FICS Teamleague


Teamleague Forum

rating cap for individual players in each sectionProposalsIndex ->

posted at 2012-11-09 22:40 by Pawnadian

After having viewed a recent TL game in the Steinitz section between a 2200+ player and a 1600+ player, it occured to me that allowing such a strong player in a lower section is not very fair to the other players in that section. With that in mind, I think there should be a rating cap for each section. Once a maximum average team rating for a section is established, no player in that section should be more than say 300 points above that rating.

posted at 2012-11-10 19:15 by Funkmaus

Not only because that 2200+ player is me.
Having someone "really strong" in the lower sections means being much worser at boards 2-4. Hardly a surprise, that my team has lost all matches with 1-3 so far. :(
So it is rather unfair to the teammates, because it drastically raises team average, but not to other teams.
I wasnt on that team from the start.
I joined the Rangers AFTER Sections have been created. This team was on the bottom of TL, with -200 points on average to the next team.

posted at 2012-11-11 17:48 by wmahan

My initial reaction was the same as Jenny's. Her team is last in the division with no chance to finish first. If the rules aren't broken, why fix them?

However, I do see a problem in the fact that she is able to play on both the top board in the league and the lowest section. Our lowest section for T52 has a rating limit of 1631. In contrast, in T45 (the last with preset limits) we had 4 U1400 teams, and in T42 we had 7 U1400 teams.

In my opinion the loss of such lower-rated teams is a detriment to the league. This seems to be an unanticipated consequence of the move away from fixed section limits. No one wants to have the lowest-rated team in the league, and in the absence of any anchor, all teams float toward the top.

Milpat previously raised this issue in the "U1400 / U1500 section?" topic of the T52 forum.

posted at 2012-11-14 12:05 by PankracyRozumek

I am not sure if I would like to change anything in the current rules, but if I wanted to, I'd rather set maximal rating difference limit between the 1st and 4th board in each team (e.g. to 300 or 400 points) rather than just set up a limit for the 1st board.

Of course such a maximal rating difference limit would make no sense for the Fischer section...

posted at 2012-12-13 17:02 by bythecliff

I second the motion to have a rating range limit within the team. Top board should not be a certain amount rating higher than the bottom board. I'd say 400 is a good limit -- there's plenty of flexibility, while ensuing that most people have opponents within reach.

posted at 2013-02-11 09:38 by OldRaptor

I agree with Funkmaus. The unfairness generated is to the team with the "ringer", not the opponents. Without having read Funkmaus's comments earlier, When my team Mysterious Knight Moves was paired against her team, Wunderblunder in Rd 1 of TL53, I had 2 thoughts. First what an opportunity it was for me to play against such a high rated player. Second what an advantage for my team! Because of her high rating, we had the rating on boards 2-4, and indeed we went into our game up 2-1. This equalized my match with Funkmaus, as we would get the match point with a WIN or a DRAW, and would get a half match point with a loss.

Before the match I calculated The odds of our team winning or drawing the match against the higher rated Wunderblunder was about 55% - 65%. Before my game, leading 2-1, I calculated that the odds of me winning or drawing (i.e. NOT LOSING) was about 35%-45%, so not so bad, considering the 331 point rating difference. And my teammates had already secured the half point match draw as a minimum result.

Our game ended when I saw that giving up 2 Rooks for a Bishop, pawn, and Rook would force a draw by perpetual check and give us the match point. Pleasantly after the combination was played, I discovered the continuation, which won her Queen for my Knight, and ended the game with a win. The continuation was not forced, but the alternate lost a Rook and the position as well.

Currently the game is the leader on the new "Game of the Week" feature of FICS for round 1.

So all things are not bad under the current system. It is up to the captain whether or not to include a super high rated player on the team, giving them a disadvantage on the other boards. The best odds overall for a team would be to have all players on the team equally rated. Of couse that is not practical, and other factors such as player relationships come into play when making that decision.

posted at 2013-02-12 03:23 by wmahan

Your math is wrong. Taking Funkmaus's RD as 75, your expected score is

which is about 13%. The glicko system does not contemplate the probability of a draw, but even in the extremely hypothetical case that you never win and only draw--which would be the best case in your match situation--you would only avoid losing 26% of the time.

posted at 2013-02-12 08:52 by OldRaptor

Which opening were you analyzing? She plays the Caro-Kann against E4, and in fact DID play the Caro-Kann against me. Did you read her finger notes, where she states that she plays conservatively waiting for her opponent to make a mistake. She states further that if both players are conservative, that her games tend to end in a draw. Further, did you figure in your analysis that she could not play for a draw, because Mysterious_Rook_Moves was leading the match over Wunderblunder by a score of 2-1? This fact forces her to change her game and play for a win. Did you consider that I could now make moves that lead to a draw, making moves that she could normally accept, but now had to avoid? My friend you obviously have overlooked the math it terms of team chess. One last observation about your math. It is obviously wrong, because if you should look at the results >>>>> OldRaptor WON>>>>> so your hypothesis "in the extremely hypothetical case you never win" was proven wrong on the very first data point. Further, since you clearly state that the glicko skystem does not contemplate the probability of a drawâ€Â¦Ă˘€Â¦ how can you possibly calculate the odds of drawing twice, verses winning once, which is what you seem to do.

I got my number of 35%-45% by looking at Funkmaus' record on FICS and making subjective adjustments for the factors listed above. Saying that my math was wrong, implies a miscalculation. When in fact I did not use the formula you presented, but relied on empirical data, game preparation, and cognitive subjective adjustments.

I do have a question though. Where does OldRaptor's RD fit into your formula?

Finally, this discussion was agreeing with Funkmaus' assessment that having a strong rated player on a weaker team HURTS, not HELPS that team. I was meerly using our gameâ€Â¦ a single data pointâ€Â¦ as a recent example to support her hypothesis. Your math does nothing to change that.

But thanks my friend for your input.


posted at 2013-02-13 05:55 by wmahan

I got my number of 35%-45% by looking at Funkmaus' record on FICS and making subjective adjustments for the factors listed above.

In other words, you made it up, rather than calculating it. It should be obvious to an experienced player, even without doing the math, that you do not have a 35%-45% chance to beat or draw a player 339 points higher. Your implication that your situation was somehow unique because you needed a draw is specious, because most players would be happy with a draw against a much stronger opponent.

The actual result of one game is completely irrelevant to a discussion about the probability distribution of game results.

I'm sorry you didn't understand my post, but I suggest you read the glicko help file I linked to if you want a more detailed explanation. It's nice that you agree with with our conclusion; I was just pointing out that your "calculation" was not correct.

posted at 2013-02-13 09:15 by OldRaptor

O.K. I'm happy to appease you and go with the 26% as presented by your argument.

I guess your assumption is that a rating is broad based and equally valid for all openings. There would be no difference if the rating were calculated separately by each opening, i.e. a person who plays the carl-kann would have the same rating as if they played the english, or the vienna.

Let's also assume that draws are always achieved because the player could not achieve a win, and at the same time was able to avoid a loss. Never is a draw achieved because that was the desired result.

Let's also assume that emperical calculations are not valid.

English is a non-precise language with words meaning different things. Webster's Online dictionary defines "calculate" as
a : to determine by mathematical processes
b : to reckon by exercise of practical judgment : estimate
among other definitions. Here is the link:

I'm just pointing out that my "calculation" was correctâ€Â¦ if for no other reason, it was MY calculation, not yours. I'm sorry that you didn't understand my post explaining that.

Now, if your arguement is that your math skills surpass that of an aging 64 year old man who has trouble remembering to zip up his pants, I'll concede that.

One last thing. Please do not be so presumptuous to assume that I have not read the glicko help file.

posted at 2013-02-13 09:38 by OldRaptor

Getting past all the irrelevant discussion, I'd like to note up one more point about gaming the system by bringing in a high rated player to an otherwise low rated team. The original conclusion by Funkmaus, supported by me, was that it hurt the team, not the opponents by giving the team a disadvantage on boards 2-4.

For the purpose of further discussion, I will assume 4 player teams (not 5 or 6), and that everybody plays every week with no defaults. This assumption is due to the fact that only 4 boards are played, and the average rating is determined by the top 4 players, specifically excluding players 5 & 6.

If a team captain wanted to game the system, one should bring in a low rated player, say with a 900 rating, to play on board 4. This would have the opposite effect of Funkmaus' conclusion, giving the advantage to the team on boards 1-3.

Obviously the only way to ensure being equally or better rated on all 4 boards would be to have the entire team each have the identical rating equal to the top rating allowed in the section. However TL does not have pre-determined rating brackets, so that would be impossible, or let's say highly improbable. (Try calculating that probability!)

posted at 2013-02-13 11:43 by bowserjrzhu

Im quite sure Funkmaus joined AFTER the sections were created, thus it did not HURT the team. It raised the team average, but it did not move the team up a section.
Look at this.

Opponents team Wunder Blunder
Brd 1:Rating 1950 2100 *Funkmaus
Brd 2: 1866 1800
Brd 3: 1790 1700
Brd 4:1779 1700

If Funkmaus was gone, Wunder Blunder would be in the SAME section, but different opps, like the 1800 would have to play the 1950

I hope you understand as i cannot write well :)

posted at 2013-02-13 12:48 by Funkmaus

Right bowserjzhu - also in Wunderblunder case I got added to the team after the sections been created.
In a 4-Board league neither "lowbie" on 4 nor "monster" on 1 makes it. With a lowbie on board 4 team is forced to make 2,5 points on top 3 boards... Hard enough.
In my TL experience as player and TD only well balanced teams (equal ratings on all boards) achieved playoffs and won the section.

Congratulations on your win again, OldRaptor - I havent expected that game will cause such a huge discussion and odds calculation before/afterwards.
To me it was just a normal game - one among the others - and I have to admit, I cant remember being crushed like that. Well done.


posted at 2013-02-13 13:36 by OldRaptor

Thanks Funkmaus.

I never expected the attention either. As we all know, one data point does not a trend make. I will fear your game as much the next time, as the first.

To me it was more than a normal game. It is rare to have the opportunity to play such an able opponent under tournament conditions. It would have been an honor to have lost, provided that I played reasonably well, without a tactical blunder.