What about using the following for a tourny rating formula:
Tourny rating given by the larger of
(1) upper decile of the last 30 games and
(2) upper decile of the last 300 games (or all games if < 300 played)
The upper decile is calculated by sorting the selected values into an incrementing sequence and taking the rating 10% from the top.
(The median works the same way but is 50% from the top, an upper quartile at 75% etc)
That would be very hard to screw around with in regard to banded tournaments and would also avoid players being punished by localised peaks in their ratings.
Originally posted by CrowleyI can indeed but to save on typing lists of 300 values, I'll simplify a bit!
I've been moaning about this for a long time now.
Sounds good, but I'm no mathematician, can you give a better explanation or maybe an example?
A practical example:
Last 30 games, ratings are 1500, 1490, 1480 … 1210 (decreasing 10 pts a game)
The 270 games before that: 1800 (50 times); 1850 (100 times); 1900 (70 times); 1900 to 1500 (a slide over 50 games).
Step 1) The upper decile of last 30: 10% of 30 is 3, so were looking for the third top value.
In order, we have 1210, 1220 ... 1470, 1480, 1490, 1500. The third top value is thus 1480.
Step 2) The upper decile of last 300: 10% of 300 is 30, so were looking for the value 30th from the top. As we have 71 instances of 1900, then the value 30 from the top is 1900.
The biggest from these two is the player’s tourney rating: i.e. 1900.
Under the current scheme, it would be the max of the last 30 which was 1500, a value well below this player's true strength.
On the other side of the problem (over-inflated rating), a solid 1200 player who gets + 40 and +40 on two games by T/O when playing a big rating-slide player ends up at 1280 under the current system. Using the above, there tourny rating would still be 1200 as you need 3 extreme values in step 1 before the tourney rating shifts. Thus players are not penalised from lower-banded tournys when a couple of 'lucky' wins happen.
Ok, I think I understand.
Isn't that a bit simple? Doesn't your system only allow for downward spikes?
What if this player had a big upwards spike in there?
Say of the 300 there were the ranges:
1200s 100
1300s 100
1400s 50
1500s 25
1600s 20
1700s 5
With the player slipping back into the 1300-1400 range with his (let's assume the player is male 😉) last 100 games.
This means his tournament rating would be in 1700s while he should realistically have a TR in the 1400 (maybe 1500) range.
Originally posted by CrowleyTrue: but how did they get to 1700 befoer the slide? Did they suddenly loose the ability to play chess? The main movements in ratings are
Ok, I think I understand.
Isn't that a bit simple? Doesn't your system only allow for downward spikes?
What if this player had a big upwards spike in there?
Say of the 300 there were the ranges:
1200s 100
1300s 100
1400s 50
1500s 25
1600s 20
1700s 5
With the player slipping back into the 1300-1400 range with his (let's assume the player is m ...[text shortened]... rating would be in 1700s while he should realistically have a TR in the 1400 (maybe 1500) range.
i) new player going to a new rating
ii) rating slide after a series of T/O
iii) slow increase as people get better.
A short spike in increased rasting would have to last at least 30 games for the long-term step 2 to care, so a feak series of winds a few months ago wouldn't stick around for a long time (thus the use of upper decile rather than just the maximum).
A significant lowering of rating would likely be caused by (ii), or maybe the choice to play lots of games too fast. That means they can still play at their high rating but are currently choosing not to, so the high number still makes sense.
A general increase in rating (i) or (iii) is covered by step 1 of the calculation and so isn't a worry (much like the current system).
So yes, it is simple but due to the nature of ratings movements, should be safe.
You said:
Tourny rating given by the larger of
(1) upper decile of the last 30 games and
(2) upper decile of the last 300 games (or all games if < 300 played)
That means a upwards rating spike would be the largest number in your proposed system until 300 games after the highest point in the spike?
It is possible for someone to jump 200 points in a month and then slide back - I've done it...
Look, don't get me wrong, I like that people are thinking about this (I've been moaning about it for a while, as I said), I just think this is too static.
I just think something like a weighted average rating would be a better value to use.
I would like to see a system where your last 100 games or so are weighted more than the ones before it and then a normal average is calculated.
Like I said before, I'm not a mathematician or statistician, but it will be more meaningful than the last 30 days value.