Rating Changes between 2006-2010 for BC vs Ottawa

[Kevin Pacey]

Re: Rating Changes between 2006-2010 for BC vs Ottawa

I remember that Ottawa had about ten adult master players, of varying ages, who went below 2200 during the ~2004-2006 CFC regular rating deflationary period (at least that's what the CFC Ratings Committee deemed it to be, before the relatively recent rating boon).

Those Ottawa [ex-?]masters I refer to could at least include: myself, Hum, Gordon, Barclay, Simic, Patterson, Beckwith and the borderline cases of De Kerpel, Nezirovic, Kirby and Gelblum, if anyone wishes to trace their regular ratings back through their CFC ratings archives.

Some of these players haven't been so active, but except for myself none seem to have their regular CFC rating currently higher than after the boon to their rating that they received at the end of 2006 (which of course doesn't disprove the theory that there is national/regional inflation of the CFC regular rating system as yet) - a boon that Roger gives inaccurately as happening in 2007 in his first post of this thread.

[Kevin Pacey]

Re: Rating Changes between 2006-2010 for BC vs Ottawa

Your theory and understanding of the openings you play appears to me to be quite good. Your willingness to do the work (work ethic) appears to be paying off for you.

At 50 you're still a young man.

I try to cut down on what I absolutely need to know, so I'm not so sure about having the work ethic, one that I certainly had when I was a young player.

Emmanual Lasker is a great model for older players, I think. He always played the man.

Korchnoi, who just keeps on going like the energizer bunny, had Lasker as a model long ago. Timman, another strong oldster, is a workhorse in his study. Both Korchnoi and Timman play lots of main line openings, to be used at least against strong opposition. They do use trickier sidelines against lesser lights though. A good way to save energy at the tournament hall.

I've been told I look at least 10 years younger than my age. Not much grey, but some hair loss. It's sad that Tom and I are fitter than some kids, considering that we're both not so fit.

[Duncan Smith]

Re: Rating Changes between 2006-2010 for BC vs Ottawa

Gary's bit of comedy about age 20 isn't really relevant; Noritsyn and Khaziyeva both won Canadian national closed titles around age 17. I suppose bringing up actual events might spoil the fiction but so be it. Magnus Carllson comes to mind for some reason, I suppose if he hasn't played correspondance Gary won't know who he is :).

Ratings are getting interesting these days because the changes came in when there was a sudden drop in regular tournament attendance with some turnover as well. I suppose there is a chance that active players under the new system are much sharper and more skilled then the group that isn't playing much. But I look at players like Charbonneau for example and he seems much stronger to me then some players coming up the CFC rating list to around his level or higher.

[Gary Ruben]

Re: Rating Changes between 2006-2010 for BC vs Ottawa

Gary's bit of comedy about age 20 isn't really relevant; Noritsyn and Khaziyeva both won Canadian national closed titles around age 17. I suppose bringing up actual events might spoil the fiction but so be it. Magnus Carllson comes to mind for some reason, I suppose if he hasn't played correspondance Gary won't know who he is :).

You're comparing a Canadian Closed to what Magnus Carlsson plays and wins. Like Canada is some chess super power or something.

Most young players don't hang around long once they hit the wall. It's great when they're young and beating older players or younger ones. Once they get a bit older and realize they don't stack up against their peers, many drift away. You don't have to tell a player s/he isn't good enough. They know themselves. The players who love the game stick around for the competition.

Carlsson and a few others are special.

Chess isn't a game or working a rating system, Duncan. It's a way of life.

[Duncan Smith]

Re: Rating Changes between 2006-2010 for BC vs Ottawa

The field was 70 players including an A class player from Manitoba. You however objected to another entry based on technicalities. The organizing committee ( which remained silent throughout the process ) created the controversy in the first place.

This was always the CFC way, create problems for people and not take ownership of their responsibilities.

[Fred McKim]

Re: Rating Changes between 2006-2010 for BC vs Ottawa

Roger: I read your rating study and think that may be the level of detail we need in order to determine a fair assessment of any inflation (national and regional) and figure out how to award bonus and/or participation and/or feedback points (I talked about this in length at the CFC forum).

However I think a better indicator to inflation requires us to also know the net rating change per game played.

So for example we might find that Ottawa players are gaining 2.0 pts per game, and the same for Vancouver (I'm just speculating here, I don't know the answer). Victoria might not come out at 0 once you did the per game analysis.

I am going to do an analysis for the really small PEI pool later today, and compare the differences ( diff per player vs diff per game).

Probably the analysis should be limited to 1600 and above, as well.

[Roger Patterson]

Re: Rating Changes between 2006-2010 for BC vs Ottawa

well, I can do that but it will take a few days (as getting the data is all manual).

[Gary Ruben]

Re: Rating Changes between 2006-2010 for BC vs Ottawa

Fred, I think the "problem" is under rated juniors who are pulling rating points from the more established players. The juniors get good, fast.

The solution is to take a players age and add two zeros for a rating, if the rating is not already at that point. In other words, when a player turns 15, he gets adjusted to 1500, unless he has more than that. At 18, which should be the cutoff for this, the rating should be adjusted to 1800 if not already there, or higher.

I know a lot of players will lose the points quickly but it helps inflate the higher end of the rating curve. Kind of a bottom up approach which stops a quicker erosion of the ratings of the established players.

A 17 year old with at least a rating of 1700 is a more attractive opponent than a 17 year old with a 1300 rating.

Not that I expect anyone would try an approach so simple.

[Fred McKim]

Juniors cause deflation

Gary: Another fix would be to reduce the game divisor for juniors from 25 to 15 or 12.5 or increase the adult divisor.

FIDE has different divisors for different age groups.

[Fred McKim]

Rating patterns (regional)

Roger: Being the smallest province it didn't take long to do a rating analysis

We generally hold 3 CFC Opens per year in Charlottetown.

My parameters for the main study are established CFC adults aged 25+ as of 2007 (post boon), and rated over 1600 (to try and eliminate those adults most likely to be improving).

We have 7 players, rated mid 1600 to mid 2200, at the beginning, age about 27 to 60, at the beginning.

There are a number of PEI Juniors that were not included as part of the core group, and are generally not strong enough to significantly take points from the 7 identified.

Total of 260 games played
Total rating gain of 8 points

Knowing all of the players very well, I would say a net gain of 0 is reasonably accurate for their skill development.

OK. That's the easy part, now we can consider the effects of participation points. These 260 games generated approx 223 points into the system.

So if there were no other players involved this group would increased by a total of 223. Instead 215 points have been diverted to other rating pools.

We have one interesting player who was showing more than normal growth during the period. He started in 2007 at about 19 (thus not part of the core group) and gained 398 points in 44 games. Subtracting about 60 participation points we can see that he took 338 points from the rest of the system. In other words he was on average at least 200 points underrated in all of his tournaments.

There were a total of 21 games between the main pool and this individual. I didn't do the exact math, but we could expect that we can pro-rate and the improving young adult took about 165 points from the core group, and the rest of his 173 points from other players.

Now considering the core group we can see that eliminating the games of the young adult we see a rating loss (not considering participation points) of only about 50 points over about 240 other games. Realizing that many of these 240 games are against each other, I woud say this was more like 50 points over 165 games. For this reason I feel that cries of inflation have to be considered carefully and we may have to break the country into pools to do the proper analysis.

If you have an area where there are no improving juniors / young adults competitive with the core group there is bound to be inflation.

If you have an area where there are plenty of competitive juniors / young adults you will have deflation.

In an area like PEI it turned out the participation level was just about spot on.

I think we may seriously need to consider having 2 or 3 divisors in the rating formula.

On the CFC forum there has been talk of having a Rating Overview Committee to look into this next year.

[Gary Ruben]

Re: Juniors cause deflation

I guess the method to use depends on what one thinks is the problem and the best way of dealing with it.

Probably it's a good idea for the junior ratings to reflect what one would expect from the age group. I used to adjust juniors ratings in CC upward to more closely reflect the players correspondence and OTB playing strength. An expert player doesn't want to play in class B events, as an example.

The advantage an adjustment has over divisors is it's faster. Something to consider when the dropout rate seems to be so high. It might help to keep members.

[Roger Patterson]

Re: Rating patterns (regional)

actually, I'm not sure it will give more information than the average rating gain. What's important is either:
(average rating gain)/(average #games per player for a region)
or
(individual rating gain)/(average #games per player for a region)

but not:
(individual rating gain)/(that person's # games).

The last one is probably what you are thinking of but is erroneous - even if you play a lot of games, those extra participation points are dissipated into the system.

Although, knowing the (average #games per player in a region) is useful info.

[Fred McKim]

Re: Rating Changes between 2006-2010 for BC vs Ottawa

Hi again Roger: I can do the same sort of measurements you did for my core group and I get a mean loss 0.9 or a median gain of 0.2

No matter how you cut it the participation points seem to have worked well here. I was not a big fan of this when it was introduced as it seemed to lack any real analysis.

However, I disagree with your seemingly quick dismissal of my analysis in the post above.

Inflation is probably best thought of as a rate. You might be determining how much inflation there might be, but I'm interested in measuring a rate of growth (up or down) and curtailing it.

For example if Ottawa has gone up roughly 100 points, I would say that is really 33 points per year, assuming conditions remain roughly the same. If we wait another year, 100 probably won't be valid any longer.

I consider inflation / game played as also significant, since we are combatting it by participation points / game.

I think a compromise analysis could use games, but limit the weight of each player to 25 or 50 games. In my core group one player has only 5 games played, while the most active has 83.

[Roger Patterson]

Re: Rating Changes between 2006-2010 for BC vs Ottawa

sorry Fred, I hadn't seen your analysis post - I was making an addendum to my previous post. Not sure why I hadn't seen your analysis - didn't refresh my browser maybe.

In any case, I've added up the activity for non junior players and get:

[edits added - correcting some math, trying to post too quickly.:-0 )

Ottawa:
average #games/player = 145
(average rtg change)/(average #games per player) = 0.7 [edited]
(median rtg change)/(average #games per player) = 0.7

Vancouver
average #games/player = 71
(average rtg change)/(average #games per player) = 1.3 [edited]
(median rtg change)/(average #games per player) = 0.7

Victoria:
average #games/player = 86
(average rtg change)/(average #games per player) = 0.0
(median rg change)/average #games per player) = 0.0


I think that would tend to support the hypothesis that ratings are inflating on a per game basis and that regions with different levels of activity are inflating differently (and have diverging ratings) [and as rating increase per game is comparable [edit] to the participation points being awarded, tends to not confirm the presence of any counteracting deflation]

Victoria (and your sample) is probably too small a sample to take too, too seriously. For that matter, policy should not be decided on only two cities either. Changes to the rating system should be made on carefully considered broad based measurement and scientifically based reaction. But... the governers decided to implement (twice) changes to increase the ratings based on hearsay evidence of deflation. Now, it appears as a result, that we (may) have a) differential inflation between regions [a bad thing], b) inflation roughly corresponding to the amount the governers added.

[Kevin Pacey]

Re: Rating Changes between 2006-2010 for BC vs Ottawa

Hi Roger:
From the link you gave in your first post of this thread,

http://www.victoriachess.com/misc/ratingstats.htm

The following is given under your motivation from your study:
"b) an observation by me that people I considered to be my peers in Ottawa seemed to be going up in rating as part of a general inflation but no such trend seemed apparent in BC"

Which people did you consider your peers in Ottawa in this reference to them?

Of the 11 Ottawa masters I listed in an earlier post of mine on the first page of this thread, only my rating seems to have risen higher at this point in 2010 than after when the 2006 rating boon occured. I can't think of many other possible peers of yours in your Ottawa days who have been very active in the 2006-2010 period.

Is it possible B.C. ratings have deflated? Or is there some regional inflation (or deflation?) for players just below 2200 (i.e. inflation is not [yet] linear)?

The study may be sound, except the conclusions may be incorrect.