2578Re: [Croquet] Peaks and averages
- Aug 11, 2014Chris,You've now heard from an Aussie that knew about the new method :-) Best memory is I learnt of the this method a few months ago. I cannot recall how I found out, maybe in reading for my role on ACA AC Selection Committee. Are you aware it applies to AC & GC world events? At time of reading I thought it was good idea but you & others have pointed out its flaws.I agree with the suggestions of others that if maximum grade is used then it should be max grade at end of tournament. Max grade at any time in a tournament particularly favours seeded block play. However would we see more instances of withdrawals because a player has reached what they think will be a qualifying grade?The reason I like max grade in past 12 months is it stops grade boosting & protecting leading to a cut off date. I think that is a problem that needs a solution.TimOn Tue, Aug 12, 2014 at 12:52 PM, chris clarke <chrisd4clarke@...> wrote:Hi Samir,These are good points. I had been thinking along the lines of 3 (I think 1 and 2 are poor). However, given your point about finding a load of weak oppos, (which I also made as a weakness of the Peak Grade GC system), I believe that it would also be poor.Option 5 is interesting, but probably too complex, so I go back to using the current grade. James' point about if you're going to use peak grade, at least use end of tournament peak is also sensible since it rewards winning.One of the things that I think is great about the Notts list is that you tend to get far more useful feedback in 24 hours than in 2 months on the WCF.As you say, the fact that no-one in NZ seems to have been aware of the change isn't great. I haven't heard from anyone in Aus who knew about it either.Chris
Date: Mon, 11 Aug 2014 12:48:35 +0000
Subject: Re: [Croquet] Peaks and averages
I agree that an average would be better, but in practice is not easy to define.
Let’s say Mark only plays two days in 2015. He start the year (on 1 Jan) with a grade of 2300. He plays 10 games on 1 April, and ends with a grade of 2400. Another 10 games on 1 July, but his grade falls to 2375 where it remains after he plays no more games in 2015. There is a cut-off for selection on 31 December.
What’s Mark’s average grade?
1. 2338, since that’s the average of his opening and closing grade
2. 2350, since that’s the average of his highest and lowest grade
3. 2370, since that’s the average of his end-of-each game grade?
4. 2362, since that’s his time-weighted average grade (3 months (Jan-Mar) at 2300, 3 months (Apr-Jun) at 2400 and 3 months (Jul-Sep) at 2375 and another 3 months (Oct – Dec) at 2375)?
5. 2358, since that’s his time-weighted average grade during the active part of the season.
If you like 3, does it worry you that if, having reached a nice high grade on 1 April, Mark could find a selection of 1400 graded-players (or perhaps some helpful 12 handicap at his club to play 10 times) and played them on 2 April? His end-of-each-game average would increase to 2380? (note that 4, 5 would not be affected by this).
Now factor in the problems caused by rankings databases not having actual dates for many games, which makes a time-weighted-average-grade approximate at best. The playing season difference will make a difference because most players will stick to a single grade for 4-6 months during their off-season, which makes their end-of-season grade more important than their in-season grade.
The “highest over 12 months” is at least easy to calculate, harder to manipulate and entirely objective (and hence easy to check that the WCF have got it right – something that was a bit of a problem with the previous system).
The CNZ summary that you posted does look wrong.
croquet mailing list. For info or to unsubscribe visit http://nottingham-lists.org.uk/mailman/listinfo/croquet
- << Previous post in topic Next post in topic >>