Dave Proctor (Anomalies?)

I would be very interested in seeing the screenshots you have of the Garmin tracker since the link no longer works as the session ended.

I've been reading the other Dave Protor run across Canada thread since it started but didn't look at anything in detail. Following your post, I downloaded all the Strava data from Dave Proctor's Strava account since his June 27th, 2018 start and ran it through my analysis tool. I have uploaded all the GPX files, resulting graphs, and analysis files onto my Google Drive in the link below.

From what I see in Dave Proctor's Strava data, I believe there is strong reason that there should be closer scrutiny of his run due to his data being highly anamalous and has indications that he may be using outside assistance to cover some of the distance. I think this warrants more collection of screenshots, photos, and data to either further this line of question or to refute all the anomalies in his data.

Dave Proctor's Run Across Canada Data So Far:

https://drive.google.com/open?id=19KbVoWuEJOJYcgFndtbqI0DXpe3vscLf

For example, creating a histogram of the speeds for Proctor's two runs on June 27, 2018 show that they both have an unusually high number of data points at 13 mph and above as indicated by the bar on the right most side of the graphs:

https://drive.google.com/file/d/1nS4_rGBahpgJDFnLoLF8ntGqItzOoV6d/view?usp=sharing

https://drive.google.com/file/d/13dxOVRulTz_xeO6-iiBmEXG2vFdi3ntC/view?usp=sharing

On July 1, 2017, Proctor uploaded two runs to Strava where both files covered roughly the same route, with the first covering ~40 miles and the second ~50 miles. The first, shorter upload contains the lower speeds expected of an ultra runner with the vast majority of data points at less than 8 mph. In contrast, the second, longer upload contains a very large number of datapoints above 8 mph and above 13 mph:

https://drive.google.com/file/d/1rS_loh3QpCE2DzmCS15usN4bWNpDHUSg/view?usp=sharing

https://drive.google.com/file/d/1c7YMdk3zayqkz6hHARq5XrzXATGC_KdK/view?usp=sharing

Note that the speed histogram from the second, longer distanced upload from July 1, 2018 has a pattern similar to the ones seen on June 27th, 2018. I think it is reasonable to say that some of the data Proctor is uploading comes from a watch that is riding in the RV that accompanies him.

Whats even more suspicious is that creating the histogram of the stride lengths from the cadence and speed data of Proctor's Strava data shows that there are very high number of data points showing extremely long and implausible stride lengths of 15 feet and above for the all the days up to the time of this posting as indicated by the far right bars of data in the graphs below:

https://drive.google.com/file/d/1eRW1typQ0GIQRdd06syfiLflCcy5cESX/view?usp=sharing

Note that Proctor's Strava indicates that Proctor is using Garmin Fenix 5X watches.

https://www.strava.com/athletes/12200779

For reference, here is all the run data from Pete Kostelnick's and Sandra Vi's runs across America that I do not question and I believe are authentic. Note that both of them also used Garmin watches and that they do not exhibit the suspicious histograms that Proctor's data shows:

Pete Kostelnick's Run Across America Data:

https://drive.google.com/open?id=1r5Q13fGdlq2gzA-NpjlxcgVN3lM6CBCY

Sandra Vi's Run Across America Data:

https://drive.google.com/open?id=0B1F3plcU7fEmZHAyd2hCOUtBczg

I have also started to analyze Proctor's longest runs from the past year before this run across Canada, including his ATY race, and I am not seeing the suspicious histogram speeds and stride lengths I am seeing in his current run across Canada which further goes to support my theory about what we are seeing here. I have not yet uploaded this analysis of the run data before the run across Canada because there are so many files but I will do so when I'm done looking through them.

Please do not send Proctor this thread yet for questioning since I wish to gather more data to either further support what we are seeing here or to refute the suspicious data.

"Do not send Proctor this thread...."

Then don't f"""n publish it, yet, you dummy.

As much as I appreciate your diligence and work ethic seeking cheaters, you seem to lack common sense. What, you think the internet is between you and a couple geeks?

Wet Coast wrote:

"Do not send Proctor this thread...."

Then don't f"""n publish it, yet, you dummy.

As much as I appreciate your diligence and work ethic seeking cheaters, you seem to lack common sense. What, you think the internet is between you and a couple geeks?

I didn't really expect LRC posters to keep things quiet, just a few more days of quiet would be nice to gather some more data before tipping them off and asking for an explanation.

But really what would be needed is boots on the ground from a distance, anonymously capturing photos and videos to determine what is going on. Canada is a big country with a lot of empty space so it wouldn't really hurt to make others aware.

Scam W.,

Now I don't know the answer to this, but perhaps you do (or someone else reading this):

Is there a possibility that over an eight or 10 hour day a battery could die and he needs another watch, would that be an upload where you can connect the runs of one watch with another? And is there a chance, running in places like Columbia River Gorge or Rogers Pass, Kicking Horse Pass - between mountains etc - that he could lose connection with the satellite? If so, don't they automatically connect the last data point to the latest data point via a straight line?

I get that sort of thing on runs about town in my little jogs and cycles.....wearing a Garmin.

For a run captured partially on two watches, you could merge them with third party tools and if there is a distance gap, it could be a straight line between the end and start of the two captures. For losing satellite connection, that could also give you some straight lines between two points but it depends on whether the watch does interpolation or uses an accelerometer for backup.

But I'm not seeing these straight lines or gaps in Proctor's Strava data from his Garmin Fenix so it's a moot point anyways.

I found one, but I think he is legit and assume - as you have answered - that these straight lines can be from connecting two data points due to dead battery (so two watches) or loss of connection.

So he wears two Garmin devices to make sure that he has tracking happening at all times. It is a Guinness World Record attempt......perhaps the linking of two watch signals is where the straight lines happen.

Wet Coast wrote:

So he wears two Garmin devices to make sure that he has tracking happening at all times. It is a Guinness World Record attempt......perhaps the linking of two watch signals is where the straight lines happen.

I don't think Proctor is using one watch for part of the run and the second watch for another part on the same day since that should cause a Garmin Live Track session to end and another to be started with only part of the track. What looks to be happening on the Live Track is that a watched is paired to a cell phone for the Live Track, which is usually blue tooth. So you can either have the cell phone loose cell connection or the watch lose connection to the phone.

Where is Asher when you need him?

Copy and pasting from the other Proctor thread:

Having had a chance to sleep over my initial analysis from yesterday and taking another look at Proctor's Strava GPS, I have come to the conclusion that what we are seeing here is very likely NOT cheating and that one of Proctor's two Garmin Fenix watches is for some reason recording highly inaccurate GPS coordinates.

For example, in Proctor's two Strava uploads for July 1st from his two watches, they both start and end at the same points and the cadence data from both are very similar yet slightly different, which is a good sign this is genuine data and that Proctor was wearing both of them. But the key difference is that the second, longer distanced upload has the tracking swinging widely from side to side of the road, even going into neighborhoods.

I believe Proctor used this same malfunctioning watch on his first June 27th runs because they also exhibit frequent swings in speed.

Of note is that I do not see the tell tale frequent cadence oscillations from 0 to 60 rpm at speed that would indicate he is riding a bicycle or in a vehicle with his arms not moving. This was a reason why I was not fully sure what we were seeing in this data yesterday.

So I am happy to report that Proctor's run seems genuine from at least what I can analyze. I would like someone, maybe me if I have time, to reach out to Proctor and his teach to try to understand why one of his two Garmin watches is record such widely inaccurate GPS coordinates.

Check July 1st

Two David's are there.

On the stream, he bonked big time towards mile 32. He began stopping every three

miles. It's understandable; he's been going out fast. He goes on (this is near Salmon Arm) to

finish the run at mile 40, averaging 13 plus minutes per mile for the run. He goes onto Facebook and says he has to get sleep and rethink run, pace,

and route.

His Strava shows 53 miles and an average pace of 8:08 for the entire run. Miles after 42 are all in 6 minute per mile range.

How do I add pics on this forum?

file:///C:/Users/Cooper/Dropbox/Screenshots/Screenshot%202018-07-01%2017.22.39.png

dunno if this will work cause in dropbox

Ok, here's the Google Doc of the anomalies.

Most disturbing are July 1st and 4th.

https://docs.google.com/document/d/1MSESnwlrfW1GRwr06vs1fnrQFgpczwhbShx8d-0JRds/edit

I've read through your doc and the explanations from my last post would apply here. Namely, what we have is a case of poor Garmin Live Track software providing wonky updates and because one of Proctor's two Garmin Fenix watches is recording highly inaccurate GPS coordinates.

The Garmin Live Track updates are provided by one of Proctor's watches tethered to a cell phone. Live Track only updates every few minutes and does not show all the data. I would discard the numbers from Live Track in favor of using the full data it is based on, ie: the GPX downloads we can get from Strava.

As I previously mentioned, the two uploads that Proctor did on July 1st, 2018 shows that one of his two watches is recording highly inaccurate GPS coordinates. One watch records coordinates that mostly stays on the roads while the other watch records coordinates that has it zigzagging offroad into neighborhoods that uses paths not physically possible:

https://drive.google.com/open?id=1OujrRtXXhhHYmrZJwJKHYbUdStRY4wSI

https://drive.google.com/open?id=15bYrbFuRZ-VaW7vsv52gq15--pV_8ufQ

For the July 3rd, 2018 upload where there is a straight line in the GPS track from 2.5 hours to 3 hours, analyzing the GPS data and cadence shows that reasonable cadence was still being recorded during this time and that during this period it seems as if his watch lost total satellite connection for some reason producing the gap seen. When his watch obtains satellite connection again at 3 hours into the run, it was about 15498 feet straight line distance to the previous point where there was a connection. 15498 feet in half an hour gives a straight line speed of about 5.87 mph, which is entirely plausible since this is lower than the ~7 mph pace that Proctor can run:

https://drive.google.com/open?id=1G7XVmXCzjU8cI1NmoBfD3yvpvDQdFcBk

So in summary, we cannot rely on Garmin Live Track and we also cannot rely on Proctor's Strava data to give intermediate pacing numbers if it is from the bad watch instead of his good watch.

I am not a pro in matters regarding the technicalities of live-streaming or this kind of analysis. I was interested in following the run and thought the Garmin live tracking was supposed to represent his run. Thank you for reviewing it. I suppose it is really meaningless to follow the live-streaming and the Strava is the accurate source. I will also conclude that he is a mighty strange fellow to average just over six minutes a mile from mile 42 to 53 and to then stop way short of his daily goal. And, then to post this to his Facebook page:

"After a few days in the mountains ⛰ Dave knows he needs to listen to his body and dial back the pace. The team is reviewing the route. If schedule adjustments are needed we’ll post an update."

The live-streaming function doesn't give you any clue to the run, then. His 40 mile, 13:24 effort was actually a 53 mile 8:08 per mile job.

I think I'll wait till the run is over and see how it all goes.

Thanks again for reviewing it.

dbjared wrote:

The live-streaming function doesn't give you any clue to the run, then. His 40 mile, 13:24 effort was actually a 53 mile 8:08 per mile job.

I think I'll wait till the run is over and see how it all goes.

Thanks again for reviewing it.

What I would do instead of looking at the Garmin Live Tracking pacing number is to do what someone else did on a previous transcon thread. They would look at the current point of where Proctor is on the Garmin Live Track map by eye and find that same point on Google Maps or some mapping software to get the GPS coordinates. Then, after half an hour or so, find Proctor's new location with the same eye method and then you can calculate the distance traveled in Google Maps. That way, you can get accurate pacing information using the advantage of a large time range to minimize errors from Live Track.