Rated players under 1000

[Paul Leblanc]

Re: Rated players under 1000

I certainly agree with the playing up aspect of John's argument. None of the BC organizers allow players to play up unless they are reasonably close to the section boundary.
However, the reason that we do this is not to prevent the very low rated player from taking rating points away from the higher rated player. It's because 90% of the time the game is no contest.

[Ed Seedhouse]

Re: Rated players under 1000

What? How is that even possible? Unless it's an arbitrary system where grade 2 kids get a 200 rating?

There is nothing in the Elo rating system that disallows a rating less than zero. If your rating goes below zero it follows that you also must have had at least one rating performance rating of less than zero.

[Mathieu Cloutier]

Re: Rated players under 1000

So are we talking about temporary ratings or permanent ratings?

I understand the temporary 200 rating after one tournament against a bunch of beginners. Sure.

But someone, playing, say, 50 games AND remaining at 200? I have a hard figuring how bad you have to play to achieve that. Most beginners I play, I would say they play at least 600 chess. A little bit of training in tactics and some openings and you get to 900-1000 in no time.

[Mathieu Cloutier]

Re: Rated players under 1000

Bonjour Mathieu,

The same thing happened, for intance, in the 2105 Ch. jeunesse du Québec. I found an old thread where you kindly explained me that there was no physical meaning to it, and then more people weighed in on advantages and disadvantages of different approaches to ratings.

Hi Vadim,

I was talking about 200 ratings, not negative PR. I just clarified my thoughts in the post above.

Cheers,

[Mathieu Cloutier]

Re: Rated players under 1000

One question;
How can increasing a 400 rated player to 1000 migrate them to stronger players? Don't let them play up until they go over say 1200. If they are good they'll move up faster. If they are bad players they stay in the U1200. I do not understand why you'd increase my rating just because you make the U1000's increase to 1000.
If a guy is only 985 why would you give them more than 15 points?

The rating points will migrate in no time.

Take a real 1000 player, have him play against a bunch of 400s and 500s and 600s who just got 'promoted' to 1000. That's a whole bunch of free tournament points for the real 1000. And as the rating system registers that as wins against equally rated players, our 1000 guy will rise to 1200 in no time, get to the next section.... and proceed to distribute back all the points because he's still a 1000 in terms of playing strength. Now you have guys in the U1400 who are inflated. Rinse and repeat until you reach the very top.

I don't know if I'm 100% accurate, but it seems like a very plausible scenario.

[Fred McKim]

Re: Rated players under 1000

Probably the most sensible approach is to have a small memory for players under 1000, so they can go up (or possibly down) quicker. I'm thinking 10 games instead of 25.

I've used Tom's training manual as a basis for teaching elementary school players. I think if you go through the material in 12-15 hours you can get 800 strength players for grade 4+, younger children will take more practice with the material. So in other words from a 0 to an 800 rating in 12- 25 hours.

The rating system should have a performance rating minimum of 0 and it looks like it might have from the tournament that produced 5 200 players.

I would gather that somebody who has learned the rules including understanding checkmate and stalemate, but no more should be around 400.

[Christopher Mallon]

Re: Rated players under 1000

If you have Chessmaster Grandmaster Edition you can choose opponents rated right down into the 200s. The lowest ones play completely random moves (but at least legal). The next step up (towards 300) know that a Queen is worth more than a Pawn. At 400 they might recognize the odd checkmate chance.

By the time someone is rated 800 they should be able to at least sometimes see 2-3 ply (in computer speak) and know the basic checkmates and a slight opening knowledge. That might be a "beginner" from an adult tournament perspective but hardly from a youth chess perspective.

For an adult it's different of course. My wife read half a chess book, played a few practice games and went to chess club for a couple of months and got an 8xx rating. There's no way a typical 6- or 8-year-old matches that learning curve.

[John Brown]

Re: Rated players under 1000

Without looking up ratings I'm assuming you that have replied are over my rating.
My idea is to eliminate an Under 1000 player. You all seem to think this is impossible. Anyone can be beaten on a given day. All the software the CFC is based on the higher ranked player always winning. If this was true then Carlsen should still be in the World Cup. The programs do not factor in a 50% human error.
The reason for 50% is we either play Black or White.
I'll give a personal example between me and Paul Leblanc. We played once and he was white and rated higher than me by the rating system formulas he should beat me.
Result was 0-1. Did I play better maybe not but the human factor was there and Paul Lost.

My suggestion was to eliminate U1000 players. Most have said it can't be done. But my personal opinion is: I think the real reason is that it does work and the higher rated players (due to the 50% error ) are afraid that these U1000 rated players will start chomping on their established ratings the same way they are chomping on the U1600's established ratings.

It's sort of like the rich and poor. Don't give the poor a chance or they might surpass the rich.
In chess it is elite vs amateurs. Give amateurs a chance to move up and they might become an elite at other elites' expense.

But either way my proposal will for sure eliminate U100 players. Whether it eats away higher ranked players is only theories that have not been tested.
The great scientists did not stop with the first test they tried but many more tests were done until they found a solution. Computers only calculate ideal solutions they do not factor in human error. That old saying to err is Human.

Hey what can hurt trying my idea and destroy the norm or tell me I'm full of wash and get more players stop playing in tournament (an income loss) against U1000 players. Revenue drops if you lose members. I feel that you'll end up with Elite players and Juniors playing in tournaments in the future if you leave the system the way it is.

[Alex Ferreira]

Re: Rated players under 1000

All the software the CFC is based on the higher ranked player always winning. If this was true then Carlsen should still be in the World Cup. The programs do not factor in a 50% human error.

Hello,

Sorry John, but that is completely incorrect. Players within a rating range of 400 should be able to occasionally score something against the other player.
My figures might not be quite right, but there's an expected ratio of points / results you should be getting based on ratings.

For example:
A player rated 2000 should be expected to score:

- 100% against a player rated below 1600 (16/16)
- 1/2 against a player of equal strength, 2000 (8/16)
- 3/4 against a player rated ~1800 (12/16)
- 7/8 against a player rated ~1700 (14/16)
- 15/16 against a player rated ~1650 (15/16)

- Nothing against a player 2400+
- 1/4 against a player ~2200

etc etc

This is of course, if playing strengths were static, math was pure, blah blah blah.
If, as a 2000 player, you score LESS than 1/4 against ~2200 players, you are actually under-performing, or should be rated lower, and the rating changes will reflect it.


Alex F.

[Christopher Mallon]

Re: Rated players under 1000

Hello,

Sorry John, but that is completely incorrect. Players within a rating range of 400 should be able to occasionally score something against the other player.
My figures might not be quite right, but there's an expected ratio of points / results you should be getting based on ratings.

For example:
A player rated 2000 should be expected to score:

- 100% against a player rated below 1600 (16/16)
- 1/2 against a player of equal strength, 2000 (8/16)
- 3/4 against a player rated ~1800 (12/16)
- 7/8 against a player rated ~1700 (14/16)
- 15/16 against a player rated ~1650 (15/16)

- Nothing against a player 2400+
- 1/4 against a player ~2200

etc etc

This is of course, if playing strengths were static, math was pure, blah blah blah.
If, as a 2000 player, you score LESS than 1/4 against ~2200 players, you are actually under-performing, or should be rated lower, and the rating changes will reflect it.


Alex F.

Actually that was the older calculation, the modern formula has a tiny chance to score even at huge difference, although with rounding it works out to being that at a 735 point difference you are expected to score 1 point in 100 games.

[John Brown]

Re: Rated players under 1000

Hi Alex;
As I said computers calculate ideal situations. But Human error will always change that. Sure on 10 games against the same player the higher rated should win. But one game one on one anything can happen. Such in the World cup most of the favorites are gone so where do your figures show on that.
I think it is time to realize that the lower rated players want change and unfortunately, I'm the only one with guts to attack the system. Yes I have an opinion but there are far more that sit on the fence to see if I triumph or burn. I'm unhappy and there are others that are unhappy but they are afraid to post. It is too bad that the little man is afraid of the big guy.
The system is flawed and the powers to be do not want to change it.

[Erik Malmsten]

Re: Rated players under 1000

I would gather that somebody who has learned the rules including understanding checkmate and stalemate, but no more should be around 400.

I've been shocked at the low level of play in kid's championship tournaments. Many can't mate with many pieces against king resulting in stalemate or 50 move rule. I think ratings below 1000 are very unstable as only realizing a few concepts will result in a large growth in results. They could realize these ideas at 400, 600 or 800 and quickly rise. On their way up their opponents over 1200 will drop 100s of rating points.

Chess growth isn't linear. While an 1800 will score 1/4 against 2000, I don't think a 2000 player scores 1/4 against 2200 as to be a master requires a change in thinking as well as knowledge. The chess ratings change less for master games. Perhaps they need to change more at the other end of the system.

[Fred McKim]

Re: Rated players under 1000

I've been shocked at the low level of play in kid's championship tournaments. Many can't mate with many pieces against king resulting in stalemate or 50 move rule. I think ratings below 1000 are very unstable as only realizing a few concepts will result in a large growth in results. They could realize these ideas at 400, 600 or 800 and quickly rise. On their way up their opponents over 1200 will drop 100s of rating points.

Chess growth isn't linear. While an 1800 will score 1/4 against 2000, I don't think a 2000 player scores 1/4 against 2200 as to be a master requires a change in thinking as well as knowledge. The chess ratings change less for master games. Perhaps they need to change more at the other end of the system.

At one time the CFC rating system would adjust a player under U1000 to their performance rating for an event (with certain limitations) if it was higher than their pre-event rating. I still think that a memory of 10 games for everyone Under 1000, would help with those players U1000 that improve every week.

The Rating System is based on the mathematical formula that 200 points difference = 75% (approximately) winning chances. So yes 1400 vs 1600 and 1600 vs 1800 and 1800 vs 2000 and 2000 vs 2200 will all yield the same averages.

[Ed Seedhouse]

Re: Rated players under 1000

Actually that was the older calculation, the modern formula has a tiny chance to score even at huge difference, although with rounding it works out to being that at a 735 point difference you are expected to score 1 point in 100 games.

The older situation was caused by using a simplified expectations table because of the difficulty, before computers came into common use, of computing the exact distribution. Computers can calculate the normal or logistic distributions quickly and so there is no need to use a simplified approach. The Elo system itself has no assumptions about a certain rating gap equating to a certain outcome. It is a purely probabilistic system so no matter how large the rating gap there is always a computable probability, however small, that the lower rated player may win.

[Christopher Mallon]

Re: Rated players under 1000

Another idea might be order of calculations... right now Provisional is calculated first, then established ratings. Perhaps after the initial "established" run, when bonus points are applied, re-run the ratings for everyone who didn't earn bonuses using the new post-bonus ratings.

That way someone rated 1000 but performing at 1600 over 5 games and earning the lifetime high bonus will be considered at their "post" rating of 1185-1304 (depending on score) for their opponents - which will be a bit less painful for the opponents.