[klaus]

Hi,

there was a recent discussion in a german bridge forum about BBOs IMPs pairs scoring method.

Is the description from the BBO- Help-Site still accurate?
Someone gave an example that he thought shows that method 1 is in use. I didn't cross-check but decided to ask someone who knows :-)

Here is for everybodies convenience the "official" BBO help text I found:

"Computing the Score
In team bridge tournaments the scoring is normally done using IMPs, but some pairs tournaments also use IMPs (instead of MPs) for scoring. This is called "IMP pairs" - clever name

Here's how IMP pairs works: Assume you are sitting North-South, play board 1 and get a score (say, for example, you bid and make 3NT and score +400).

Your +400 is then IMPed (using the IMP scale) against all the other North-South pairs who played board 1. Let's assume that each board is played 16 times, just like on BBO. There are 2 ways that the IMPing can take place at IMP pairs:

-1- Your +400 is compared 15 times with the scores of the other 15 North-South pairs (using the IMP scale). The IMPs you won or lost in each of these comparisons is added up and that is your partnership's IMP score for board 1.

-2- The scores of all 16 North-South pairs are first added up and then divided by 16 to compute the average score for board 1. Then your +400 (and the scores of all the other North-South pairs) are IMPed against this average using the IMP scale.

BBO uses method 2 (mostly because it keeps the numbers smaller and easier to comprehend)."

I can post the score-calculation in question if needed.

Thx and bye

Klaus

[Trpltrbl]

It seems that 1 ) is Imp-pairs and 2 ) is the one they use to compare regular Imp games.

Mike

[McBruce]

I'm surprised at this; I thought the smart folks behind BBO would know better than to use Butler scoring without tossing the top and bottom 2-3 results out before computing an average. Cross-IMPs is far better than Butler and isn't so difficult to do.

[inquiry]

I disagree that method "2" is the one used for pair imps. Clearly, cross imps are used... Comparing the results of imps earned or lost at all the other tables. You can go to myhands site, pull up any traveller and test it yourself.


Contract ___ Score___ BBO Imps __Diff.____ "imps earned by first pair
5SExx-1_____200_____10.93__________
4SE-1_____50_____8.8____ -150_____4
3HW+2____ -200_____3.67____ -400_____9
1HW+4____ -200_____3.67____ -400_____9
2HW+4____ -230_____2.87____ -430_____10
4HW=____ -420____ -1.93____ -620_____12
4SE=____ -420____ -1.93____ -620_____12
4HW=____ -420____ -1.93____ -620_____12
4SE=____ -420____ -1.93____ -620_____12
4SE+1____ -450____ -2.8____ -650_____12
4HW+1____ -450____ -2.8____ -650_____12
4SE+1____ -450____ -2.8____ -650_____12
4SE+1____ -450____ -2.8____ -650_____12
4HW+2____ -480____ -3.67____ -680_____12
4HW+2____ -480____ -3.67____ -680_____12
4SE+2____ -480____ -3.67____ -680_____12

Total imps for 5Sxx-1 is 164 imps, divide that by 15 tables, you get the 10.93 reported. Similar checks for the other people show how to calculate the other imps....

[McBruce]

I'm glad to hear that. I only wish I could convince ACBL TDs that cross-IMPs is better!

Good to see that BBO divides by the number of comparisons and not the number of results. I have seen arguments for the latter and I am not convinced. If you make eleven tricks in 3NT and fifteen other pairs make ten, you should be +1.00 IMPs, not +0.94 IMPs.

[mikestar]

I imagine Butller himself would admit that cross IMPs are better now that we are scoring by computer. The Butler method had some advantages in the days of hand scoring.

[JRG]

Thank you Ben for putting up the example showing how the scoring is really done.

The Online Help had an almost exact quote from Fred (I think it was even from a post here in the Forums). I suspect perhaps he and Uday changed the algorithm (or perhaps intended one and did the other).

In any case, I have corrected the Help file.

Thank you Klaus for noticing this problem and posting here - much appreciated.