Following thread in average 5/10, I think I open this new thread.
Reading decrepit's text:
The rules say, a total distance not greater than 500m and a diversion at the start and finish points no greater than 50m. That's the way Mal conceived it (I think it's Mal's idea).
The GPS we use record data at 1 or 2 second intervals, so it's impossible to get exactly 500m, apart from the very odd time, 500m coincides with a measurement. That may have been Mal's reason for not saying exactly 500m
So no software is lenient on the plus side of 500m. I think I've seen an alpha as low as 70m on realspeed, that wasn't shown on GPSarPro, but I can't tell you what the minimum distances any software allows.
PLEASE FEEL FREE TO WRITE THIS:
1. What than is ideal alpha, what is ideal path, feel free to attach some drawings of perfect alpha.
2. What if that line is not tipical "U" ride then crossline ride ?
3. Anything about perfect alpha is welcome for me
4. Why is @500 "UP TO 500m" when I have in gps results also @250
(what is the difference between these 2 alphas?)
Also I have @1km, I suppose that is kind of long distance alpha, right? 500-500met
THANX IN ADVANCE
Hopefully Sailquik will answer this one as he has studied it indepthly & is Australia's best Alpha 500 record holder @27.9.......
1. Not sure there is a perfect alpha, think it depends what sort of gear your on, and what your skills are. I'm an ancient leightweight wave sailor, so I do much better on small gear I can turn hard and spend a minimum amount of time in the gybe. Others may do better on bigger gear that keeps on the plane easier and have a higher minimum speed, but spend a bit longer in the gybe with a bigger arc.
2. My fastest alphas aren't a perfect "U" the turn tightens up as I slow down, if you can complete the gybe in under 50m you can go in square and out slightly downwind. If the lines cross it means at least one leg is upwind, this will slow you down, you want the separation to be as close to the 50m maximum as possible.
4. cause @250 is up to 250m and of course the @1km is up to 1km. The shorter distance will be lower because it takes a while to get back up to speed. The 1km could be faster or slower depending on the wind, and water conditions.
This was one of the best alphas I've seen:
You will notice:
1) Good entry speed (35 knots in this case)
2) 25m radius corner, not too asymmetrical but just tightening up a bit as you slow down
3) Exit leg parallel to entry leg, ideally at 49 metres between them the whole length of the track
They are a real art form to get right, I'm still struggling with mine, but the only way to get better is to practice, analyse, and then practice some more 
Great illustrations Nebs, what was that "best alpha I've seen" you don't give a speed for it about 27/28kts I'd guess
Perfect alpha 500? It's pretty simple really.
1. Go bloody fast (over 35 knots?) square to the wind for about 250 meters.
2. Chuck a killer gybe losing as little speed as possible with a little less than a 25m radius.
3. Accelerate back up to warp speed instantly on the other tack parallel to the first, and
4. Hit the 50m proximity circle with a few cm to spare at exactly 499.5m total distance.
Jacques are the closest to this I have seen so far. 
I'm still working on it. 
1. Answered already
2. Fine. The name "Alpha" comes from the Greek letter, which looks like a crossed track. (I tried to paste one here, but in this font it just looks like an 'a')
3. See other answers
4. As I discovered when coding for Alpha 500, if you don't specify a minimum travel distance, then you end up with really fast alphas that are actually just the first bit of the gybe, and are really short (<50m.) To deal with this, ka72.com insists that the track length needs to be more than twice the radius figure. Which means an Alpha500 score on ka72.com will always be measured over at least 100m of sailing (and not more than 500m.)
Different software has different parameters. As you noticed, RealSpeed seems to have a lower tolerance.
In addition, I believe RealSpeed has some extra smarts that other packages don't have. If your return track crosses the 50m radius, but you don't actually record any points inside the radius (your return track just touches the radius) then Realspeed will pick it up. KA72.com doesn't.
These are pretty marginal cases that don't happen often, but perhaps someone has an example to share?
