I have reread Mal's explanation http://www.intellimass.com/RealSpeed/Alpha%20Racing.htm but I am still struggling to understand some of my results. I would be greatful if someone could explain it to me.
In the first alpha it all makes sense. Get back to within 50m of start point within 500m total distance.


The second one confuses me. Why does the line not end at the front of the proximity circle? Is it that by running to the back of the circle it adds a greater section at higher speed and therefore a better average?
My question then becomes which profile is better? A broader gybe or tightergybe with a longer run out?
Some more examples of tracks by the alpha gurus would be cool.

Looks like a length issue on the second one, though would be good for Mal or sailquik to comment!

You are correct Lao Shi. Any point within the proximity circle can be used to calculate a result. If your speed is accelerating after the turn then it is more likely that the last point in the circle will be chosen because it will produce a higher average Alpha.

Ahhh, now you get to the crux of the question grasshopper.

The beauty of the Alpha challenge is that there are many different ways to approach the best answer. They will depend on the local conditions, the wind strength, the type of board, fin and sail and whole lot of other things I have not thought of yet.

These things I can say though with a maybe a tiny little bit of confidence:
I suspect the optimum approach if you have the conditions for it is closer to a square reach each way.
The fastest gybe will give you the best Alpha. ie: where you lose the least speed and start accelerating soonest with the least amount of downwind loss consistent with the maintaining speed part. How to achieve that will vary a lot depending on all the things I mentioned first.
If you lose speed by pinching one way more than you have to to make the proximity circle it is most likely a disadvantage. But if you miss the proximity you don't even get a result!
Judging the best place to gybe is definitely a major skill in a fast Alpha.
Judging the best radius to gybe is also definitely a major skill in a fast Alpha.

Sorry guys, that is probably as much help as a cool drink in a snowstorm.

I've analysed a lot of alphas for the challenge, by a variety of people.

The fastest ones seem to have both legs dead parallel, with a lovely 50m diameter gybe in the middle. Whether this is across the wind I can't say.

Mine on the other hand look like your first one -- the gybe isn't as tight so the legs aren't parallel. Needless to say my PB is 23 knots, nothing like the fast guys.


I've often been surprised by which alpha comes up as the fastest on analysis though, frequently it's not the one I plane out of, it's the one with the fastest entry speed. I think that your speed across the wind is equally, if not more important than not losing speed through the gybe.

What also helps is a somthing to judge you gybe radius eg bank or beacon or race mark

Select to expand quote

vando said...

What also helps is a somthing to judge you gybe radius eg bank or beacon or race mark

i agree with Vando,it is important to know where you are in the gybe thats the killer at hardies where your sailing huindreds of meters off shore and with an on shore wind you cant see your tracks. a temp bouy system would be a good gauge?
i have noticed most of my best alphas have not been the fastest gybes on the day,allways seem to miss the proximity but then thats the beatuy of the challenge

Of course, I should have said that the fastest gybe is critical given equally fast speed on the reaches.

Here is my best Alpha 500 so far (optimism ) :

Note that I just hit the proximity on this one. It is a bit further downwind and upwind than most of my other good ones so the extra speed in this case may have more than compensated for the slightly upwind and slower return. Also note that if I had not gybed at this exact moment (pure luck) I may not have hit the entry point proximity which is a slightly upwind course. A bit of luck always helps.

One of my favourite feelings in windsurfing is carving into a gybe at 30+ knots on flat water. 35 knots in this particular case. Oh what a feeling!!!



Here is another Gybe from that same session where I clearly came back too tight but also fluffed the gybe and lost too much speed (thats my weaker tack ).

It definitely helps to have some reference marks like a buoy or even a couple of empty orange juice bottles anchored with bricks.

Analysing Alphas from the GT-11.

Unfortunately, it seems easy to get too may errors in GT-11 Alpha trackpoints and they can be a bit less obvious to spot. Thus far, RealSpeed does not have a specific filter to help flag these type of Alpha error but I know Mal is working on it.

The GT-31 was worse for these at first but we think the latest firmware with 'smoothing' off helps a lot. It is still wiser to use the Doppler values from the GT-31 though.

Here is a good example of a track with an invalid set of points. It is easy to spot when you have the Doppler speed track on in the speed graph. In a normal graph from the GT-11 the two lines are pretty much the same except that the trackpoint speed usually rises a little faster than the Doppler accelerating out of the turn. In this case it is clearly way off.




Here is the best run again with the Doppler speed shown. This is a normal track and an example of the type of relationship between the Doppler and Trackpoint speed that you should see.

Regarding alphas on the GTC site,

Can someone please write a short article on how to analyse your alphas, specifically with a picture of what a spiky alpha looks like (a big V) versus a real alpha. This is one of the two main problems we have with people's data on the site (the other one being garmin 2 second speeds).

Most of the article is already written written if we can steal some of sailquik's post above (hint hint )

Good stuff guys.
Keep the samples coming.

My 22.47 ave in the first example only had a max speed of 29 ish so entry speed must be a start. Also my min speeds are down on 12-13. I will keep trying.

I'm implementing the Alpha 500 check in my web site at the moment, and came up with this beauty in one of my test files. The purple track is out of RealSpeed, and is the top Alpha 500 in the file. I already sent it to Mal to have a look at, but in this track, the start point is over 60m from the end point, not under 50m.

My calculator wouldn't pick up this speed, but Realspeed does (on my computer.)

And while I'm at it, I'm seeing a lot of these sorts of Alphas in Realspeed now that I look. See the "Distance" column to the right, the distance for this Alpha is less than 210m.

My calculator doesn't kick in until distance reaches 500m. Realspeed seems to cap out under 500m (this is the shortest one, but see that the others are all under 500m too.)

This is with both GT-11 and GT-31 files using the latest version of Realspeed. I am a bit suss on my setup though, I am beginning to suspect that Realspeed may not be entirely happy running under a 64 bit OS.



Also, in response to the earlier query, here is a very suspect-looking Alpha 500, with a very sharp point, and a 14 second trackpoint.

Hey Dylan, an Alpha 500, is a maximum of 500 metres, not exactly 500m, so any distance under 500m is valid, (happens when tracks go wide at the end) it's at a disadvantage of having a greater percent of slow gybe speed in the equation so won't be as fast as one that goes the whole distance.

I took some the advice posted here today and got some pretty good alphas in just moderate winds. I gybed tight, and kept it square.

Select to expand quote

Dylan72 said...
Also, in response to the earlier query, here is a very suspect-looking Alpha 500, with a very sharp point, and a 14 second trackpoint.

I just happened to speak to Mal last week about this -> there are two problems here:
1. occasionally the GPS's "loose" a single data point (or more as in this case), but RS wasn't detecting them. **
2. there isn't an option in RS to have max value for the rate of angle change from one data point to the next point.

** in this example of 14secs, RS will detect it if you set the "Time threshold" option to its minimum value of 10.

Select to expand quote

decrepit said...

Hey Dylan, an Alpha 500, is a maximum of 500 metres, not exactly 500m, so any distance under 500m is valid, (happens when tracks go wide at the end) it's at a disadvantage of having a greater percent of slow gybe speed in the equation so won't be as fast as one that goes the whole distance.

Yes, after I posted this, I went back to the intellimass site and re-read the Alpha definition, and the lightbulb went on.

I see that now, and will work on it tonight. (Luckily I hadn't released my Alpha calculator yet!!)

I still think the first one I posted, where the start and end points are more than 50m apart, is weird though.

(Back to work...)

Select to expand quote

nebbian said...

Regarding alphas on the GTC site,

Can someone please write a short article on how to analyse your alphas, specifically with a picture of what a spiky alpha looks like (a big V) versus a real alpha. This is one of the two main problems we have with people's data on the site (the other one being garmin 2 second speeds).

Most of the article is already written written if we can steal some of sailquik's post above (hint hint )

Steal away Ben!
I don't know if I can add much to that though other than some more examples. I will have a think about it..... owww, that hurts already..........

Oh, I find the most helpful graph is the speed graph with speed and Doppler speed on in the 'graph fields' menu (Right click on the speed graph). Used in combination with the map view most issues are pretty easy to spot.

Another hint: Check the satellite numbers line in the speed graph as well. In the bad example above you will notice that the number of sats tracked dips just before the trackpoints go haywire. I finally figured out that it is quite easy for the GPS worn on the upper arm to get tilted down as I sheet into the gybe. This might explain the loss of sats for that short moment. I have had great clean tracks with the GPS's helmet mounted without the loss of sats issue.

Ahh cool, thanks Sailquik

The 'V's that I'm talking about are like this one:




Most often if a newbie posts a high alpha, it's because they don't understand that this is a spike.
As far as I can tell, even if you tick the "Strict (no invalids)" box, Realspeed still treats this as a valid alpha. Perhaps this is a bug in Realspeed?

Another easy way of detecting "false" alphas, (for Gt11s anyway) is to set up a doppler version as well.
If the highest doppler alpha is only slightly slower than the trackpoint alpha, and is the same gybe, then it should be valid, but if the 2 gybes are different, or if there is a marked speed difference then there is a problem.

Select to expand quote

decrepit said...

Another easy way of detecting "false" alphas, (for Gt11s anyway) is to set up a doppler version as well.
If the highest doppler alpha is only slightly slower than the trackpoint alpha, and is the same gybe, then it should be valid, but if the 2 gybes are different, or if there is a marked speed difference then there is a problem.

Spot on decrepit!

Select to expand quote

sailquik said...

decrepit said...

Another easy way of detecting "false" alphas, (for Gt11s anyway) is to set up a doppler version as well.
If the highest doppler alpha is only slightly slower than the trackpoint alpha, and is the same gybe, then it should be valid, but if the 2 gybes are different, or if there is a marked speed difference then there is a problem.

Spot on decrepit!

He is not just a pretty face, our Mike form the crab lickers squad

Select to expand quote

mineral1 said...
>>>

He is not just a pretty face, our Mike form the crab lickers squad

Why thank you kind sir, but please tell, where else do you think I'm pretty???