Handicap racing now pointless?

Originally posted by andy101 so setting the handicap for the top of the fleet (if it does have the effect) is not really a bad thing imo

Well, this is the sort of thing one can argue about endlessly, and there are no right answers.

My personal opinion, which is no more likely to be "right" then anyone elses, is for club sailing we should handicap for the majority. For a big event like the Sailjuice series I can see why they would want to handicap for the potential winners, but my concern is that statistically that's a hard target, especially for classes where there isn't that much data. My limited statistical ability isn't up to it, but most I'm sure they are better at the analysis than I am.

Originally posted by rb_stretch I thought if the very best are just ahead of the average that suggests lower deviation, not higher.

Normally, if you'll excuse a statistics pun, yes.
In this case the distribution is highly skewed, and the best don't finish that far ahead of the mass, but the "still room for improvements" finish a long way behind the mass.

Originally posted by Neptune My confusion was what happened to the other 25%, as those numbers add up to 75%.

They don't need to in this case: I'm just using top 25%, middle 25% etc to define, well, really, points on the graph. I should have said quarters really I guess.


Edited by JimC - 14 Jan 12 at 12:07pm

Originally posted by JimC Even more complicated than that! Our Laser fleet, compared to our RS200s and Solos, has a greater deviation, but also much more skew in the deviation curve. Effectively that means that very best Lasers (and some of ours are top ten in masters worlds standard) are less far ahead of the average sailor than the best Solos and 200s, but also that the those of "still learning" group who have the greatest potential for improvement are rather further behind the average than in the other classes. If (and that's a very big IF) that picture is true of Lasers nationally, it means if you set a handicap for the best 25% it will be fair for them, generous for the middle 25% and harsh on the lowest 25%. If you set a handicap for the middle 25% it will be harsh on the top 25% and very harsh on the lowest 25%. If you set a handicap for the lowest 25% then it would be generous for the leading 25% and *very* generous for the middle 25%.

Probably showing my poor understanding of statistics here, but I can't quite make sense of this.

I thought if the very best are just ahead of the average that suggests lower deviation, not higher. This then fits with your suggestion that if you set the PY for the best, then you are being generous with the average. And the average is really what PY is about in my books, cause the best will be class racing.

Originally posted by rb_stretch From what I remember of JimC's graphs the Laser has a much narrower deviation than other classes, so setting PY for the winners means the whole laser fleet moves up the rankings.

Even more complicated than that! Our Laser fleet, compared to our RS200s and Solos, has a greater deviation, but also much more skew in the deviation curve. Effectively that means that very best Lasers (and some of ours are top ten in masters worlds standard) are less far ahead of the average sailor than the best Solos and 200s, but also that the those of "still learning" group who have the greatest potential for improvement are rather further behind the average than in the other classes.
If (and that's a very big IF) that picture is true of Lasers nationally, it means if you set a handicap for the best 25% it will be fair for them, generous for the middle 25% and harsh on the lowest 25%. If you set a handicap for the middle 25% it will be harsh on the top 25% and very harsh on the lowest 25%. If you set a handicap for the lowest 25% then it would be generous for the leading 25% and *very* generous for the middle 25%.

quote Marke:

 

Exactly, On one race like the BM (or whatever) the PY system isn't going to work very well, it never will. IMO the kerfuffle, hyperbole and aspersions cast in this thread have been way ott,  given the situation I think some "have protested too much" .

Edited by GK.LaserII - 13 Jan 12 at 11:14am

MaxB
Yep surprised me as well - but the numbers don't lie (or perhaps just a little bit !).  I ran 2 years of our club races using an implementation of the YR2 algorithm - and then applied a CSF normalising adjustment remarkably similar to the scheme Jim was alluding too with his graphs.  We had the same objective as the SJ - the best sailors should win when the wind/course conditions are suitable for their boat - we expect the wind/course variation to even out over a season.  For the fleets where we have enough data for the results to be significant - there was a remarkable match with the SJ; mostly within +/5 and all with +/-10.

I think there may be a couple of explanations:
  -  we try to race mostly at the top of the tide so often the tidal effects are small for most of the race
  -  being an estuary the slower boats aren't always hampered as you might expect by the tide as they can short tack and gybe up a shore while the faster boats are out in the tide.  Quite often an RS200 can keep with the B14s/RS800s upwind in light weather by short tacking.  Over a long series I think the swings and roundabouts even themselves out - and we still like everywhere else have the effect of every boat having its day.

The Sailing Committee decided to use the SJ numbers for the Steamer for consistency with the other big handicap events - and to be honest we didn't think the RYA numbers were working to give an equal chance to good sailors over the 20+ years it has been held.  I should say that I had nothing to do with the work the SJ folks did, but I heard about the Great lakes activity and they were kind enough to give me an insight into the data they were using. 

We have also adjusted our numbers for local club racing.  We haven't used the SJ numbers directly for every class - the PY sub-committee (yes we now have one!) decided on a scheme that uses a number halfway between our calculated numbers and the SJ ones, with some further alterations where either our data or the SJ data was not sufficient.  Like the SJ its early days yet - only three races run under the new scheme - but like the SJ it looks promising with a far wider range of classes appearing in the top third of the fleet.  A laser even won a trophy event - albeit in conditions that suited the laser - but I can't remember that happening before in 10 years and we have some good laser sailors.  The plan is to try and review our adjustments every 6 months.  I am currently trying to modify our race management software so that it can switch between RYA, local and personal handicaps as required.

We also provide the club members with an online tool to estimate their personal or class PY using a database of results that is updated automatically as each race finishes.  I know that some crews review their average 'achieved' PY after each series to see what the trend is - for instance I can see that I sailed significantly better across the Spring and Summer series than the Autumn series - there must be a temperature effect I haven't taken into account :-).  A few more tweaks to make and the system will also automatically upload each series when it completes to the new PY 2012 system.

regards

Mark


Edited by marke - 13 Jan 12 at 1:02am