[awm]

Since the UK uses the NGS to rate players and pairs, there must (implicitly) exist a rating of games / clubs (so as to compute the appropriate adjustments). Is this made available anywhere? I can see a number of uses, for example if someone wants to pick a difficult club evening to challenge themselves, or an "appropriate to level" event, or wants to compare the caliber of opposition in online vs. offline play. I could also see some reasons the EBU might not want to make this information public of course.

There's been a lot of discussion in another thread about the Friday evening game at Young Chelsea being atypically strong, and I'm curious as to whether there are metrics to back this up (in particular since Young Chelsea was not the bridge club that had been recommended to me in case I ever go to a club in London).

[helene_t]

I don't think there is a rating of clubs, there is a rating of fields which within the club fluctuates from evening to evening. Mitchell-movements are very popular so there may also be different strengths of the EW and NS fields on a given evening.

I suppose one could calculate the average strength for e.g., wednesday night at a given club, but I donøt think that is published unless the club decides to do that themselves.

[shyams]

https://www.bridgewe...id=display_past

At the bottom of each result page, one can find the NGS ranking for the regular participants.

I looked at the past 4 Friday evening games --- the NGS for the participants was around 59% on all four Fridays.

[awm]

 shyams, on 2025-April-01, 05:57, said:

https://www.bridgewe...id=display_past

At the bottom of each result page, one can find the NGS ranking for the regular participants.

I looked at the past 4 Friday evening games --- the NGS for the participants was around 59% on all four Fridays.

I saw this too, but I don’t have a good baseline for what the number means. I tried looking at a couple other clubs to compare but they didn’t seem to have the NGS number included in the results.

[shyams]

This is an extract from a PDF on the EBU website (https://www.ebu.co.u.../full-guide.pdf)


Page 5: NGS Grades

An average strength player therefore has a grade of about 50%, but there are no limits to the values of grades. The strongest players have current grades of over 60% and the weakest under 40%. The spread of grade values has a bell-shaped distribution. (For the statisticians, it’s approximately Normal with a standard deviation of around 5 to 6%.)

Using my limited knowledge of stats, if a field is 59% and the Std deviation of grading is (say) 6%, then the quality of the field at YCBC is in the top 15% of all EBU clubs.

[helene_t]

shyams, on 2025-April-01, 06:55, said:

Using my limited knowledge of stats, if a field is 59% and the Std deviation of grading is (say) 6%, then the quality of the field at YCBC is in the top 15% of all EBU clubs.

The SD across clubs will be much lower than teh SD across players, so YC may well be the strongest of all EBU clubs.

[mikl_plkcc]

awm, on 2025-April-01, 06:05, said:

I saw this too, but I don’t have a good baseline for what the number means. I tried looking at a couple other clubs to compare but they didn’t seem to have the NGS number included in the results.

The NGS Grade is the expected matchpoint percentage when placed in a field of a matchpoint duplicate across the whole country. For example, a player ranked 59% is expected to get 59% in a matchpoint duplicate played by all players in the country.

Therefore, an average player has a grade of 50%, and the numerical grades are divided into ranks of the 13 cards in 2-percent bands. Therefore, the middle card 8 represents 49% - 51%, and the top honour (A) represents 61%+. Wimbledon Bridge Club is holding a beginner / intermediate tournament where the upper grade limit is a 5!!! (My current grade is a 9 despite that I have only achieved a positive score once over my Friday evenings at Young Chelsea, and I thought that getting an honour grade would already be club-level good)

[awm]

Indeed, this suggests that a score of -1 imp/board at Young Chelsea is above average for the country as a whole! So you may just be playing in a tough game.

[mikl_plkcc]

awm, on 2025-April-01, 09:33, said:

Indeed, this suggests that a score of -1 imp/board at Young Chelsea is above average for the country as a whole! So you may just be playing in a tough game.

EBU suggests that 1 cross-IMP is approximately equal to 6.12% in an MP game, although the factor of 6.48% is traditionally used by EBU.

[mycroft]

As I mentioned (similarly) last time, the median player with a 59.00% NGS mature rating is 1244 in the EBU.

If the game averages 59% rating (assume mean here, but probably it should be close), then effectively about half the field are in the top 1000 in England. There are 24900+ players with "mature grades" (1000+ rated boards).

I've played in fields with that kind of rating. Very very rarely in a club, though. Sure, you get one or three pairs that are "national level" in the strong games, (and if they come worse than third, something's really wrong); this is a game where every second or third round is, and some of those don't make average on a random night (and "national level" in England is stronger than, say, "national level" in Canada or Mexico, or Switzerland, if you don't count the imports).

[akwoo]

Off-topic, but something that I'm interested in that probably only NGS at present provides the data to study.

All Elo-like ratings rely on what I'll call a "transitivity function" of some sort. Let's say we take a bunch of players who all average 60% in the average club field. Now split them into their own "expert game". Into this expert game comes a super-expert player who averages 55% in the expert game. What would you expect this super-expert to average in the average club field? (Presumably it's more than 60% and less than 65%)

NGS makes some assumption for the purposes of calculating adjustments when the super-expert decides to play in an average club field.

Is there a way of statistically studying how accurate this assumption is, and perhaps even statistically adjusting this assumption to be more accurate (so that this super-expert's rating doesn't vary depending on which field they're playing regularly in)?

[paulg]

The Friday night game of the Young Chelsea in the late 1970s/1980s was easily the strongest club game in the country. It was played in the smoke-filled basement of the club in Earl's Court and featured most of the strongest players in London and their visitors.

It was a very intimidating place to play as a visitor. And expensive if you were not used to London prices, but fortunately all the abuse was free. It was the place to test your game and prepare yourself for the tournament world. Considering how much alcohol was consumed in an evening, it is very surprising how high the standard was.

I think it was this era that established the reputation of the club which it maintains to this day, although times (and location) has changed and the behaviour has improved immensely.

[awm]

 akwoo, on 2025-April-01, 18:21, said:

Off-topic, but something that I'm interested in that probably only NGS at present provides the data to study.

All Elo-like ratings rely on what I'll call a "transitivity function" of some sort. Let's say we take a bunch of players who all average 60% in the average club field. Now split them into their own "expert game". Into this expert game comes a super-expert player who averages 55% in the expert game. What would you expect this super-expert to average in the average club field? (Presumably it's more than 60% and less than 65%)

NGS makes some assumption for the purposes of calculating adjustments when the super-expert decides to play in an average club field.

Is there a way of statistically studying how accurate this assumption is, and perhaps even statistically adjusting this assumption to be more accurate (so that this super-expert's rating doesn't vary depending on which field they're playing regularly in)?

I suspect there is some non-linearity when the gap between the pair’s ability and the field is sufficiently high. For example if you stuck one of the UKs top pairs in a beginner game, the NGS would suggest that they should average around 100%, which they will not achieve in practice.

Most likely the system will penalize very strong players who play in a really weak field, and boost very weak players who play in a very strong field. Of course, this might be desirable in terms of the influence on playing habits and for most players probably has little impact.

[pescetom]

 awm, on 2025-April-01, 00:53, said:

Since the UK uses the NGS to rate players and pairs, there must (implicitly) exist a rating of games / clubs (so as to compute the appropriate adjustments). Is this made available anywhere? I can see a number of uses, for example if someone wants to pick a difficult club evening to challenge themselves, or an "appropriate to level" event, or wants to compare the caliber of opposition in online vs. offline play. I could also see some reasons the EBU might not want to make this information public of course.

Getting back to the original discussion, I struggle to see why the EBU should not want to make the club rating public, or even rank the clubs. It's not as if the clubs are individuals with a right to privacy. I guess EBU might be worried that poor clubs will be more clearly perceived as such and thus be more likely to disappear, but that is not necessarily bad news if people have simply moved to a better game. Or maybe they have a problem of unaffiliated competition and are worried that poor clubs might lose motivation to remain in the system.

[awm]

While it’s possible there are business concerns for the clubs (and information that is “private to a business” likely does need careful handling), I was thinking more along the lines of gaming the system for simultaneous pairs by finding a club with a very weak field to “visit” just for the simultaneous.

It might also be embarrassing for EBU if certain prestigious tournaments turn out to have relatively low ratings. And there might be some people upset if (for example) clubs in one region turn out to be much stronger than another (I think the strength of some London clubs is well known, but what if hopping across the Welsh border gets you a much stronger game for example).

[helene_t]

In principle you shouldn't be able to game the system as you are compensated for the strength of the opposition.

One way to game the system is, I think, to play with strong foreign partners, the system can't guess the strength of someone without an EBU history so they will be biased towards the middle, or maybe biased towards your own level.

[awm]

I was not talking about gaming the NGS system (indeed, the best way to do that is playing with underrated partners, foreign players being a likely example), but rather trying to produce a very high percentage in a simultaneous pairs. Obviously for this purpose you want to have very weak local opponents who will give you top results. In fact if it’s matchpointed across the field, you do not even care about the strength of players sitting your direction — you want weak players the opposite way.

I admit that I don’t know for sure how simultaneous pairs work in the UK — do they actually try to adjust the scores based on strength of opposition? Usually in the US and Switzerland the winners of such events played in weaker clubs, although we cannot measure this as scientifically as they can in England.

[pescetom]

awm, on 2025-April-02, 10:27, said:

I admit that I don’t know for sure how simultaneous pairs work in the UK — do they actually try to adjust the scores based on strength of opposition? Usually in the US and Switzerland the winners of such events played in weaker clubs, although we cannot measure this as scientifically as they can in England.

In Italy they do not adjust scores based on the strength of the opposition (or the weakness of the pair).
The masterpoints reward is strongly biased towards those doing very well at national level, which can vary between being well placed in the top 40% in the most important tournaments and being well placed in the top 15-25% is less important tournaments.
There was a period some years ago when strong pairs would play in weaker clubs to maximise masterpoints (as well as to win easily), but nowaways they sell award masterpoints generously in ways such that this no longer has much attraction.

[mycroft]

Heh, I was once told (in court, no less!) that obviously my partner and I were expert bridge players, because we scored higher in the (then EPSON) World-Wide Pairs than "Sabine Auken, a world champion."

Yeah, because he and I played a system the opponents were unfamiliar with (Precision) in a 4-table Howell in (small city 90 minutes out of Town) in a field where even at the time, I would expect to come in second playing with any random player there - only because the people who I hitched a ride with would be first; and Ms. Auken played with a client in (very likely) the strongest game in Germany. Sure...

[paulg]

awm, on 2025-April-02, 10:27, said:

I admit that I don't know for sure how simultaneous pairs work in the UK — do they actually try to adjust the scores based on strength of opposition? Usually in the US and Switzerland the winners of such events played in weaker clubs, although we cannot measure this as scientifically as they can in England.

There is no adjustment of scores for a simultaneous pairs in the UK, that would effectively be a handicap competition. The EBU, SBU and WBU all prohibit awarding master points for handicapped events although you can earn master points for the scratch (unadjusted) scores.

I suspect my local club is typical for those who use handicaps occasionally: the unaltered scores for an evening earn master points, but the trophy associated with the evening is based on the handicapped results. Handicaps seem a lot more popular in Scotland than England from my experience.

Master points can be earned for stratified or flighted events, in a way you'd expect.