[Al_U_Card]

Is the glass half full or half empty?

[awm]

I'm convinced that any method for generating individual ratings is flawed. The issue is that bridge is very much a partnership game. Established long-term partnerships will almost always do better than pick-ups. Fantoni and Nunes may well be the best partnership in the world, and Meckstroth and Rodwell are top notch as well... but throw together a team of Meckstroth-Fantoni and Nunes-Rodwell (in those partnerships and with only a few minutes to discuss) and I wouldn't pick them as favorites to win a top international event.

Individual ratings also have negative social effects (see okbridge) in that they discourage people from playing with or against those with very different ratings. They may create incentive for people to have many "aliases" with different ratings, perhaps to distinguish between casual play and serious play.

On the other hand, partnership ratings might be an interesting feature to add. An easy way to compute a rating might be to take the number of masterpoints won as a pair in tournaments, and divide by the total number of masterpoints that could have been won for placing first in all tourneys entered.

In fact I bet the jlall-jdonn partnership comes out looking pretty good on that measure.

[DrTodd13]

We have to first say what the purpose of the rating is before we should say what type of rating is good or bad. Ratings may exist for several different reasons: 1) a way of tracking your progress, 2) a way of identifying potential partners and 3) a way of selecting teammates, opponents, or people to watch.

In the first category, you may care about random partnership performance, established partnership performance, or both.

In the second category, a partnership rating is useless and you really just want to see how they did with random people. Two people could both have the same delusional view of bidding and work well together and could play terribly with everyone else.

In the third category, you may again care about both random and established partnerships depending on whether the potential pair in question are random or established.

So, I don't think it is possible to say whether individual or partnership ratings are "better." They are different and each have their own uses. In my view, the problems with ratings systems are the extent to which people will lie, cheat and steal to increase their rating. I believe I have a solution for this problem and I'm willing to sell the answer to the highest bidder.

[Winstonm]

Al_U_Card, on Jan 4 2006, 03:03 PM, said:

Is the glass half full or half empty?

Neither. The glass is too big!

Winston

[microcap]

I find pigpenz idea the best, as it lets the free market work! But a simpler way is just to add pluses and minuses to the various options, especially the advanced and expert ones. While self rating isnt perfect, more choices should lead to better accuracy.

In the meantime, I recommend that you just edit player's notes to mark experts and world class players that are nowhere close.

[uday]

Quote

I believe I have a solution for this problem and I'm willing to sell the answer to the highest bidder.

I'll bite -- was that a joke, Todd ?

[DJNeill]

Hi all,

Quote

I believe I have a solution for this problem and I'm willing to sell the answer to the highest bidder.


I'll bite -- was that a joke, Todd ?

I think he is referring slyly to ebay's peer-submitted ratings?

Dan

[pigpenz]

all rating systems have their flaws especially ones built up on attendance like the ACBL does. Jlall could quit bridge right now and come back at the age of fifty and probably be ranked as a novice LM and still be better than most ( i hope) that may in 30 yrs have 20,000 masterpoints more than him, accounting for mastpoint inflation

[helene_t]

DJNeill, on Jan 12 2006, 12:47 AM, said:

I think he is referring slyly to ebay's peer-submitted ratings?

Hmmmm .... some websites ask users to rate the information they provide. In the same way, BBO might ask players to rate each pick-up partner after the tourney. I wonder if such a system would encourage abuse: it might make sense to rate your favourite p as an moron (to scare off rivals) and to rate the morons as favorites (to mess up potential opponent pairs). Alternatively, an objective rating could be based on how succesful players are in keeping the same unfamiliar partner for a long sesion, or for attracting pick-up partners as selected partners. But it is undesirable if players start selecting the partners that will cause their rating to increase, rather than selecting those they like to play with.

Actually, when I went to high school, the teachers experimented with a peer rating system for pupils. The system prooved to promote corruption so it was abandoned.

But as for self-rating: I think the terms "beginner", "intermediate" etc are ill-defined. Yes, there are definitions but I'm sure many players don't read them. Besides, the definition of "intermediate" is subjective, and there is no room for non-beginners that are below intermediate (in fact a very large group, imho).

So maybe, when you first create your account, you should be asked more concrete questions like what kind of tourneys you have won.

Alternatively, the whole self-rating could be abolished. Imho it's rather worthless.

[Al_U_Card]

helene_t, on Jan 13 2006, 08:09 AM, said:

So maybe, when you first create your account, you should be asked more concrete questions like what kind of tourneys you have won.

Getting interesting. Say 10 questions, each worth 10 points.

number of years playing (1 point for every 5 years)
number of hands played per month (1 point for each 100)
avg. monthly imp score ( times 10)

then the next 7 questions are about real performance/results

Ever won a sectional event (10 points)
Ever won a regional event (10 points)
etc. depending on your area/organization.

So most players would be in the 30's to 60's and only top pros and real experts would get above the 60 mark.

[Elianna]

Quote

number of years playing (1 point for every 5 years)

For many people around me, the number of years playing should be worth MINUS points!

[Al_U_Card]

Elianna, on Jan 13 2006, 10:32 AM, said:

Quote

number of years playing (1 point for every 5 years)

For many people around me, the number of years playing should be worth MINUS points!

Okay then, how about after 30 years, you get 10 and then you subtract a point for each additional decade?

[DrTodd13]

DJNeill, on Jan 11 2006, 02:47 PM, said:

Hi all,

Quote

I believe I have a solution for this problem and I'm willing to sell the answer to the highest bidder.


I'll bite -- was that a joke, Todd ?

I think he is referring slyly to ebay's peer-submitted ratings?

Dan

I've actually implemented a peer-review system for bridge, like Dan says ala EBay. The system weights recent reviews more highly than older reviews...I'm currently using a pretty complicated piecewise function to do this but the concept is simple. Likewise, the system weights reviews more heavily depending on the number of boards played with the person but only up to some limit. The idea is that you can't make a very good judgement about another person after 3 hands but your opinion of them will likely become more accurate up to some threshold. After that, your opinion can't get that much more accurate. Currently I'm using a linearly increasing weight up to 20 boards and beyond that the weight doesn't increase but I'm currently working on trying to figure out whether 20 is the right number or not...doesn't have to be perfect, just close. The system would also indicate how many reviews each person had so you can see whether their early peer-rating might be influenced by a lack of reviews.

I think this system allows the rating to adapt over time as previous reviews will age and count less and less. It also prevents a small handful of steady partners from arbitrarily increasing or decreasing your rating by limiting their influence by the fact that their opinion counts as much as someone who has played only 20 boards with you. The obvious approach would have been to weight their judgement based on total number of boards without any sort of cutoff but that enables the sort of abuses discussed elsewhere in this thread. If the purpose of these ratings is to allow others to decide who to play with, then in a sense, only the opinions of pickup parnters should matter. Your regular partners will continue to play with you if they like you and they'll stop if they don't so the only issue is whether the first 10 or 20 boards with a new partner are likely to be enjoyable.

Also, my idea was to have separate ratings for bridge ability and amiability. You may have a great player who is a jerk or vice versa. The same system would be used for amiability. After each session with a player, you get a chance to rate them on a scale of 1 to 10 on ability and amiability. The ratings effect of vicious people who leave a table after a short number of hands with a pick-up partnership will be minimized by the fact that the weight of their rating will be low due to the fact that they haven't played many hands with the person. If you've played a lot of hands with the person then presumably you like them and in this case you'll probably see them more favorably than others might. So, I think these two characteristics may cancel each other out and give you something close to an accurate rating.

Todd

[Gerben42]

May work for American players but I would have no idea where to rate myself in this point system. What's a Regional or a Sectional (don't answer that, the point is that on this side of the pond, no one knows and no one cares...)

Here there are "national championships" (of which I have won none) and other tournaments, of which I have won many... The level of the tournaments depend on location, if there is prize money or only other prizes, and who happens to have time.

[pigpenz]

I think on BBO they have the ratings from 1-10 and jack to ace, but to accumulate points you have to play in $$$ events. Why not give the same type of points for people who play in anytype of tourney wether $$$ or free. I remember at first I used to want to kib some of these people who had qeeuns kings etc cause i thought they would be good to kib, but what i have found out some of these people are ones who play like 8-11 games a day in acbl events. So in these cases points are more for attendance( not really)

[Winstonm]

Maybe best is self rating on interest level:

Friendly Don't Care.
Friendly Kind of Care.
Serious But Friendly
Serious and a Jerk
Jerk and Really Serious
A Serious Jerk.


Winston

[awm]

You could probably go with a series of questions something like this:

1. Have you ever won an open tournament event which included 50 or more competing pairs (or 50 or more teams if a team event)? If no, then go to question 2. If yes, go to question 4.

2. Do you often (say around half the time or better) finish above average in open games at local clubs or tournaments? If yes, then rating is "Intermediate." If no, then go to question 3.

3. Have you ever played in a duplicate bridge game with open participation? If yes, then rating is "Novice." If no, then "Beginner."

4. Have you done consistently well in national-level events, excluding those that restrict entry based upon bridge level, age, or experience (several top ten finishes for example, or one national win)? If no, then "Advanced." If yes, then go to question 5.

5. Have you won more than one national championship, or won a zonal championship (this includes events like the Spingold and Vanderbilt in the NABCs in north america, and European Bridge Championship events, etc)? If yes, then "World Class." If no, then go to question 6.

6. Have you ever played in a world championship event with teams determined by the various zones, excluding those with age restrictions (for example the Bermuda Bowl, Venice Cup, or Olympiad), or ever won an open world championship (for example World Open Pairs, World Mixed Pairs, Transnational Teams)? If yes, then "World Class." If no then "Expert."

Of course, these questions are somewhat biased by my own view of the world (in particular I don't think playing on your national junior team automatically qualifies you for world class status, I know some disagree). But likely some series of such questions could be designed to better figure a level than just asking people who often don't look up what the terms are "supposed" to mean.

[mike777]

Gerben42, on Jan 14 2006, 06:13 PM, said:

May work for American players but I would have no idea where to rate myself in this point system. What's a Regional or a Sectional (don't answer that, the point is that on this side of the pond, no one knows and no one cares...)

Here there are "national championships" (of which I have won none) and other tournaments, of which I have won many... The level of the tournaments depend on location, if there is prize money or only other prizes, and who happens to have time.

Wow this seems like very wierd criteria to base level of tourneys on. Very wierd.

Location determines level? Who has time determines level of tourney?

Based on my reading of British and Aussie bridge magazines over decades this does not seem to be the criteria at all. I would guess the very elite tourneys have been elite for years and while the sponsor's name may or may not change over the years the level of the tourney is the same.

[Bon_An]

IMO,there are 3 parts on bridge playing,bidding,defence & declarer playing.To most players other than true world class,they should be evaluated by at least 3 corresponding committee consisted of certified teachers.
Any member would apply for a test(match) on bidding,defence or declarer playing,then their grades are definite.

That's for BBO on-line ranking system.

This post has been edited by Bon_An: 2006-January-17, 20:53