, but all of my peak speeds are spikes on the graph. It’s the nature of Ultimate – cut in or cut out, walk around, cut, etc.
I’m not going to say your fantasy is impossible. Improbable sure. But it is clear that sampling once per second with an accuracy of plus minus 5 to 10 meters will be very insufficient to calculate brief bursts of speed of several seconds with meaningful accuracy.
Again my forearm device run through a paired iPhone is clearly reporting a peak speed roughly 3 mph faster than my pace. Looking through my runs history ranging in duration from 50 minutes to just over 2.5 hours, many of which report very consistent mile splits with no sprinting and every effort to stay at a consistent pace, that (false) peak velocity happens near the same number every run.
These devices are not designed for short sprint or burst speed data reporting. At that granular level noise swamps signal.
Accuracy might be plus/minus five meters, but what about resolution? Can the GPS receiver calculate position to a higher resolution than five meters, even though that position will have an error of up to five meters from the actual position?
This seems to be a measure of how fast you swing your arms. That’s going to be pretty constant when running in a straight line. In Ultimate there is a lot of sudden changes in direction as well as a lot of relatively wild arm motion as you try to block or intercept a throw. Think of someone guarding a basketball player after they have picked up their dribble or a cornerback trying to break up a pass. I can easily see a relative arm movement much faster than normal running.
A receiver calculating once per second is not going to notice a relative arm movement. In any case, for these particular measurements (which, as I mentioned, I remember the situation) I was running in a straight line, chasing down a disc and a receiver.
Let’s say you want to find someone’s top speed over 5 meters, starting fresh, with a running start. I don’t think the speed is very different from people’s 100m average speeds (for trained people). Maybe a little bit higher.
So just check out average 100m times for people of various ages and fitness levels, and you get a good idea of how realistic it is.
" His average ground speed during this feat was an astonishing 37.58km/h. Additionally, he achieved a top speed of 44.72km/h in the 60-80m stretch, an astounding performance fitting for the world’s fastest man. It’s no surprise that Usain Bolt won the gold medal."
That is a 27 mph top speed. About 18% faster than his average speed on 100m.
So, maybe 18 or 19 mph, I guess. Could be.
Do some practical measurements.
Try the “max treadmill speed” as I said earlier to get a baseline for what “pretty fast” feels like, and what its real speed is.
Then go out to the track and set up a couple of markers a a short distance apart (e.g. ten or twenty yards) and do a ramp up to an explosive short sprint, going way past that top treadmill speed as you pass the markers. Hit a stopwatch as you pass each (might need a friend to do this).
Surely there will be clock error (especially if using your phone as a stopwatch), but if you have a long enough distance the error bars in speed calculation ought to be reasonable.
One thing I realized some years back: even though I run regularly, I still cannot run as fast as tween children. I would take my kids to the park and we would chase each other around and they always could beat me in short bursts. I don’t do sprints or other speed training, so it’s most likely down to not training for it, but I took it as a sign that age was having its effects.
Sorry but it is not. The below is spot on. It is not using the accelerometer.
In any case the point was that even my long slow run, a very slow consistent pace other than stopping at red lights and certainly no sprinting, has some spot sampling that calculates me at 3 mph faster than I was going. This (noise) evens out over the one mile splits.
Using measurements from these devices for bursts of speed is something like using a yardstick to measure the diameter of a grain of sand.
Anyway. Found a recent article using state of art technology measuring young adult male sprinters hitting max velocities of about 9.44 m/s. That comes to about 21 mph for young adult male competition level sprinters.
Guessing what fraction of that young male sprinter max velocity you think a 50s aged week end athlete is capable of, and then what fraction of that would be achieved while playing Ultimate Frisbee is going to be less inaccurate than using the Garmin data.
You can measure out some markers, and then just place your phone a distance away and set it to film your run. You can then accuratly get times from the video.
Nice! It has the added benefit of be a geeky technique that will leave us feeling like we did some science.
That’s a cool idea. I also have a treadmill at home, but I don’t know what the max speed is.
As I mentioned, in my mile run around a track, my watch didn’t show this behavior.
Anyway, I appreciate all the input. I’m happy to keep the discussion going, but I think we’re well past FQ territory. Weather permitting, I’m playing again this weekend and I’ll turn on all three GPS system sensors and see if that makes a difference. I’ll try out the treadmill idea and I’ll see if I can do something with yard markers and a phone.
You could also see if anybody else out there has a GPS tracker and make a point of marking each other, then compare your top speeds when you’re both going all out.
Sure, as if they could keep up with me… 
Oh ho ho, maybe you really are going 20 mph!
One mile is a much smaller set of sample than an hour of Ultimate Frisbee. I doubt my weird number occurs once a mile. It does occur at least once an hour. Plus I would suspect the algorithms are built for rates that are more steady than rapidly changing …
Anyway, yes 20 mph is possible, even reasonable … as a max velocity for a competition level young adult male sprinter. Lesser numbers expected for athletes who are not sprinters, with age, and less intense regular training.
That’s the factual answer possible.
Oh. Another idea if you want to be scientifical.
Go to the track. Warm up a few minutes then do a series of sprint intervals from a steady clip as separate measurements. See if you are consistently hitting near that 20 mph number. See if the numbers are just consistent.
It doesn’t answer the true number question but informs on accuracy some.