US Trials for Turkey

[jdonn]

This really seems simple. Matchpoints is more accurate at choosing the best matchpoint pair than imps is at choosing the best imp pair (in the same amount of boards). I agree with that. But matchpoints is NOT more accurate at choosing the best imps pair than imps is at choosing the best imp pair. Unprovable I know, but I would stake anything on it.

If anyone can come up with a method to validate this claim (this isn't impossible, given enough data) I am happy to bet that a 3-session MP event is better at determining the 6 best IMP pairs than a 3-session IMP event.

Then why has this not always been done?

Because usually there is more time than 3 sessions to select two teams?

I think 4 sessions has been extremely common in the past. Is 3.5 sessions the cutoff where imps are a better selection method?

 

Frankly I don't see why they don't do more than 3 sessions now when playing online should allow for more sessions than usual instead of fewer, but I'm sure there must be a reason.

[hrothgar]

This really seems simple. Matchpoints is more accurate at choosing the best matchpoint pair than imps is at choosing the best imp pair (in the same amount of boards). I agree with that. But matchpoints is NOT more accurate at choosing the best imps pair than imps is at choosing the best imp pair. Unprovable I know, but I would stake anything on it.

If anyone can come up with a method to validate this claim (this isn't impossible, given enough data) I am happy to bet that a 3-session MP event is better at determining the 6 best IMP pairs than a 3-session IMP event.

Han / Josh

 

One way to approach this problem would be to analyze the variance in board results for (comparable) IMPS pairs and MP pairs events.

 

As we've discussed before, variance in board results negatively impacts a tournament's ability to identify the "top" pairs. In theory, if we could characterize the impact of the change in format on variance, we might be able to arrive at some ball park estimates for other parts of the model.

 

For example, we might be able to say

 

Switching from MP pairs to IMP pairs increases variance by X

 

In order to "break even", the IMP pairs format would need to be at least Y times better at identifying good IMP players than a MP pairs format.

 

If Y is very very large, stick with MP pairs.

If Y is very very small, go with IMP pairs.

[helene_t]

That's easy enough, the difficult thing is to estimate the correlation between a pair's performance at MPs and its performance at IMPs.

[mtvesuvius]

Also, what about U21s if attendance warrants? or is the USBF just planning on sending an U26 team? Also... This is trials for the Team Events... What about pair events?

[kfay]

Why not just send the team that wins the qualifiers in DC?

[karlson]

I really don't get all the griping about MP. Surely a pair that is significantly better than the rest of the field is going to be fine regardless of the format?

 

I'd be really happy with the fact that random gifts won't account for 20 imps myself.

 

By far the best aspect of having it online is that people who would never normally spend the money and time to come out to DC or whatever for the trials are going to have a chance to play. Who cares if they have no shot at winning, at least they get an idea of the level they would need to get to, and maybe some of them won't think it's a hopeless endeavor.

[olien]

Well, I think the idea of MP trials is not the way to go. As has been previously mentioned, what is the point of an MP trial for an IMPs contest?

 

Anyways, the idea of having to require people to go to a club is impractical. Who will have all day on Saturday and half a day on Sunday to watch? Family should be excluded as potential proctors because of the potential conflict of interest.

 

Maybe I'm paranoid, who knows, but the entire format and setup seems to be a farce, too many holes that people could complain about after the fact.

[JanM]

Why not just send the team that wins the qualifiers in DC?

Because that is too late an event. The DC Trials are July 22-24, the Istanbul Championship is Aug. 15-23

[mike777]

Why not just send the team that wins the qualifiers in DC?

Because that is too late an event. The DC Trials are July 22-24, the Istanbul Championship is Aug. 15-23

As a side note if the 2009 World youth championship is in August of 2009 when is the 2010/11 youth events and when are the trials? Is there enough time?

 

Should these DC trials be for 2010/11; perhaps not, just a suggestion.

[georgeac]

Are all USA juniors open to play? If so, and a junior pair that enters(myself and one of my friends) does well enough to make the team but has no right really being on the team by qualifications, what happens then. It seems selection would be better than having the chance that some unqualified juniors make up the final team.

[whereagles]

Interesting

Interesting indeed. Matchpoint scoring to select the pairs for a teams event!? I am lost.

 

Roland

Welcome to Portugal. We got that as well in a stage of the team trials. Absolutely ridiculous.

[PeterGill]

I think the over-reaction to the use of matchpoints is backed up more by emotional reactions than by actual facts, especially by some younger posters.

 

The Australian Youth Pairs is the curtain raiser of our annual Youth Bridge Week, which has been held every January for 41 years. Saturday is 2 sessions of matchpoint Pairs Qualifying, then 2 sessions of the Pairs Final (match pionted) and Consolation on the Sunday, followed by the barbecue and cricket match.

 

I just looked through the results of the Youth Pairs from 2001 to 2009 at

http://www.abf.com.au/youth/events/ayc/pastindex.html

 

The top-scoring six or seven pairs in the Final in each of these nine years

are consistently among the best ten pairs in the event, with the top-scoring

three pairs usually being among the best five pairs in the event.

 

This data strongly suggests to me that matchpointed yotuh events do

sort out the best players efficiently in a quick time. I think it takes a

(much) longer time for imp pairs events to sort out a youth field efficiently.

 

I would expect similar results if I did a similar analysis of the World Junior Pairs, which also is matchpointed.

 

Peter Gill

Australia.

[karlson]

Are all USA juniors open to play? If so, and a junior pair that enters(myself and one of my friends) does well enough to make the team but has no right really being on the team by qualifications, what happens then. It seems selection would be better than having the chance that some unqualified juniors make up the final team.

Why would you think you're unqualified?

 

The answer is, you have a great experience and maybe next time won't feel like a fluke.

[JanM]

Why not just send the team that wins the qualifiers in DC?

Because that is too late an event. The DC Trials are July 22-24, the Istanbul Championship is Aug. 15-23

As a side note if the 2009 World youth championship is in August of 2009 when is the 2010/11 youth events and when are the trials? Is there enough time?

 

Should these DC trials be for 2010/11; perhaps not, just a suggestion.

The DC Trials are for 2010. We do not yet know where and when the 2010 World Junior Championships will be held.

[georgeac]

Are all USA juniors open to play? If so, and a junior pair that enters(myself and one of my friends) does well enough to make the team but has no right really being on the team by qualifications, what happens then. It seems selection would be better than having the chance that some unqualified juniors make up the final team.

Why would you think you're unqualified?

 

The answer is, you have a great experience and maybe next time won't feel like a fluke.

maybe i worded it poorly. it was late last night. anyways. maybe not that i am underqualified but shouldnt be the point of the trials be to send the best possible teams with a chance of winning. i know there are lots of juniors that are much better than i am. so shouldnt they be the ones to go to have the best chance of winning the event? just saying that one pairs trials doesnt seem like the best way to do this. with the time constraints though i guess its the only way for this year.

[JLOL]

So Jan is it confirmed we have to go to a club to do this?

[JanM]

It is confirmed that each player will have to be monitored. If the monitor isn't a director at a recognized club, the arrangements will have to be approved by the USBF Junior Committee.

 

The monitor's responsibility will be to confirm that the player did not use a cell phone and did not have any Instant Messaging or Email programs running on his or her computer. Hopefully, the Conditions of Contest will be more precise than this, and may include a requirement that no program other than BBO is running on the player's computer.

[mike777]

'may include a requirement that no program other than BBO is running on the player's computer.'

 

 

This cannot be correct.

[Lobowolf]

The top-scoring six or seven pairs in the Final in each of these nine years

are consistently among the best ten pairs in the event, with the top-scoring

three pairs usually being among the best five pairs in the event.

 

This data strongly suggests to me that matchpointed yotuh events do

sort out the best players efficiently in a quick time. I think it takes a

(much) longer time for imp pairs events to sort out a youth field efficiently.

 

I would expect similar results if I did a similar analysis of the World Junior Pairs, which also is matchpointed.

 

Peter Gill

Australia.

Not sure I get the logic here. Matchpointed adult events sort out the best players pretty efficiently, too, but I can't think of any major IMP events offhand that sort out qualifiers by matchpoints.

 

The Spingold gets some awkward number fields; perhaps 6 sessions of pairs, averaged between teammates, then the best 8 teams play knockout matches?

[PeterGill]

I meant that if the rather short World Junior Pairs was used to identify the

best youth pairs in the world at imps, I expect the process would be valid.

 

Relative to adult events, youth events scored by IMPs can have unusual

variations, because youth pairs are more likely to give up and throw imps

all over the place (running youth events for decades allows me to make that

statement with 100.0% confidence), for example giving a pair 50 imps in

5 boards (some effect, but less effect, at matchpoints) and the relative wildness

of youth bidding can introduce extra randomness factors more so to youth imps events than youth matchpoint events.

 

I'm not claiming that I would use matchpoints for this particular event -

I'm just presenting some data based on my experience.

[jdonn]

Relative to adult events, youth events scored by IMPs can have unusual

variations, because youth pairs are more likely to give up and throw imps

all over the place (running youth events for decades allows me to make that

statement with 100.0% confidence), for example giving a pair 50 imps in

5 boards (some effect, but less effect, at matchpoints) and the relative wildness

of youth bidding can introduce extra randomness factors more so to youth imps events than youth matchpoint events.

Well that's nice.

 

"Since my anecdotal data tells me we can't trust you to try to win all the time because you are so immature, we are going to treat you like babies and change the form of scoring to prevent your results from greatly impacting who wins."

 

If a pair is way behind, they may rightfully try to swing. Just like in the real event. Playing well against pairs who do that is a skill that is valuable in the real event, as is actually doing it yourself. It's just one of a number of skills valuable at imps that matchpoints doesn't measure well (it's not all about safety plays).

 

I have no doubt there would be a very strong correlation between who does well in matchpoint and imp events. But of course, that is not the point. It's just so silly for someone to claim (not that you are necessarily are) that matchpoints is a better indicator of skills at imps than imps is, unless they want to admit that essentially every country in every event ever around the world has done a lousy job of picking their teams but suddenly the USA has seen the light.

 

I don't understand why suddenly there is such a motivation to remove randomness from a junior event. If they play randomly, then randomness is part of the game. With just one notable exception, has there been a single team that is not the best available or at least very close that has won in the trials due to excessive randomness? It's the exact same thing as the cell phone ban. Let's focus on problems that are rampant rather than perceived problems that don't exist on a large scale, especially when the solution creates a perception of unfairness or lost credibility.

[mycroft]

Josh's comment about "suspicious" was exactly what I was thinking...

 

How many non-youth can throw out the score for 4Cxx+1 off the top of their head (strangely enough, I had to score that one yesterday, for the first time in my life. I'm getting old)? I recall a Rigal story in an NABC bulletin a while back where that result occurred, and the defender bounced out the result right away. "Trust a junior to know that" was the gist of his comment.

 

I think that a 3-session MP pairs event will generate a really good set of pairs. Especially if (vice the old thread), they can (being the US) qualify 6 pairs in two teams, as they can settle out who's on what team afterwards. Will they potentially miss a heavily IMP-skewed pair? Sure. Might they conversely have a "MP is god" pair qualify, and have them continue to play MPs in Turkey? Sure.

 

I don't think it's the best way of doing things - I *really* dislike it, in fact - but given the costs of anything better, it might be the least worst option.

 

[Edit:]I can't imagine anyone reasonable getting five full tops from a random, swinging-to-qualify pair at MPs not qualifying themselves. And it's not just the juniors - didn't we just have the German pair in Beijing swinging to qualify, and dropping 60 IMPs in half a session?

 

Again, if skewing the results by swinging to qualify and failing is a problem, fix the Conditions of Contest.

[brianshark]

1. You can't take a tougher stance on "suspicious activities" because it makes it more likely you'll wrongfully accuse innocents. In fact, it's even harder to identify cheating in junior events as compared with top class open events because of the chance of more random actions, some of which will work even if a lot of them don't.

 

Just moniter the players and you won't even need to worry about cheating. You won't even need to ban kibs because the players are being watched so can't be on the phone to their buddy in the gallery.

 

2. Matchpoints is a totally different game than imps. I'm sorry but if you think the only difference between the 2 is the importance of making overtricks in MPs and the importance of making safety plays in imps, then you don't fully appreciate the difference between them. A lot of the decision making is very different.

 

For example: In situations where you are competing for the part-score and are vulnerable, it often pays to bid 1 more in certain situations knowingly risking 500 or 800 because it's only a bottom whereas a numerically significant amount of the time, your bid will work because you make, or because they don't double you, or didn't matter because them bidding to the level they did was going to be a near top anyway. You would never dream of overcompeting in some of these situations at imps because the size of the potential imps gain would be small and the size of the potential imps loss is great. And similarly for doubling tight part-scores the opponents have bid when you are sure that if they make they get a good score anyway.

 

And undoubtedly, you should be testing the imps decision making abilities of the trialists, not the matchpoints decision making.

 

In my opinion:

- Moniter the players but take no additional action towards cheating than you would in any other context.

- Play imps. Matchpoints is ludicrous.

- If you are worried about variation, pre-duplicate the boards to get a reasonable datum to remove a certain amount of the random element while mentaining the imps pairs format.

[awm]

I think there is a problem being addressed here. The goal is to simultaneously send a good team and give every strong player a reasonable chance to go.

 

The options seem to be:

 

(1) Just pick a team. This will end up with a good team, but there are potentially lots of strong players who don't know the right people and will end up with no chance to go.

 

(2) Run a team event of some sort to select the team. However, this suffers from some of the same problems as the first, because there is a core group of strong players which includes most but not all of the best competitors. This is basically the remnant of the team from previous years, and they will pick people who they like and respect to add to their team. While there might be some good players out there who they don't know well or mis-estimate the skill for, there probably isn't an entire team of six like this, and even if there were they might not all know each other etc. So the team trials ends up with one really strong team, and a bunch of teams with one or two players who might potentially be better than some members of the really strong team, but can't carry an entire team of four-six to beat the strong team.

 

(3) Run a pairs trials. This gives everyone a chance to go, but especially if it's a short format there is a lot of randomness and a good chance that a weaker team is sent.

 

Obviously none of these ideas is really satisfactory. The best idea seems to be running an IMP pairs over a very long period, but it seems to be determined from on high that the qualifier cannot take more than three or four sessions. USBF has certainly done the "pick a team" and the "run a team event" things in the past, and these have both resulted in strong teams at least most of the time; however the junior program is really in need of "new blood" at this point with a lot of the top players who have won championships in the last few years now being too old. Basically the problem of "pick a team" is that we end up sending mostly the same people every year, and now that a lot of them are over the age restriction it is much less clear who to pick than it was in previous years.

 

Certainly it is true that MP pairs has some different features than IMPs. The strategy in both bidding and play is a little different. But take a look at Justin's blog (www.justinlall.com) -- a lot of the best "MP pairs" just play it like IMPs with very minor adjustments! And if we look at the winners of major IMP events, most of them have had success at MP too (and vice versa). So it's not like we are telling the juniors to play poker to decide a bridge team -- IMP and MP are two very similar events. Yet MP has the advantage of being far less random than IMP pairs over the short term. Suppose you bid a slam on one board that's on a finesse where most of the field doesn't get to slam. At IMPs, this one board is going to swing something like 26 IMPs (win 13 if it makes, lose 13 if it fails). That is often the difference between first overall and not even making the top ten or twenty. Do we really want our team to be potentially decided by one random swing on one board?

 

In fact the one year that the "best team" didn't win the team trials, something very similar to this actually happened. And this is much more possible at IMP pairs than at teams as well, because you don't have your teammates to rely on (instead the "datum" is generated by the field which probably includes a lot of weak pairs especially early on).

 

It also makes sense to do things in different ways for different events. If we "pick a team" every time then probably the same team goes to every event. This might win some events if it's a good team, but once the players "graduate" from youth status we'll be left with no replacements who have substantial international experience. And it also seems unfair to youths who might be good players but just aren't part of the right cliques and never get a chance to go to anything. If we run a pairs trial every time, we end up with more random and potentially weaker teams. But it makes sense to say that every few events we want some "new blood" and we run a pairs trial. Even this year, we have several different youth events upcoming and I think this one is the only "pairs trial" event.

[jdonn]

I think Adam makes a very good argument for simply picking the team without even realizing it, as long as I refute the negatives. There isn't worry that strong players wouldn't be discovered because if they are strong enough to matter then they have some non-junior results to show for it. And there isn't worry that the players of the future wouldn't get enough international experience because it's just not true. I had gone to Europe four times to play bridge before I was on the team, turning down several other opportunities along the way. And at the start of that I wouldn't have even been a particularly strong candidate as a future team member.