[HighLow21]

sfi, on 2014-March-02, 16:44, said:

Congratulations, you are a full 50% better than Jimmy Cayne according to the ratings. That might actually be true, but it sure seems like an indictment of the system itself more than anything else.

It's not true--I've watched him, and rate him as one of the best in the world. He also plays regularly against the best in the world. He's way better and way more seasoned than I am. (He's also a lousy banker, and made millions while he abandoned his failing company in order to play golf and bridge, but that's a separate issue.)

In statistical terms, jec is a complete outlier--in a regression that, in my mind, has a very strong R-squared. The reasons that he's an outlier are readily identifiable and thus removing him in the sample is justifiable.

To retort by pointing out one outlier and thereby implying that my argument is invalid, however, is completely specious.

My response:
(1) For every one jec that plays only with and against world-class competitors, there are 1,000 players in the rest of the pool for whom knowing the stats would be very useful.
(2) If jec regularly played against average opponents his IMPs average would unequivocally be higher than mine.
(3) Even on bboskill.com, there is a feature allowing a player to be recognized as a "VIP." They even explain what this means, and the reasons behind it, on the website. jec is a very-well known VIP--perhaps the quintessential VIP.

Your move.

[sfi]

You really want me to go through all the internationals I know who have lower ratings than yours (all of them that I checked)? No thanks, but it's pretty clear that the data is poor when one of Australia's best players manages a rating of -0.02 despite winning another title on the weekend.

[bluecalm]

Quote

In other words, some information is more helpful in determining who is good and who is not than no information. Sure, without perfect information, there will be prediction errors such as the allegorical ones you listed. But they will be dominated by the correct predictions made in the face of actual information and analysis.

But publishing the ratings even if 100% objective would be bad for the game imo even without cheating considerations. It's nice that people think they are good, they continue playing and have fun. I don't see any benefits from objective rating in current state of internet (and live) bridge. You want to prove you are good ? Win something. Want to have good partners ? Go convince them you are good (player and partner) and/or win something. That's it.

Quote

jec

I don't want to debate how good jec is but the odds are heavily stacked in his favor: he plays with the biggest stars of the game on regular basis and his team is about always stonger than the opponents.
It's the same for many other players. In bridge only partnerships score points so any kind of rating is going to discourage playing with weaker players in both teams and partnerships. This is reason enough to not go in this direction.

[helene_t]

fwiw I think it would be a good idea to get rid of the skill level. And the country, now we are at it. If BBO wants to make some of the profile fields filterable (languages spoken, bidding systems played) then by all means have them in the profile. Other than that, free text and the picture are sufficient IMO.

[OldPlayr]

Playing individual games on BBO, to quote Forrest Gump, is a “box of chocolates – you never know what you're going to get”. If it matters that much to you, only play with a regular partner or with partners that you are familiar with. Otherwise, chill out and smile at whatever happens. It's only a game!

Self-ratings are worthless. Complicated rating schemes are usually flawed. Anyone who is as good as many people here claim to be should be only be playing for money. Money games rate themselves.

As for me, I have no partner, so I play only in individual tournaments and pick-up games in the main club. When I lose, it's obviously due to a poor partner. When I win, it's clearly my skill. Having poor players out there is key – winning all the time would be boring

[HighLow21]

OldPlayr, on 2014-March-02, 18:40, said:

Playing individual games on BBO, to quote Forrest Gump, is a “box of chocolates – you never know what you're going to get”. If it matters that much to you, only play with a regular partner or with partners that you are familiar with. Otherwise, chill out and smile at whatever happens. It's only a game!

Self-ratings are worthless. Complicated rating schemes are usually flawed. Anyone who is as good as many people here claim to be should be only be playing for money. Money games rate themselves.

As for me, I have no partner, so I play only in individual tournaments and pick-up games in the main club. When I lose, it's obviously due to a poor partner. When I win, it's clearly my skill. Having poor players out there is key – winning all the time would be boring

Agreed with everything you said... I was simply stating my full position on the issue and backing it up with argument, evidence, and facts... since I figured that's what this post was for in the first place!

And yes, whenever I go down in a cold contract it's always my partner's fault.

[PhilKing]

HighLow21, on 2014-March-02, 15:58, said:

- Simple Bayesian analysis would indicate that someone with a -1.2 IMPs average is FAR more likely to be a truly bad player than someone who is Advanced and regularly plays way out of their league. There may be 1 of those for every 10,000 terrible players out there. That 1 person could simply create a new login if he/she wanted to start fresh.
- Similarly, you can get a great result, in theory, by "bunny bashing." But seriously, just try to pull that off in the long run. You have to find a willing, good partner, and then selectively allow only bad players to your table to whip on them. They have to stick around or be replaced by other bad players. In my experience, this is completely unsustainable in the long run. Anyone who has good results over a decent number of hands actually knows what they are doing.

I might be being a bit thick, but in my mind, a simple Bayesian analysis is a contradiction in terms.

Take the -1.2 player: to achieve this against run-of-the-mill players would be an astounding feat, but to do so with a weak client against world-class opposition would be standard. For every "pro" that takes on the best with his donkey, there are 10,000 astoundingly bad players but they are usually playing against eachother - that's Bayes', and that shifts the odds.

[barmar]

OKbridge has Lehman ratings. They're objective. AFAIK, no one wants them here.

[1eyedjack]

To my mind, a major problem with a candle in the darkness is that the uninformed, which would make up 95% of BBO population (and closer to 100% of the whiners), would look at that candle and see an arc-light.

[Bbradley62]

barmar, on 2014-March-02, 22:49, said:

OKbridge has Lehman ratings. They're objective. AFAIK, no one wants them here.

I think "no one" is an overbid.

[mycroft]

...and the ways to game the Lehman system, like any other system, are legion.

The only thing that might work is the "hidden ranking match" discussed a number of times. It's always there, it's always running, it's only used to match you with "random" partners and opponents, and nobody can see it. Ever.

And, of course, given those criteria, it may in fact already be in place. If you can't know by regulation, BBO can't tell you, can they?

[dustinst22]

Bbradley62, on 2014-March-02, 23:58, said:

I think "no one" is an overbid.

From another planet overbid.

[1eyedjack]

dustinst22, on 2014-March-03, 17:35, said:

From another planet overbid.

Just an overbid.
It should not be hard to find someone who wants Lehmans introduced. Indeed given the number of members it would be extraordinary if there were literally no-one.

And I also have no doubt that anyone with an agenda who wanted to drum up support could find a pocketful of players to add their endorsement

But I am not aware of any reliable impartial research in the public domain regarding the size of the support base for such a change, and we should take care not to confuse personal strength of conviction of its merits with a presumption of widespread support.

[Zelandakh]

Hello shoemaven and welcome to BBF. The self-assessment skill ratings should not be about judging how good you think you are but rather using the criteria SimonFa's post (#10). For example you self-rated as Advanced ("Someone who has been consistently successful in clubs or minor tournaments.") Is that an accurate description of your face to face level?

I took the liberty of looking up your results for the last month and found 50.35% and -1.28 IMPs per hand, apparently at tables consisting of GIBs and intermediate level players. Just looking at this result in isolation, a program that issued warnings for overstating skill level might decide that this qualified. Of course it might just be that you were unlucky this month, or had terrible partners, or that the opposition was better than it looks at first glance.

And that is precisely the issue. If BBO starts issuing warnings to players that have self-rated in good faith then they will quickly find that many of them become ex-players. Even WC players have bad sessions from time to time - what would Benito Garozzo think if he got a warning in the post one day? In other words, while your suggestion appears reasonable from first glance it is actually fraught with problems and would almost certainly lead to a lower quality service. As an example, one result might be that players are more reluctant to play with pick up partners in the MBC, or if they do will leave the table (or boot partner of they are host) anyone that they think did not play optimally. How would you feel if you had worked hard to set up a squeeze ending only to find that an honour was misplaced and you went down where a simple finesse would have worked, even though your line had a far higher success rate, and then to top it all your partner booted you as a "n00b"? And getting similar reactions at 5 or 6 tables in a row without doing anything wrong. Now think about how such an experience would come across to someone new to BBO.

Perhaps you can now try and look at things from a wider viewpoint and see the failings of your suggestion. I doubt that anyone would be against a mechanism that supported more accurate ratings on BBO; but only if that rating system did not introduce issues, many of which have turned up time and time again on sites that have tried this. One of the reasons BBF has been so successful is precisely because there is no formal rating system. Until someone comes up with a method that avoids the issues, that will remain the case. Better imo would be to improve the current self-ratings to split the current intermediate level, which probably accounts for 80% of members if they rated correctly, in two. That would allow the "better intermediates" to distinguish themselves from the rest without having to overstate their level to Advanced. But this has also been suggested before and I am not holding my breath that such a change is happening any time soon.

[gordontd]

Zelandakh, on 2014-March-05, 09:12, said:

Even WC players have bad sessions from time to time - what would Benito Garozzo think if he got a warning in the post one day?

Zia told me he had a game on Bridge Club Live, a popular English online site, with a friend of his and they scored about 42%. Just as he was leaving he was amused to receive a message from another player saying "I bet the real zmahmood wouldn't be impressed with that score!"

[lycier]

helene_t, on 2014-March-02, 17:58, said:

fwiw I think it would be a good idea to get rid of the skill level. And the country, now we are at it. If BBO wants to make some of the profile fields filterable (languages spoken, bidding systems played) then by all means have them in the profile. Other than that, free text and the picture are sufficient IMO.

I agree with you.

[mycroft]

gordontd, on 2014-March-05, 09:34, said:

Zia told me he had a game on Bridge Club Live, a popular English online site, with a friend of his and they scored about 42%. Just as he was leaving he was amused to receive a message from another player saying "I bet the real zmahmood wouldn't be impressed with that score!"

Heh, and I bet that was exactly correct - the real ZMahmood wasn't impressed by it :-)

[lycier]

helene_t, on 2014-March-02, 17:58, said:

fwiw I think it would be a good idea to get rid of the skill level. And the country, now we are at it. If BBO wants to make some of the profile fields filterable (languages spoken, bidding systems played) then by all means have them in the profile. Other than that, free text and the picture are sufficient IMO.

Especially "get rid of the country" will do good to our chinese since almost of our chinese know Euramerican players look down on us in the heart,this is a fact.so actually many of chinese players have to select others flag,the main choices are USA ,Canada ,singapore and Hongkong etc.

[mgoetze]

HighLow21, on 2014-March-02, 17:29, said:

In statistical terms, jec is a complete outlier--in a regression that, in my mind, has a very strong R-squared. The reasons that he's an outlier are readily identifiable and thus removing him in the sample is justifiable.

LOL, I see not much has changed here in my most recent absence. HighLow21 is still here and he is still spouting *****.

FWIW, I will reiterate a suggestion of mine: optionally display a rating based on matchpoint results from robodupe tournaments. BBO can earn some money thanks to people playing robodupe just to get rated, and if someone manages to consistently score 55%+ in robodupes you know they have some minimum level of competence.

[johnu]

lycier, on 2014-April-12, 10:53, said:

Especially "get rid of the country" will do good to our chinese since almost of our chinese know Euramerican players look down on us in the heart,this is a fact.so actually many of chinese players have to select others flag,the main choices are USA ,Canada ,singapore and Hongkong etc.

The obvious solution is to change your self rating to World Class to fool the rest of the players Of course, after a couple of hands (or maybe just 1), your real level will start to become apparent.