[sceptic]

Ok, This does not include world class just the right to be called an expeart

Set up loads of 64 board tourneys and you have to get over 70% in at least 20% of them to call your self an expert rating only starts when you have played 10 tourneys and you have to play 10 a year

by default if you score less than 50% in more than one of these tourneys you have to call yourself advanced until you get your rating back up, this may not work in an indy but should work for a pair

you can play about with the percentages but it has to be better than selfratings

(A few more bottles of Tanglefoot and I will come up with another fool proof scheme for ratings)

[dogsbreath]

hi
there's more to bridge than match-pointed pairs
rgds
Dog

[DrTodd13]

I went to a pairs event at a normal night at my local club. I saw a lot of people I didn't know but I did know who were undoubtedly the best pair in the room. Head and shoulders above the rest. Their result? 46%. It just so happened that all their good decisions turned out wrong on this set of boards thanks in part to trying to draw inferences from people who didn't know what they were doing. Put these people in an all expert tournament and they would do very well but if you have an all-expert tournament, no one will score 70%. This doesn't sound like a way to define expert to me.

You are an expert if people consider you an expert. As far as I'm concerned, there is no objective definition of "expert."

[Wayne_LV]

   

[Free]

Imo: bad idea... Defining an expert would test you on bidding and play skills. However, people prefer different systems to get through the auctions, have different styles, different leads, different carding methods, everything is different! So how can one make an objective test? It's impossible.

It's like Todd says: you're an expert if people consider you an expert. But implimenting such system might get people taking advantage of this feature and consider everyone a loser. Still I think that this is 'probably' the best way to have some improvement. I've even thought about several rules to counter abuse. It could be pretty accurate, but it would cost even more bandwith while loading the lobby. So first we need a new lobby structure before such system could be used.

The rules are:
- everyone can SET any player's level he has played against (and with), by 'edit player notes for xxx' (like the neutral/friend/enemy system). With a radio button you can select the level. Default is obviously neutral which doesn't count.
- if a player selects too many "novice" his votes don't count. Perhaps a percentage would be useful. The same might be used for world class settings, but some people always play with WC players so perhaps it's not such a good idea.
- it's probably necessary to create a database instead of cross referencing everone's settings to every player in the lobby.
- it might be useful to block players catching a zero to select 'novice' for his opponents.

My general thought is that it's quite complex, and still not foolproof. However I think it would probably give a better representation of players, and perhaps there are some adjustments possible to make it even better.

[Free]

Wayne_LV, on Oct 8 2006, 10:43 PM, said:

How about this in lieu of a rating system that puts players in a category.

The BBO hand records are kept for all players for 1 month back for both MP and IMP.... all hands played, tournaments or club rooms.

Suppose the players profile displayed the IMP won/loss per board average AND the MP % average for the past month. The data is available already and it would be a simple programming chore to retrieve the data from the archives and display it on the players profile.

If would not even be necessary to update the averages displayed in real time. It could be done as a batch processing job say once a week.

I know win/loss records are not always a true indication of a players skill but in the long run better players win and poorer players lose. Also "N/A" might be displayed for players that have not played sufficient number of hands at MP or IMP to get a meaningful average .. say 50 or more hands over the last month for an average .. display "N/A" for players with less hands played.

Consider pretty new players learning the game with a true expert. The new guy makes mistakes like hell, the expert doesn't but gets his poor scores recorded in myhands.

This gives me a new idea: you might call in GIB to judge what happened at the table, and get a GIB-score (if that is possible). Problem again is the bidding systems: if I'd play moscito, GIB would give my auction a 0%. But perhaps the line of play could be analysed, which is also quite a good indication of a player's skills.

[whereagles]

Experts quite often make blunders. It takes a lot of concentration to make few mistakes.

[keylime]

Bob Hamman stated, "An expert makes 2 mistakes on every hand.". I firmly believe that.

[whereagles]

keylime, on Oct 8 2006, 10:20 PM, said:

Bob Hamman stated, "An expert makes 2 mistakes on every hand.". I firmly believe that.

A bit of an overstatement. But 2 mistakes per session is not uncommon, especially against tough opposition.

[Echognome]

Let me say why this is a silly idea. I won't claim to be of any status, yet I think I'm a decent player. I sometimes play with very capable partners and score quite well, except for when we play against star opponents where sometimes we do well and sometimes poorly. Other times I play with juniors I know because I enjoy playing with them, even if they are not very good. Yet when I play with certain juniors that are still learning the game (or trying a new system), then we won't do as well. Should I have any incentive not to play with them? I think the incentive should rather be to improve the game encourage young players to play. Self ratings are not ideal, but IMO they are much better than some "objective" criterion of imp scoring, etc.

[Winstonm]

Perhaps a better solution is to reduce the categories available to simply: World Class/ Non-World Class.

World Class would be easier to manage, as the names of those who have represented their countries in World events are recorded somewhere - everyone else falls under the next umbrella.

[Free]

Winstonm, on Oct 9 2006, 12:00 AM, said:

Perhaps a better solution is to reduce the categories available to simply: World Class/ Non-World Class.

World Class would be easier to manage, as the names of those who have represented their countries in World events are recorded somewhere - everyone else falls under the next umbrella.

We already have that: stars or no stars

[golfacer]

Wayne_LV, on Oct 8 2006, 05:43 PM, said:

Suppose the players profile displayed the IMP won/loss per board average AND the MP % average for the past month.  The data is available already and it would be a simple programming chore to retrieve the data from the archives and display it on the players profile.

I think it would be a bad idea, for several reasons (listed in no particular order):

1. The averages are not often a reliable indication of a player's skill level. One who plays in team matches vs. stars would be compared evenly to one who plays in the main club vs. beginners. By the time you adjust the averages for location of play and ability of partner/opponents, you would have the rating system that so many people do not want.

2. I think a meaningful sample size would be much, much more than 50 boards. A very large percentage of BBO players would not reach that minimum over a one-month period.

3. It would encourage many people to act unsocially to "protect" their averages. It would give players more incentive to act rudely, leave/quit, and cheat. However, maybe one can make the case that it is better to know as soon as possible if your partner is capable of being rude.

4. It would encourage more people to use several BBO names, whether to "reset" one's averages or to use one with some partners and one with other partners, or other reasons.


It would also not seem to be right for random people to judge one's skill level over several boards. The "votes" are only as accurate as the people who are voting; again dependent on one's partners. There is also a sample-size problem here: not enough boards played in one session for one to vote accurately, and not enough different people who would be able to vote. I can imagine people receiving a negative vote for taking a 90% line of play and going down in a contract instead of taking a standard 50% finesse, or if someone opens 1NT with a 5-card major and gets a bad result. The biggest difference between using this system for bridge and for Ebay is anyone is capable of buying something and seeing if it arrives quickly and in one piece.


Then what is a good solution? Sorry, but there does not appear to be one. I think the proposed ideas would cause many additional problems and not solve anything. I think seeing one's self-assigned rating can be somewhat useful, if not taken literally.

[DrTodd13]

A while back, I posted my implementation of a peer-reviewed ratings system. In it, I weighted someone's rating of you based on the number of boards you had played together up to some maximum. So, if you played 3 boards and quit then your rating would count only 1/30th as much as someone who had played 90 boards with someone. I think this takes care of the problem of somebody getting pissed and leaving. If everyone who plays with you gets pissed and leaves after 3 boards then you really must be terrible or an asshole or both and as such their rating of you is fair.

[Wayne_LV]

   

[Gerben42]

These statistics make no sense since I and many others usually play with their friends who are of the same level. Also some players can cope well with a mix of partners and will score well with strangers and others cannot. So your statistics are completely useless, I'm afraid.

[FrancesHinden]

Free, on Oct 9 2006, 12:08 AM, said:

Winstonm, on Oct 9 2006, 12:00 AM, said:

Perhaps a better solution is to reduce the categories available to simply: World Class/ Non-World Class.

World Class would be easier to manage, as the names of those who have represented their countries in World events are recorded somewhere - everyone else falls under the next umbrella.

We already have that: stars or no stars

Some stars are not world class players (under the strict definition of having been selected to represent their country in a world championship).

Some world class players are not stars (for all sorts of reasons).

The current definition of an expert is perfectly valid: "wins national events". People call themselves expert who haven't won national events (or won a non-expert event or something) but any scheme will make it possible to pervert the results.

[Walddk]

Expert: Someone who has enjoyed success in major national tournaments.

This definition, as stated in Rules of this Site, may be flawed, but you don't get much closer than that in my opinion. As Frances points out, we have loads of members who rate themselves as experts although they haven't won anything significant.

Roland

[paulg]

Walddk, on Oct 9 2006, 11:07 AM, said:

Expert: Someone who has enjoyed success in major national tournaments.

This definition, as stated in Rules of this Site, may be flawed, but you don't get much closer than that in my opinion. As Frances points out, we have loads of members who rate themselves as experts although they haven't won anything significant.

Roland

I'm a lot more relaxed about my self-chosen expert rating now that I am living in Scotland. Although I've had my moments in England, it's a lot easier to achieve success in major national tournaments here and play against, and with, the world-class players of the Scotland team.

So size does matter, but I don't really care what people call themselves as you can quickly tell. What I do care about is seeing BBO kept free of any rating system.

Paul

[Mr. Dodgy]

Agree BBO should not adopt a statistically based rating system, and almost all of the reasons given previously in this thread.

That's despite being quite proud of the little '4' in the corner of my profile.

And a mad fan of stats and stuff. There are, currently, 6714 hands in 'my' myhands database, and I have every partner rated LOL.

If you want to track these things, you can. If you REALLY want to track these things, get BridgeBrowser (not a paid advertisement). Statistical analysis, however, will be flawed as others have noted.

I agree entirely there are probably too many A++-rated players. I have from time to time 'checked' another player's myhands and been none-too-surprised to find the 'expert' in question had worse figures than mine. That doesn't PROVE anything, of course, although my figures aren't great LOL.


Regards,


Justin