[EarlPurple]

I am sure this kind of things has been discussed before, but I have failed to find anything on it in searches, because it is difficult to know what to search for.

I am thinking of a pairs scoring that is somewhat hybrid between Match Points and IMPs. The issues with both I see as follows:

1. In MPs, there is a lot of "gambling" involved over tiny differences - having to guess between NT and major games, for example. Plus obviously a lot of emphasis on overtricks and magic scores. The plus side and what I like about this scoring is that every hand matters.

2. IMPs is good for long team-of-4 matches. For pairs tournaments though of limited numbers of boards, too much emphasis on one or two "big" boards and often out of your control if the opps bid a good slam against you missed at other tables which essentially can destroy your entire tourhnament chances.

With a hybrid scoring, it is the same as MP in essence but with slight changes. With computers scoring, it is of course easy to calculate the results of a hand.

Option 1: (simplest). Same as MP except that a difference of 10 counts as a tie. So if you score 430 you tie with scores of 420. If you score 120 in 1NT+1 you tie with scores of 110 or 130, although those with 130 still beat those with 110. You still beat anyone with 100.
You score by either winning, tying or losing against other pairs thus determining your score on the board.

Consequence on tactics: You'll play pretty much the same but won't necessarily strive for the NT games or slams, and be content with playing in the major suit instead. Of course you may suffer a ruff and find you make a trick fewer than the NT makers, but in general you'll be more content to play in the major.

A score of -100 (going down in a part-score) would tie with those making -90 in 1NT or -110 in 2 of a major or 3 of a minor.

Option 2: (more complex). As above but now we introduce a "small win" and a "big win". A small win is 3-1, a big win is 4-0, a tie is 2-2. As before, equal score or difference of 10 is a tie. A small swing is considered a small win. I think the maximum for a small swing could be considered 40 or 70.

Consequence on tactics: If you are an overtrick short you will make a "small" loss on those who made the overtrick. Similarly if you concede an overtrick trying to beat the contract. However you will gain a lot more for actually beating the contract and getting into the right "zone". You would have to weigh up the odds accordingly. You'll get 25% if you concede the extra ovetrick but 100% if you beat the contract (against others who pick the opposite tactic). In general, in an "everything equal" situation you are less likely to gamble for an ovettrick.

I think in addition, the general feeling that you don't lose the same for losing an overtrick as you do for losing the whole contract will make the scoring feel more "real" to the spirit of bridge.

However unlike full IMPs, you will be able to recover from one or two bad results more easily.

[Cyberyeti]

If I was going to do this, I'd run a modification of patton teams scoring possibly with some sort of clever calculation being used for the datum against which the swings are calculated rather than the result from the other table that you can use at teams, possibly the standard Butler datum.

[helene_t]

Patton against the field (like in ximps ) is fine

[EarlPurple]

This topic discusses various Patton style scoring systems:

http://www.bridgebas...ch/page__st__20

The pure "Patton" system is normally used for team games. I am saying it would work well in a pairs tournament (where you play a few boards against a pair then move on, and score against the field).

In MP, you score "board a match" against every other pair playing the same hand. In X-Imps you play IMPs against all the other pairs (which then gets averaged). My suggestion would be a form of Patton on every hand against every pair.

The first option is similar to board-a-match with a 10-point swing counting as a tie. The second is closer to regular Patton but I have only introduced two levels of "victory" on a board.

[Cyberyeti]

 EarlPurple, on 2013-June-23, 09:38, said:

This topic discusses various Patton style scoring systems:

http://www.bridgebas...ch/page__st__20

The pure "Patton" system is normally used for team games. I am saying it would work well in a pairs tournament (where you play a few boards against a pair then move on, and score against the field).

In MP, you score "board a match" against every other pair playing the same hand. In X-Imps you play IMPs against all the other pairs (which then gets averaged). My suggestion would be a form of Patton on every hand against every pair.

The first option is similar to board-a-match with a 10-point swing counting as a tie. The second is closer to regular Patton but I have only introduced two levels of "victory" on a board.

I was suggesting that you aggregated the results of the round and used a Pachabo like system.

[EarlPurple]

There are variations on the theme. I knew of the system having been used in teams events.

When I say "what do people think" it's not so much which is the perfect way to implement it perhaps as much as whether such tournaments would be a good idea. I like the idea myself as it combines what I like about both forms of scoring and eliminates much of what I don't like.

A situation in MP is when you are competing a part-score and the opponents are vulnerable. Say we bid to 2♠ and they compete to 3♦. We might bid on to 3♠ but if we double them and it's touch and go whether we'll beat them, we're shooting tops and bottoms. In rubber of IMPs you would never go for the extra 100 risking doubling them into game, but at MP it is a potentially good tactic if you think the chances of beating them are a bit more than 50%. Certainly in a Patton form of scoring you'd need better odds than that.

+200 beating 3♦ versus +140 making in 3♠ is a small win (3-1) in my second proposal but -670 would be a full loss. If you go down -100 against their -110 that is considered flat, so pass also becomes an option.

[TylerE]

What I thin would be cool is a form of normalized imps.

Something like this:

Score at ximps as usual, drop the top and bottom 10% of scores. Now, scale so those at the 90th percentile (e.g. those that are right on the cutoff for the boards that got dropped) as +/- 10.0 "imp-equivalents". Now scale all the scores (including those originally dropped. In effet, this will be MPish (an overtrick could suddenly be worth 5 imps or something), except 10pt differences won't matter much, and if you have a really extrodinary result (Like only one in the field to bid and make grand), it's worth a bit more than an average board.

[Siegmund]

I have kicked around the idea of smaller, 4- 6- or 8-point "IMP" scales... things like 10-40=1, 50-240=2, 250-490=3, 500+=4... but never really been captivated by any of the ideas I have tried. I am a big believer that the 10-point difference should count. It's one of my many annoyances with IMPs as used today.

But a big question is, who will such a system make happy? I like matchpoints. I would probably like matchpoints better than a hybrid system. Other people like IMPs. They would probably like IMPs better than a hybrid system.

To gain any traction, there would have to be some specific goal, for which the new approach is nearly optimal. I can't think of one, off the top of my head.

[helene_t]

 TylerE, on 2013-June-24, 14:30, said:

What I thin would be cool is a form of normalized imps.

Something like this:

Score at ximps as usual, drop the top and bottom 10% of scores. Now, scale so those at the 90th percentile (e.g. those that are right on the cutoff for the boards that got dropped) as +/- 10.0 "imp-equivalents". Now scale all the scores (including those originally dropped. In effet, this will be MPish (an overtrick could suddenly be worth 5 imps or something), except 10pt differences won't matter much, and if you have a really extrodinary result (Like only one in the field to bid and make grand), it's worth a bit more than an average board.

I have thought of normalized imps but I am afraid the the tactical considerations would be quite different from something in between mp and imps. on very flat boards the 10 and 90 percentiles will be zero IMPs so an extra overtrick could win a lot after normalisation. on more normal boards the tactics would be similar to imps. maybe the importance of judging the flattness of the board would be even more crucial than at MP. I am not saying this is bad but it is like a third kind of bridge and I think that if you want to make something for a single event then you are better off doing Patton which has easy-to-understand tactical implications.

[EarlPurple]

when I thought about it, a much simpler version of what I am trying to achieve would be ximps with the existing scale but cap the IMP scale at 6, so any swing 220 and above is the maximum 6 imps.

That will overcome the issues of certain "big" hands because there are potentials for part-score swings in far more hands. Your tactic will be in general closer to IMP tactics.

[CSGibson]

I object to any form of scoring which I cannot understand/explain succinctly to my parents. Imp scoring approaches the threshold of complexity I can tolerate - this is way beyond.

[Zelandakh]

I daresay you could also achieve your objective by changing the premiums - eg: partscore = 0; NV game = 100; V game = 200; NV slam = 200; V slam = 300; NV grand = 400; V grand = 600. That would seem to increase the importance of the points for tricks without losing the incentive for "bidding up" completely.

[webwiz]

Does anyone know what form of scoring is used in the Speedball Pairs tournaments which use Imps? What is the optimum bidding style in these tournaments? Playing teams with imps we do not risk the contract for the sake of a possible overtrick and sometimes we will make a safety play giving up a trick unnecessarily in order to guarantee the contract. In match pointed pairs we will try to play in NT if possible, particularly if the alternative is a minor, and will strive for overtricks even if it risks going off in a cold contract if a suit misbehaves badly etc etc

[ArtK78]

 webwiz, on 2013-September-27, 11:30, said:

Does anyone know what form of scoring is used in the Speedball Pairs tournaments which use Imps? What is the optimum bidding style in these tournaments? Playing teams with imps we do not risk the contract for the sake of a possible overtrick and sometimes we will make a safety play giving up a trick unnecessarily in order to guarantee the contract. In match pointed pairs we will try to play in NT if possible, particularly if the alternative is a minor, and will strive for overtricks even if it risks going off in a cold contract if a suit misbehaves badly etc etc

The scoring in the ACBL BBO IMP pairs games is XIMPs, which means that every result is IMPed against every other result and then the average is taken. I don't know if any extreme results are thrown out.

Example: Suppose there are six results on a hand. +420, +420, +170, +140, -50 and -100. The +420s would have IMP results against the other 5 pairs as follows: 0, +6, +7, +10 and +11. The average of these results is +6.8, and that is the score that the +420s would get. The +170 would have IMP results against the other 5 pairs as follows: -6, -6, +1, +6 and +7. The average of these score is +0.4, and that is the score that the +170 would get. This assumes that none of the extreme results are discarded before making the calculations.

I bid and play the hands as if they were being played in a heads-up team match, but with the understanding that I have to get good results on every hand (or, certainly, almost every hand) to wind up plus enough IMPs at the end of 12 boards to win.

[arete]

This is a great idea. Similar ideas have been suggest but never gain traction.
Pros
Makes Pairs more like bridge
Helps solve the Major suit/ notrump Dilemma
Seems fairer

Cons
Harder to understand (at our club, half the players do not even understand matchpoints)
Change

Can we define it in detail and give it a try. With computers you could score both ways (although I know you might have different strategies).
I like Option 2 but Option 1 is okay too.

[ArtK78]

I have to agree with the sentiment that playing pairs scored at IMPs sounds like a great idea. As arete said, it "makes pairs more like bridge."

However, it is my experience that the idea sounds much better than it works in practice.

I have played in a number of IMP pair tournaments in face-to-face competition. The reality is that, unlike a matchpoint pairs tournament, each board is not worth the same as the other boards. Boards on which games and slams can be made or can be beaten usually count far more than part-score hands, since the number of IMPs at stake are much higher. In an IMP pairs tournament, you can suffer a huge loss on a hand through no fault of one's own merely by playing a board against the wrong pair - the only one to bid and make a slam, for example.

Typically, your results on a handful of important hands will determine whether you have a winning session or a losing session.

At matchpoints, you can be fixed on any given board, but at least the damage is limited to one board. At IMP pairs, a -13 on one board may take 3 or 4 very good results for your side to recover.

I suppose you could say that this is true in "real bridge," if you mean by real bridge a heads-up team match. But in a heads-up team match you have your teammates at the other table and you can hope that they are able to achieve the same result as the opponents at your table. In an IMP pair event, all you have is "the field."

[TylerE]

 ArtK78, on 2013-September-27, 13:41, said:

In an IMP pairs tournament, you can suffer a huge loss on a hand through no fault of one's own merely by playing a board against the wrong pair - the only one to bid and make a slam, for example.

Personally, I think that happens just as much at mps. It's just partscore randomizing that kills you, rather than game/slam bidding.

[ArtK78]

 TylerE, on 2013-September-27, 13:50, said:

Personally, I think that happens just as much at mps. It's just partscore randomizing that kills you, rather than game/slam bidding.

But at IMP pairs, one huge loss may cost you the equivalent of 3-4 "tops." At matchpoints, the worst you can get is a zero. One top gets you back to average.

For example, playing IMP pairs on BBO, I can get 4 or 5 extremely good results on part score hands and the IMP results will be +2.3, +3.4, +1.4, +1.6, for a total of +8.7. But then if you go down in a vulnerable game which is on a guess (or, more likely, a guess which is a 65-70% favorite), or you go down in a very good game that very few others will bid, and you wind up with -8.7. And you are back to zero which, in IMP pairs, is worth exactly that - nothing. If those results occurred at matchpoints, they might be 80%, 90%, 65%, 70%, 10%, for a total result of 63%. Now, 63% may not win the event, but at least you are not out of the running.

Personally, I like IMP pairs on BBO. I find that better bidding and better play is rewarded more at IMP pairs than it is at matchpoints. But, there are reasons that I used to refer to IMP pairs as "matchpoints on steroids."

[Vampyr]

 ArtK78, on 2013-September-27, 13:41, said:

I have to agree with the sentiment that playing pairs scored at IMPs sounds like a great idea. As arete said, it "makes pairs more like bridge."

However, it is my experience that the idea sounds much better than it works in practice.

Quite. A world-class player once commented to me that IMP pairs is like a team game in which your teammates are the worst pair in the room.

[Zelandakh]

 ArtK78, on 2013-September-27, 14:50, said:

Personally, I like IMP pairs on BBO. I find that better bidding and better play is rewarded more at IMP pairs than it is at matchpoints. But, there are reasons that I used to refer to IMP pairs as "matchpoints on steroids."

Do you really think any scoring on BBO is a good yardstick? My experience is that almost any result is possible there and that just bidding (and making) an obvious game gives approximately +4-6 IMPs. Of course that is not dissimilar to club bridge but I would not call that a good yardstick either.