02 Nov 13
I realise this is an "old chesnut" but consider the following.
I am playing someone whose current game load is 1400+
His 1 year rating is above 1700 so he is better than me by some distance.
However because of his massive game load he loses a lot of games by timeout.
His current rating is 834!!
If I lose my rating will plummet because of this.
That is why the rating system cannot be relied on and I personally take no notice of my rating.
I've heard all the arguments before so i don't expect any change
Rant over.
02 Nov 13
1 edit
Originally posted by vendaI cannot think of even one good reason for a person to falsely lower their rating. Maybe to enter and win tournaments that are intended for lower-rated players. Does that really make them feel like they accomplished something?! Seems pathetic but then, some folks are. Even the most mentally challenged person is eventually going to get tired of it, I would think.
I realise this is an "old chesnut" but consider the following.
I am playing someone whose current game load is 1400+
His 1 year rating is above 1700 so he is better than me by some distance.
However because of his massive game load he loses a lot of games by timeout.
His current rating is 834!!
If I lose my rating will plummet because of this.
That is wh ...[text shortened]... tice of my rating.
I've heard all the arguments before so i don't expect any change
Rant over.
A person's average rating is perhaps a better indication of their true skill.
Would it be better to use that rating for any game or competition that has rating limits specified?
The average and "tournament entry" ratings are effected by this too, but it would require dumping many more games in order to make them drop, especially as the total number of completed games continues to increase.
02 Nov 13
Originally posted by SwissGambitI remember the rating floors suggestion from a while back and thinking it had some merit.
Like I always say: rating floors. They would avoid this problem.
I also remember a suggestion where ratings were calculated on the rating level of each player when the game started and not when it finished which also seemed like a good idea.
I can't remember the arguments against but I know there were some and I suppose there would have to be a lot of protests before the system was changed.
- Joined
- 07 Jun 05
- Moves
- 5301
03 Nov 13
Originally posted by vendaI had a look at the tournament in question, your graph and that of your opponent.
I realise this is an "old chesnut" but consider the following.
I am playing someone whose current game load is 1400+
His 1 year rating is above 1700 so he is better than me by some distance.
However because of his massive game load he loses a lot of games by timeout.
His current rating is 834!!
If I lose my rating will plummet because of this.
That is wh ...[text shortened]... tice of my rating.
I've heard all the arguments before so i don't expect any change
Rant over.
On the tournament: he has already lost 4 games on timeout, so if you keep playing, there is a chance that you get points for nothing, just by taking the skull.
Your own one year rating is 1655, so not that far from his 1713.
On average rating, you are at 1569, whereas he is at 1294. If you actually lose to a player whose average rating is so much lower than yours, you'd be be better off thanking him for helping get your own rating back in line!
What is particularly interesting, given that you mention 1700, is that your opponent's rating peaks at 1713, and then falls back again, repeated several times with slightly lower numbers. My interpretation is that he is not actually rated at 1700 at all, but is storing up wins and postponing losses to boost his peak rating.
Rating floors. Why not rating ceilings for a change? Your average rating for the past 100 games is too low, so you cannot increase above 200 points more than it. It would stop people trying to manipulate the games timing to give unreasonable peaks.
I quite like Coletti's idea:
"What if rated games had to be played with someone within 400 points of your own rating? All other games would be automatic non-rated".
None of the rating inflation caused by floors, and all of the benefits.
In short:
1) his "skill" rating is hardly 1700, but more likely to be lower than yours.
2) His graph yoyo's all over the place within a time period of a few weeks - chose your time to resign if you think you are losing, and you get to chose his rating.
3) If your opponent manages to beat you, thank him for finding the flaw in your playing, and for getting your rating back into line.
4) Like I always say (@Swiss ;-) ) : Rating floors cause rating inflation, and should be avoided at all costs.
03 Nov 13
Originally posted by gezzaYou've done a lot of research there my friend and I thank you for your interest.
I had a look at the tournament in question, your graph and that of your opponent.
On the tournament: he has already lost 4 games on timeout, so if you keep playing, there is a chance that you get points for nothing, just by taking the skull.
Your own one year rating is 1655, so not that far from his 1713.
On average rating, you are at 1569, whereas he i ...[text shortened]... s say (@Swiss ;-) ) : Rating floors cause rating inflation, and should be avoided at all costs.
However, I don't agree with all your analysis.
His graph is predominently a downward slope.
I attribute this to his frequent time out losses because of the number of games he's playing.
If you click on a couple of realatively recent peaks on his graph he has beaten players of rating 1603 and 1662 so he's not a poor player.
I don't understand why you think my rating is out of line.
My graph is relatively stable.
I still believe the system would be better if the rating gain or loss was based on the rating at the start of the game & not the end.
- Joined
- 07 Jun 05
- Moves
- 5301
03 Nov 13
Originally posted by vendaHiya.
You've done a lot of research there my friend and I thank you for your interest.
However, I don't agree with all your analysis.
His graph is predominently a downward slope.
I attribute this to his frequent time out losses because of the number of games he's playing.
If you click on a couple of relatively recent peaks on his graph he has beaten players of r ...[text shortened]... etter if the rating gain or loss was based on the rating at the start of the game & not the end.
I'm always good for discussions about rating; we all have our weaknesses.
I do try and do my research, so that at least I appear to know what I am talking about.
There is a general downward slope from 23rd Oct, but if you were to have looked from 4th Sept until 4th Oct, then the picture is different. I guess I am saying that your opponent may well have a very different score in two weeks time. I'm also not finding his wins against 1662 and 1603 (Game 10020744 was a loss, as was Game 10198137).
I really don't think your rating is out of line. As you say, your graph has been stable for months. I tried to say that you should thank your opponent if you lose, as he has found a flaw in your play, which you can improve on.
Based <b>purely</b> on average rating (I will not comment on games in progress), the higher rated player (you!) <b>should</b> win. If you do not, then your rating should drop a little - that is what I intended to say. I know that against an 800 player there is a large drop, but wait 2 weeks and his rating will have yoyo'ed back up again, and the drop will be less.
Basing on rating at the start of a is subject to similar problems. A returning player will have a rating that rises, but his opponents will lose points based on the low rating when the match started (even though he is now rated higher).
I guess that I am trying to say that the rating system is not perfect, but some suggestions could break it for everyone (hyper-rating inflation).
Cheers, Gezza
04 Nov 13
Originally posted by gezzaYou always say it, but never show why, inflation is a bad thing. I expect nothing will change this time around. 😴
I had a look at the tournament in question, your graph and that of your opponent.
On the tournament: he has already lost 4 games on timeout, so if you keep playing, there is a chance that you get points for nothing, just by taking the skull.
Your own one year rating is 1655, so not that far from his 1713.
On average rating, you are at 1569, whereas he i ...[text shortened]... s say (@Swiss ;-) ) : Rating floors cause rating inflation, and should be avoided at all costs.
- Joined
- 07 Jun 05
- Moves
- 5301
06 Nov 13
Originally posted by SwissGambitWhy my dear Swiss, I must admit I don't recall you ever asking why inflation is bad.
You always say it, but never show why, inflation is a bad thing. I expect nothing will change this time around. 😴
I recall arguing whether it occurs, and also the extent, but never whether it is a bad thing or not.
06 Nov 13
Originally posted by gezzaWell, then perhaps your memory is going. 😞
Why my dear Swiss, I must admit I don't recall you ever asking [b]why inflation is bad.
I recall arguing whether it occurs, and also the extent, but never whether it is a bad thing or not.[/b]
I generally concede that inflation occurs, and that floors cause inflation. I'd be ignorant of simple math if I didn't. But in a rating pool, the primary thing that matters is the relationship to the rest of the pool. The number means nothing by itself. Thus, we can live with some inflation.