[GMarco24]

Recently I've played the hand which is quite normal... Bidding, declarer play and defense were good, anyhow possible results were 3♦= or 3♥-1 and leads to -110 or -100 for my side. Neither side can't make any game, the board was quiet at many tables, but...
One table played 7NTXX-12 on opps side. That makes my perfectly normal score very, very bad. Actually I missed the game there!
Do you have any ideas how to solve this, because it makes scoring funny sometimes?
My suggestion is to exclude some of the top and bottom results, and then calculate cross-imp as you were before. So, instead of 16 pairs, maybe 20 should play the board, exclude 4 results and calculate the cross-imp.
Otherwise, you are in this situation:
-Always when your opps make something (anything!) you'll lose 2-3 imps. ( There was a board where i lost 3.5 imps where opps made 6nt. They had 38 pts, and 16 tricks on 7nt. And -3.5 is because 3 pairs played on second level)
-One angry player can ruin the results for everyone on one board
I'll be gratefull if you at least think about my words.
Thank you for making wonderfull playground for us

[gwnn]

It doesn't actually matter. Your cross imps score will even out anyway. But your cross imp score doesn't matter!

[helene_t]

Yeah, don't worry about it.

Matchpoints would be slightly better IMHO but apparently everybody prefers xIMPs. Their own choice.

Btw, the 7NTxx-12 would be even more devastating at butler scoring!

[Cascade]

helene_t, on May 28 2009, 03:29 AM, said:

Yeah, don't worry about it.

Matchpoints would be slightly better IMHO but apparently everybody prefers xIMPs. Their own choice.

Btw, the 7NTxx-12 would be even more devastating at butler scoring!

In Butler it is routine to remove the top and bottom result(s).

[kenrexford]

Cascade, on May 27 2009, 10:36 AM, said:

helene_t, on May 28 2009, 03:29 AM, said:

Yeah, don't worry about it.

Matchpoints would be slightly better IMHO but apparently everybody prefers xIMPs. Their own choice.

Btw, the 7NTxx-12 would be even more devastating at butler scoring!

In Butler it is routine to remove the top and bottom result(s).

Yeah, but in my local club the next score is 7NTXX-11.

[Gerardo]

Makes no sense to remove results in cross imps. How do you assign scores to the removed results?

[Cascade]

Gerardo, on May 28 2009, 02:00 PM, said:

Makes no sense to remove results in cross imps. How do you assign scores to the removed results?

I am not saying that it makes sense but they could be cross-imped with the non-removed scores.

[GMarco24]

Gerardo, on May 27 2009, 09:00 PM, said:

Makes no sense to remove results in cross imps. How do you assign scores to  the removed results?

Maybe we could remove top 2 and bottom 2 results for them too... Of course, their result stay, so remove the next two

[bid_em_up]

Gerardo, on May 27 2009, 09:00 PM, said:

Makes no sense to remove results in cross imps. How do you assign scores to  the removed results?

The same way you do now.

Remove the top and bottom score, then calculate the average result, then IMP all results (including the removed ones) against this average.

Currently, I believe that all results are averaged then the IMP's calculated from that average. This allows a big score in one direction or another skew the imp results from what should be a "normal" result.

(But I could be wrong).

[1eyedjack]

bid_em_up, on Jun 4 2009, 06:03 PM, said:

Currently, I believe that all results are averaged then the IMP's calculated from that average. This allows a big score in one direction or another skew the imp results from what should be a "normal" result.

(But I could be wrong).

This is at odds with my understanding, which is that each result is IMP'd in full against each other result, the the net total of all these results divided by the number of comparisons.

(But I could be wrong)

[bid_em_up]

1eyedjack, on Jun 4 2009, 12:52 PM, said:

bid_em_up, on Jun 4 2009, 06:03 PM, said:

Currently, I believe that all results are averaged then the IMP's calculated from that average.  This allows a big score in one direction or another skew the imp results from what should be a "normal" result.

(But I could be wrong).

This is at odds with my understanding, which is that each result is IMP'd in full against each other result, the the net total of all these results divided by the number of comparisons.

(But I could be wrong)

No, you are right. The way I thought it was done is (evidently) the Butler method.

[barmar]

Butler scoring, AKA IMPing against the datum, was common when IMP pairs was scored by hand, since it's very tedious to calculate all the cross-imps. Cross-imps tends to be common with computer scoring, since they don't get tired of doing all those additions and subtractions.

[Cascade]

I don't have the source but I also believe that someone has shown that cross-IMPing is more sound than butler.

I recall a situation where improving one's score actually resulted in a lower butler score. Butler Problem

[NickRW]

I could be wrong, but I seem to remember reading other material on the merits of Butler versus X-IMPs on the same site. Anyway, wherever it was, I do recall that the conclusion was that X-IMPs are marginally better.

Nick

[matmat]

GMarco24, on May 28 2009, 08:41 AM, said:

Gerardo, on May 27 2009, 09:00 PM, said:

Makes no sense to remove results in cross imps. How do you assign scores to  the removed results?

Maybe we could remove top 2 and bottom 2 results for them too... Of course, their result stay, so remove the next two

i would vote for removing the top 8 and bottom 8.