I ran into an interesting difference between how alphas are calculate in GPSResults and GPS Action Replay. It appears that GPS Action replay always takes a run of 500 meters (or the closest it can get), while GPSResults will also consider runs - even much shorter runs, such as 100 m.

Most days, that difference does not matter, since the full 500 m runs will be fastest, and therefore selected by both programs. But just 3 days ago, I had GPS Action Replay give me a top alpha of 7 knots, while GPSResults gave more than 20 knots (details at boardsurfr.blogspot.com/2016/12/never-trust-software.html).

Looking at the GPS-speedsurfing.com and GPSTC web sites (respectively Hardie's explanation of alpha on this forum from 2007), it seems that the two sites use different definitions. GPSTC uses " total distance covered that is less than or equal to the maximum distance", while GPS3 uses "one run of 500 meters" (without the "less than"). Apparently, GPSResults implements the first definition, while GPSAR implements the second.

Since both GPSTC and GPS3 allow data analysis with either program, it seems that the difference in definition was not intentional (and perhaps even not noticed). It would be nice to hear from the respective "authorities" what the rule should be...

Alpha definition is always 'up to and including' X Meters. It is supposed to be the same in both GPS-TC and GPS-SS.

Shorter is usually slower, but sometimes it is all you get.

The same definition was used by Yann and Manfred in consultation with Mal Wright who invented the category. I think any differences you see will be in quirks of the way the program implements it.

KA-72 also used the same definition of course.

The original idea was that there would be different categories of 250m, 500m and 1000m (or more). But as it has turned out, the 500m has become the most used, possibly because that was the one originally chosen to highlight by GPS-SS and used in GPS-TC.

Here is a web archive link to the original pages with some examples but I can't get the Alpha definition page to appear:

web.archive.org/web/20140125053444/http://intellimass.com/TrackData/GPS%20Alpha%20Racing%20Top%2010.htm

Select to expand quote

boardsurfr said..
I ran into an interesting difference between how alphas are calculate in GPSResults and GPS Action Replay. It appears that GPS Action replay always takes a run of 500 meters (or the closest it can get), while GPSResults will also consider runs - even much shorter runs, such as 100 m.

Most days, that difference does not matter, since the full 500 m runs will be fastest, and therefore selected by both programs. But just 3 days ago, I had GPS Action Replay give me a top alpha of 7 knots, while GPSResults gave more than 20 knots (details at boardsurfr.blogspot.com/2016/12/never-trust-software.html).

Looking at the GPS-speedsurfing.com and GPSTC web sites (respectively Hardie's explanation of alpha on this forum from 2007), it seems that the two sites use different definitions. GPSTC uses " total distance covered that is less than or equal to the maximum distance", while GPS3 uses "one run of 500 meters" (without the "less than"). Apparently, GPSResults implements the first definition, while GPSAR implements the second.

Since both GPSTC and GPS3 allow data analysis with either program, it seems that the difference in definition was not intentional (and perhaps even not noticed). It would be nice to hear from the respective "authorities" what the rule should be...

and then if you use GPSAR 'jibes analysis' in the 'Speed' section, set distance at 250 metres (each leg) and scroll down 'width' column to find the 50m gybes, this analysis uses something different again as the speeds are quite different to posted (usually lower).

Select to expand quote

Te Hau said..

and then if you use GPSAR 'jibes analysis' in the 'Speed' section, set distance at 250 metres (each leg) and scroll down 'width' column to find the 50m gybes, this analysis uses something different again as the speeds are quite different to posted (usually lower).

Well, Jibe analysis is different from alpha analysis, so I would expect different results (jibe analysis makes both legs 250 m long, while alpha can move the end points around by up to 50 m if one leg is faster, and the legs are closer together).

But the alpha analysis in GPSAR tries to stick very close to 500 m at all times, even if it gives slower results. Here are two more examples - the first one from a 5 Hz
GW60 file:

For the 3rd alpha, GPSAR uses 496 meters, but GPSResults uses only 470.2 meters. For the 5th one, GPSResults uses a 9.2 m shorter distance. The speed differences in this case are small. Here's an example from a GT-31 file:


In this case, the length difference for #4 is large (428.5 vs. 497.8 m). The shorter distance that GPSResults uses gives a speed that 0.66 knots higher.

These were the first two files I grabbed to check, after seeing Andrew's question if this may be specific to the file I had first noticed it on. That does not seem to be the case.

At those great flat water spots that you guys seem to have a lot of in Australia, such differences will probably be uncommon, since longer runs will usually be faster. But at spots with short flat regions where you slow down again 150 m after the jibe due to chop, you may get better alpha results with GPSResults.

Definitely boardsurfr, GPSAR, not doing short distances can rob you of a good alpha, the worst case is where you go too wide just before the GPSAR min distance.

On KA72 I think the rule was that the length of the Alpha segment has to be between 100m and 500m. I have a feeling it may have changed slightly with the introduction of a wider/higher resolution range of devices.

When I originally developed the site software, I foolishly assumed that the alpha was a 500m segment, but the results I got with that approach were entirely inconsistent with the other systems on the market (namely GPSResults and Realspeed which were the systems I was benchmarking off at the time.) Quickly re-reading the definition cleared things up for me, but as far as I am aware there is no defined minimum length for the Alpha. As a rule of thumb, I used 2x the radius (2 x 50m, or 100m).

I've run other variants of Alpha for special competitions. An Alpha5 (an Alpha with a 500m segment limit, but 5m radius instead of 50m radius) has been a feature of the WindWanderers online speed competition for a few years.

Hello,

Boardsurfr warned me about that difference a few days ago, and I have just changed the computation of the Alphas in GpsarPro 5.19 (available for download) so that it relies on the same rules as others.

I personally preferred the previous one of GpsrPro, which provided Alpha 500 of length 500 (minus at most one GPS point), Alpha 250 of length 250, etc., rather than having some Alpha 500 shared with some Alpha 250, or even some Alpha 500 of length 100. Some people also asked me recently to have an Alpha 10000 so that we get 5000 + jibe + 5000, which could not be consistant with these rules. So I will see if I make a special category for that in a next version of the software (built on the "old" version of GpsarPro Alphas), which would be not named Alpha to avoid confusions.

Regards,
Yann

The story continues.
In addition to the Alphas, the Betas. I don't know how to include images in messages, so this is the message from my forum :

gpsactionreplay.free.fr//forum/viewtopic.php?f=9&t=449

Yann

Big thanks to Yann, who has worked a lot on GPSResults. Alpha results are now the same in GPSAR Pro, and he also fixed other bugs. In addition, he added some new features. The "Trackpoints table" in GPS Action replay now also shows SDOP number (+- column); selecting a top speed or region in a track now also selects the same region in the track points table, similar to GPSResults (if the "Auto select" checkbox in the speed results is checked).
That particular feature may be mostly of interest to geeks like me, but the new "Open SBP with auto-split" is a really useful function for all GW52 and GW60 users who recorded more than one session before downloading the data. Downloads as usual at gpsactionreplay.free.fr/index.php?menu=2

Select to expand quote

boardsurfr said..
Big thanks to Yann, who has worked a lot on GPSResults. Alpha results are now the same in GPSAR Pro, and he also fixed other bugs. In addition, he added some new features. The "Trackpoints table" in GPS Action replay now also shows SDOP number (+- column); selecting a top speed or region in a track now also selects the same region in the track points table, similar to GPSResults (if the "Auto select" checkbox in the speed results is checked).
That particular feature may be mostly of interest to geeks like me, but the new "Open SBP with auto-split" is a really useful function for all GW52 and GW60 users who recorded more than one session before downloading the data. Downloads as usual at gpsactionreplay.free.fr/index.php?menu=2

Many thanks to Yann. An already excellent program getting even better.

Have to agree this is a great version, it's only with GPSARPro you can do this.



A speed overlay of two different tracks.
Here we have the GW52 (red) worn on my head and the GW60 (yellow) on my wrist, I think the peak at gybe minimum on the yellow track, is the flipping action of my hand temporarily increasing speed.

The cross referencing between track points table and graphs is also very handy, as is the +/- values.
Nice work Yann

Thanks Yann for you excellent program and continuing work on GPSAR-Pro.

For those sailors who are interested in trying this new division idea, and who are users of RealSpeed, I would like to point out that it is easy to create this division in RealSpeed (and many others) like this:

Thanks for the comments.
I have just enhanced the software (version 5.22) so that it is compliant with recent UBX files.
Since tracks are now sometimes very large (at 10Hz or more), I have also enhanced the 2D display which is now really faster, even whith time zone selected in the speedgraph. It is an improvement also for people using several tracks at the same time for replays. So I suggest current users to (freely) upgrade to this version.

Select to expand quote

decrepit said..
Have to agree this is a great version, it's only with GPSARPro you can do this.
EDIT,
woops, just had an email from Manfred, my above statement is wrong!!!!
Sorry Manfred, GPSResults can also do comparisons of different tracks.