Matchpoints or cross-IMPs?

[VixTD]

I play and direct at a large club that is lucky enough to be able to run games every evening (barring Sundays and some Saturdays), and also a couple of afternoons a week. As the members are spread so thinly among the many sessions, some games are down to four, three, or even two tables. If I am directing a club game and only six or seven pairs turn up, I try to persuade the players to agree to score by cross-IMPs, as there is then more chance for your score to "count". They usually agree, after I've explained how the scoring works and how they should adjust their game.

 

Other directors in the club have taken this up as well, and now the committee has decided for a trial period to score all games by cross-IMPs if there are fewer than eight pairs (although the TD may choose not to if they think the players don't want to or don't understand).

 

We've discussed this policy in the club committee and at a directors' meeting (tournament directors', that is - it's not that sort of club). One of the TDs thought it would be a shock to the players, as many of them have never played teams, rubber, or any IMP-scoring game. Another wanted to know if there was any study or analysis to show that IMP scoring was "better" when there are few score comparisons.

 

What do your clubs do? Does anyone have any opinions on which scoring system is "better" (and what "better" means), and whether the number of comparisons makes a difference?

[manudude03]

When I was in Lancaster, we occasionally had an IMP game when there were 2 tables. My general experience here in Northern Ireland is that people simply don't care. Most of them don't even understand how matchpoints work anyway.

[mgoetze]

I think matchpoints is "better" (meaning the better players win more often) for any number of tables. At IMP pairs you are always going to be at the mercy of who your opponents are etc. on the boards that "count", whereas at matchpoints all the boards count equally.

 

Incidentally it strikes me that there should be special movements for very few tables - if you have 4 pairs, then of the boards you play against pair A, half should be with pair B sitting in your direction and half should be with pair C sitting in your direction. Does anyone actually do anything of the sort?

 

Of course the best thing to do with 4 pairs is to see whether they can agree to divide themselves up into 2 teams.

[StevenG]

I prefer MPs. At cross-IMPS, wild results produce significant randomisations - and there are a lot of wild results in a club game.

 

You sometimes see results something like 5 pairs getting -3 IMPs for the par score, with one pair getting +15 when their opponents self-destructed. 5 x 40% and a 100% is less unfair. As manudude says, the average club player doesn't understand scoring anyway.

[WellSpyder]

I sympathise with the rationale behind the OP. I always feel that match points can be pretty random when you only have a couple of tables in play. But at others point out, the same is true of X-imps.

 

Which is better? Well, consider the case of just two tables. How often would you want a teams-of-4 match to be scored by point-a-board (BAM) rather than imps? It might be interesting occasionally for variety, but everyone seems happy to accept imps for teams matches. So why would you necessarily want to do anything different when comparing with 2 other tables rather than just one?

 

To answer the OP about what clubs actually do, in my experience they generally stick to the announced scoring method, whether that be imps or mps. But with fewer than 3 tables, I would expect them to switch to imps anyway, and I have certainly seen this done.

[campboy]

I think mgoetze is probably right that the better pairs win more often at MPs, but in normal club events I don't think that should be the primary criterion. I strongly prefer to play some form of IMPs with two or three tables, because I find it annoying to play/defend a contract carefully and then discover it was a waste of time because something silly at the other table(s) meant my MP score was determined before the opening lead.

[Stephen Tu]

I think mgoetze is probably right that the better pairs win more often at MPs, but in normal club events I don't think that should be the primary criterion. I strongly prefer to play some form of IMPs with two or three tables, because I find it annoying to play/defend a contract carefully and then discover it was a waste of time because something silly at the other table(s) meant my MP score was determined before the opening lead.

 

What, at IMPs you never had your IMP score determined before your opening lead, and found out that your careful play for the overtrick was irrelevant because teammates went for or took 800 or something like that?

 

This thing happens at both forms of scoring. At IMPs you have the extra luck factor of one total luck 50-50 slam board wiping out the result of many other boards. You outplay the opps on 5 straight boards but they win the game because they guessed right on the sixth.

 

MP reduces luck factor, period. IMPs is kind of a compromise between old fashioned total point scoring and point-a-board. I think it is popular for teams because it reduces the luck factor of big boards, but there is still enough luck factor to get lesser teams a chance to win. If it was all BAM, better teams would win even more consistently than they already do, and weaker players would get more frustrated from never winning.

[mgoetze]

Which is better? Well, consider the case of just two tables. How often would you want a teams-of-4 match to be scored by point-a-board (BAM) rather than imps? It might be interesting occasionally for variety, but everyone seems happy to accept imps for teams matches. So why would you necessarily want to do anything different when comparing with 2 other tables rather than just one?

If you're playing teams, your opponents bid the thin game that makes and your teammates don't, you chose the wrong teammates. If you're playing IMP pairs, your opponents bid the thin game that makes and the other table doesn't, you got fixed.

[Cyberyeti]

A somewhat weird solution we used to use in a now defunct club I used to play in for many years was this: (we used to get 6 or 7 pairs most weeks)

 

Play teams, IMP score with the two pairs sitting in the opposite direction to you at the other tables.

 

If you get a plus and a minus or a flat board and anything, score zero

 

If you get two plusses or two minuses, score the smaller gain/loss.

 

This tends to mean that if you step out of line for good or bad, you keep your score.

[campboy]

What, at IMPs you never had your IMP score determined before your opening lead, and found out that your careful play for the overtrick was irrelevant because teammates went for or took 800 or something like that?

It happens occasionally, yes. But even an overtrick when there is a big score at the other table can often gain an IMP. Making your contract rather than not will almost always gain. I don't mind the play turning out to be irrelevant occasionally, but with two- or even three-table MPs I find it happens far too much.

[mycroft]

Vix: please double all my numbers to translate from American to English (although you already knew that).

 

I hate "top on a board is 1" when I'm not playing BAM (and therefore get to choose my "teammates"). A 2 top isn't much better. But IMPed pairs is even more determined not just by your opponents at the table, but by your "teammates" - and with only one or two comparisons, there's not much "field" for you to get support from. I think you have "choice of bad options" - that's just the nature of small games.

 

Disclosure: I won a three-table MP game on Friday. Almost certainly sheer luck, as we were both asleep. I will say that I had zero confidence in my estimations, which was reasonable, as I had estimated 10% low...

[Stephen Tu]

It happens occasionally, yes. But even an overtrick when there is a big score at the other table can often gain an IMP. Making your contract rather than not will almost always gain. I don't mind the play turning out to be irrelevant occasionally, but with two- or even three-table MPs I find it happens far too much.

 

What would you rather have:

1. Your superior play save you 1 imp so that you lose 12 imp on a board rather than 13. Yay, your play on the 10th board actually mattered. However this board wipes out your carefully gained 1 trick better play on 9 previous boards, so you are now below average on these 10 boards. This 10th board counted more than the previous 9 combined.

 

2. Score the same boards BAM, now you have 9-1 advantage, running 90% game. But it turned out your play on the 10th board didn't matter.

[Phil]

1. X imps is terribly complicated for a club player.

 

2. People get depressed seeing negative imps but 46% won't faze them.

[campboy]

What would you rather have:

1. Your superior play save you 1 imp so that you lose 12 imp on a board rather than 13. Yay, your play on the 10th board actually mattered. However this board wipes out your carefully gained 1 trick better play on 9 previous boards, so you are now below average on these 10 boards. This 10th board counted more than the previous 9 combined.

 

2. Score the same boards BAM, now you have 9-1 advantage, running 90% game. But it turned out your play on the 10th board didn't matter.

In a club event, if you think it's going to be anywhere near as few as 1 board in 10 where the play is irrelevant, or as much as 1 board in 10 where you're booked for +/-12 whatever you do, our experiences are clearly quite different. I find there are lots of boards where your play might make the difference between two smallish but positive numbers (or two smallish but negative numbers) at IMPs and is a waste of time at BAM. There doesn't have to be a big number at the other table to make the play irrelevant -- 200 on a partscore board, for example, which could still see your play changing -3 to -7 at IMPs.

 

Of course there is more luck involved at IMPs as to where the big swings fall, but I'm happy to take my chances. As I said in my first post, I don't think maximising the chance of good pairs winning should be the primary aim of club events.

[Stephen Tu]

In a club event, if you think it's going to be anywhere near as few as 1 board in 10 where the play is irrelevant, or as much as 1 board in 10 where you're booked for +/-12 whatever you do, our experiences are clearly quite different. I find there are lots of boards where your play might make the difference between two smallish but positive numbers (or two smallish but negative numbers) at IMPs and is a waste of time at BAM. There doesn't have to be a big number at the other table to make the play irrelevant -- 200 on a partscore board, for example, which could still see your play changing -3 to -7 at IMPs.

 

So basically you are only concerned about whether play makes a small difference of a few imps on each board, even though as a fraction of the total IMPs available on the evening it becomes irrelevant to the large swing boards. So it makes a difference on your score for the board, but it makes very little impact on your placement for the evening.

 

In BAM, even though it may turn out your play didn't make a difference because "already won/lost at other table", at the time of the play, *you don't know that*. So you are forced to concentrate for every trick, because for a lot of boards it most certainly will matter. Don't tables reach the same contract on a fairly decent percentage of the boards?

 

If you want superior card play to count as a significant part of your overall score, you definitely want to be playing BAM not IMPs. If you want the play to simply have mattered after-the-fact on as many boards as possible, even though the overall placement is mostly unaffected, then I guess you are right to prefer IMPs. To me that's weird. But whatever.

[barmar]

If you're playing a 7-board Swiss Team match, one swing board can totally decide the match. I played a regional Swiss a couple of weeks ago, and we won our matches against the far superior teams, and lost against some teams I would normally have expected to beat -- even GLMs sometimes make mistakes that cost them 10 IMPs.

 

But in an IMP Pairs club game, with 24-26 boards, a couple of bad swings is less likely to ruin your whole session. And X-IMPs gives you a little field protection (although not much with only 4 tables), you might not be the only one making that mistake.

[Siegmund]

I try to persuade the players to agree to score by cross-IMPs, as there is then more chance for your score to "count". Another wanted to know if there was any study or analysis to show that IMP scoring was "better" when there are few score comparisons.

<br style="color: rgb(28, 40, 55); font-size: 13px; line-height: 19.5px; background-color: rgb(248, 248, 248);">

There isn't, there is, and it isn't.

 

Whether IMPs or matchpoints, only comparing 2 results leaves twice as much variation in the final scores as comparing a large number of results does. More comparisons is better.

IMPs and matchpoints remain different games, in the same way, whether there are 2 comparisons or many.

[helene_t]

I think mgoetze is probably right that the better pairs win more often at MPs, but in normal club events I don't think that should be the primary criterion. I strongly prefer to play some form of IMPs with two or three tables, because I find it annoying to play/defend a contract carefully and then discover it was a waste of time because something silly at the other table(s) meant my MP score was determined before the opening lead.

I think that's the point. XIMPs is not "better" than MPs although it is probably so that the advantage of MPs (in terms of how often the better pair wins) becomes smaller as the field becomes smaller. But it is a psychological thing.

 

Playing XIMPs in a small field you won't know if your overtrick matters or not but I don't think that is so bad because you know you are playing IMPs so you shouldn't be too woried about overtricks anyway.

 

If you want everything to matter you may prefer to play Total Points but then you have the issue that you can be down 5000 points before the last round so you know that it doesn't matter even before you pick up your cards.

[SteelWheel]

Here in ACBL-land, nothing about scoring any bridge game makes much sense to me.

 

For example, in local club games, the idea of scoring via any method other than matchpoints is virtually unheard of.

 

However, in most Regional tournaments, the (by far) dominant method of scoring for nearly all events is by imps (not x-imps, but imps nonetheless). And why is this? Because the product that ACBL sells is masterpoints.

 

Hence, nearly event is a "bracketed round-robin Swiss", where they just start cutting the field into seven or eight teams per bracket, and have each team play against all the others. After all, it awards more masterpoints, since each bracket rewards those precious gold points to at least three teams.

 

Heaven forbid that any pair play their best all day long in a matchpoint event and have "nothing to show for it". Must award gold points!!

 

I'm a lonely voice, I guess. I became a Life Master a long time ago, and it wasn't all that tough to get there, even under those old, draconian rules (y'know, the ones that required that you be able to beat LMs in order to become one?).

 

Modern tournament bridge is so deathly dull with the heavy emphasis on formats which will award the most masterpoints possible (NABCs mostly excepted). But hey, where is someone ever supposed to learn how to play matchpoints (vs a strong field), or IMP pairs, or (gasp!) board-a-match, if the clubs and the non-national tournaments don't put such events on the schedule?

 

ok, rant over. Go back to your normal business...

[helene_t]

In Lancaster I played in the same club as Manudude but other than that, XIMPs is rare in the UK.

 

In the Netherlands, some clubs have a shedule with say mathcpoints in the first week of the month and XIMPs in other weeks. Most people don't care but there are some who prefer XIMPs because they see the club evening as training for team events which are almost always IMP scored.

 

I sometimes play at a club that uses patton scoring but that is very rare, I think there are only two or three such clubs in the whole country.

 

It doesn't seem to be related to the size of the field.

 

I tried to make some simulation experiments to investigate the hypothesis that MPs have a bigger advantage in large fields than in small ones but I think my model was dodgy so I won't link to it here :)

[VixTD]

There isn't, there is, and it isn't.

 

Whether IMPs or matchpoints, only comparing 2 results leaves twice as much variation in the final scores as comparing a large number of results does. More comparisons is better.

IMPs and matchpoints remain different games, in the same way, whether there are 2 comparisons or many.

It's not clear to me exactly what you're disagreeing with, Sigmund, but it seems to be everything.

 

I thought it was self-evident that the fewer tables you have the fewer score comparisons there are using any across-the-field scoring method.

 

When I said that your result on a board has more chance to "count" in the final scoring method when there are few comparisons (when there's a small field) if you are scoring by cross-IMPs rather than matchpoints what I meant was something Campboy alluded to earlier. If there's only one other table in play your result has only two chances to register any matchpoints: (i) if it matches the opponents' score, and (ii) if it is greater than the opponents' score. At cross-IMPs if your opponents get an unbeatable score at least your good efforts get some reward for reducing the score deficit on the board.

[Stephen Tu]

When I said that your result on a board has more chance to "count" in the final scoring method when there are few comparisons (when there's a small field) if you are scoring by cross-IMPs rather than matchpoints what I meant was something Campboy alluded to earlier. If there's only one other table in play your result has only two chances to register any matchpoints: (i) if it matches the opponents' score, and (ii) if it is greater than the opponents' score. At cross-IMPs if your opponents get an unbeatable score at least your good efforts get some reward for reducing the score deficit on the board.

 

It may "count" in terms of your raw score for an individual board being altered. But at the end of the day, in terms of your overall score and overall ranking, these imps hardly matter at all. So it "counts", yet it doesn't really count in the end. It's like you are concerned about what comes out to rounding errors when total score for the evening is computed. And it renders a lot of boards where big swings aren't possible relatively meaningless.

 

At IMPs, your careful play to set up the double squeeze for the 2nd overtrick on board 2 counts 1/13th as much as the opps bidding slam on a finesse that happens to work on board 1. At MP, your good play has evened up the game. At IMPs you are still way behind.

 

Opps get a gift 800 on a partial board. As a good player, what would you rather have, a chance to lose 12 imps instead of 14 because your play on this board "matters", or a chance to recover the entire deficit with just an overtrick with good play on the next board?

 

IMPs basically renders a lot of boards very unimportant, because realistically only can have a 1-2 imp difference depending on how well played/defended the board is. While some boards have +/-13 imps in play, 26 imp swing vs. 1-2, sometimes simply on the result of whether a 50-50 finesse works or not. If you come up on the wrong side of that, it wipes out better bridge played on many, many other boards.

IMPs is just a board importance magnifier vs. MP. It's true that as barmar said that over 24-26 board pair session you have more of a chance to recover than over a 7 board Swiss match (more boards > fewer boards in evening out luck), but MP is way better still, as you only need one board to recover from a bad result.

 

MP is just less luck oriented than Imps. X-imp pairs is simply the most random kind of game in bridge. Teams has less luck than pairs because you have control at both tables. But Imps makes some boards worth much more than the others. It makes superior play in the form of extra overtricks/undertricks worth very little vs. what they are worth in MP. Basically from least luck to most luck is team MP > team IMP > pairs MP > pairs IMPs. Imps is just easier to play because you know what your goal is (beat/make contracts), and you don't have to stress over every single trick quite as much when it's clear the contract is making.

[Siegmund]

I think it is also self-evident that the considerations in a IMPs vs MPs decision are exactly the same in any size of field.

 

In a large field, a serious MP error will cost you the whole board and a small error will cost you a few matchpoints.

In a 2-table field, a serious MP error will almost always cost you the whole matchpoint and a small error will sometimes cost you the whole matchpoint . The relative sizes of the two errors remain the same, in terms of expectation, it is just harder to see when you only have one comparison.

 

Similarly, you will see some boards at IMPs that cost you 0 IMPs on a good day and 12 IMPs on a bad day when there is only one comparison, but cost you about 6 when there are a bunch of comparisons and half the field beats the game.

 

Maybe your players like IMPs in small fields... I just don't like small fields, and shake my head in amazement at the popularity of team games with only 2 comparisons when a lot more are available.

[barmar]

In a large field, a serious MP error will cost you the whole board and a small error will cost you a few matchpoints.

In a 2-table field, a serious MP error will almost always cost you the whole matchpoint and a small error will sometimes cost you the whole matchpoint . The relative sizes of the two errors remain the same, in terms of expectation, it is just harder to see when you only have one comparison.

That's what we call "field protection". In a large field, you can expect that you'll have some company in a small blunder. If the hand hinges on a 50-50 guess, you can expect half the field to guess right (scoring 75%) and half wrong (scoring 25%). But in BAM, you either get 100% (you guess right, opponent guesses wrong), 0% (vice versa), or 50% (you both guess the same) -- there's no field protection to smooth out the extremes.

[barmar]

IMPs basically renders a lot of boards very unimportant, because realistically only can have a 1-2 imp difference depending on how well played/defended the board is.

I think IMPs tends to be more oriented towards accurate bidding (especially game decisions), while MP is more about taking the maximum number of tricks in whatever contract you land in. The hands that tend to be unimportant at IMPs are the ones that are really easy to bid, since most pairs end up in the same contract,