For you to get a PB is an awesome accomplishment as you have been pushing the limits for so long !!!

What algorithm does Dylan's KA72 site use? It is disconcerting when the software that us noobs rely on has glitches... 

Nice Avatar picture!! That NP sail really stands out!

Well done Kean. Congrats on the PB.

Thanks for the support guys. It took me a few days to suss the jibe ARoona, I knew it was good for some high alphas but it was a case of thinking through the dynamics to get it to work. You need to enter through some nasty chop and standing waves at high speed off the wind to get the initial stage set up so you can then head upwind at speed in the butter smooth water.

With swells coming through you have to time one to jibe on so you can come around tighter to set up the exit. Once back to speed and having completed the 50m seperation it's tricky to get a nice bearaway on a gust to complete the sequence. As the area is in close to the shore and some cliffs the wind swirls around and isn't constant, a bit more westerly in the direction is the key to it filling in so you can go in further on the butter.

If GPSResults gave a lower alpha for a shorter distance, it's most likely because at least one of the points in the alpha run exceeded the filter limits, or because a data point was missing. It's not uncommon to see time points with 0.4 second distance to the previous point instead of the expected 0.2 seconds in GW60 data. GPSResults will not calculate short-distance speeds (including alphas) through such points, or through points that were filtered out. This should be easy to see in GPSResults; all invalid data points are shown in brackets.

GPSAction Replay uses a different philosophy; it will calculate alphas and short-distance speeds even if data points are missing. It also does not use acceleration of SDoP filters. Additionally, GPS Action Replay also allows you to edit the distance settings for alphas, so you can compute alphas with (e.g.) 70 meter distance. 

Barton, I'm not sure why you mention ka72.com. It's different again from both GPSResults and GPSAction Replay, and Roo's last upload to ka72.com is from 2016. In at least one GPS trace with missing data points , I have seen ka72.com agree with GPSResults about the lower numbers, while GPSAR gave higher numbers.

I have had discussions with Dylan, Yann (GPSAR), and Manfred about some of these issues. They all will fix any obvious bugs if you can send them test data. However, they also have different ideas for some issues, e.g. how to treat missing data points. Manfred tends to be most strict. I think for 5 Hz data, stopping calculations because one data point is missing is too strict.

Boro and Denis recently had some issues with artifacts. Both Dylan and Yann are working on adding SDoP filters which would identify the worst of these (GPSResults already has SDoP filters). Identifying artifacts from "medium bad" data as Boro had would require changes that seem unlikely to be implemented anytime soon, though.

In view of the large alpha differences in 4 of the top 5 alphas, though, it seems that GPSResults viewed large parts of the tracks as inadequate. Either that, or the "Proximity limit" in GPS Action Replay was larger than 50.

Peter, my posting about KA72.com was misleading/confusing. I sometimes analyze my results using GPSResults but rely on KA72 for my normal posting. I didn't mean imply that Dylan's algorithm had a glitch or anomaly and was actually hoping to learn that it worked correctly even when GPSSResults and GPSAR gave differing results. That said, for my result numbers a wet thumb in the air provides adaquate information  .

Barton, all software has bugs . I have seen bugs in all three programs. The authors usually correct them as quickly as possible - sometimes within a day, sometimes it takes a lot longer. But besides that, there will be some differences in the results that remain because of differences in opinion. 

It may seem a bit strange that I look into issues like the ones Boro and Denis had in such detail. They probably affect only one or two percent of all sessions, if that, and most sessions don't matter for the rankings. But with 40 teams posting x 2 sessions x 6 categories, we have about 500 sessions per month that determine the rankings. Even a 1% rate of false results would mean that up to 5 of 40 rankings are wrong. With Boro and Denis actually delivering results in the top 10 in several categories, a higer level of diligence is appropriate.

Roo, congratulations on PB, great alpha. Is that a "Triple Crown" or you have to get all three numbers (40/30/25) in the same year?

Thanks Denis, triple crown in the same year is the goal. I'm sure you and Boris can do it with your eyes closed! I want a nm of 30 knots with a 25 knot alpha in it, my 28.11 knot one included the 25 knot alpha so only 1.89 knots to go! Even better if I can get the 40 knot run as part of it, then it would be a "Golden" triple crown.