[bluejak]

I have received an email!



Hi David

I have some problems understanding the exact procedure for calculating self-inflicted damage in a pairs tournament. Suppose we have the following frequency table in a 7-table tournament before the TD makes his ruling.

1 x +1430 = +6 mp
4 x +680 = +1 mp
1 x +500 = –4 mp
1 x +300 = –6 mp

The TD adjusts the +300 to +680, but N/S committed a serious error, without which they Could have scored either +800 or +500, each with a probability of 50%. Now, what is the precise algorithm that I should apply? Simply create the hypothetical frequency table with the “Could” result?

1 x +1430 = +6 mp
0.5 x +800 = +4.5 mp
4 x +680 = 0 mp
1.5 x +500 = –5.5 mp

Thus, N/S Could have obtained –0.5 mp instead of the Actual result of –6 mp, so the self-inflicted damage is 5.5 mp.

This is the frequency table after the adjustment:

1 x +1430 = +6 mp
5 x +680 = 0 mp
1 x +500 = –6 mp

Therefore, N/S score 0 mp – 5.5 mp = –5.5 mp (almost all damage self-inflicted). The other N/S pairs score 0 mp in total. This is slightly odd, because we have one N/S pair scoring a rock bottom –6 mp, and another N/S pair scoring –5.5 mp, although next-to-bottom would be –4 mp, but maybe it is acceptable.

Suppose N/S Could have scored +800 without the serious error, i.e. all damage (and more) is self-inflicted. Now common practice is to let the score stand and use the original frequency table for N/S scoring. Thus, N/S score –6 mp instead of –5.5 mp in the previous example, but all of a sudden the other N/S pairs score +6 mp in total, and specifically our +500 pair goes from –6 mp to –4 mp although no N/S table result was changed from the previous example. This feels wrong, so let’s use the above method instead of common practice:

“Could” frequency table:

1 x +1430 = +6 mp
1 x +800 = +4 mp
4 x +680 = –1 mp
1 x +500 = –6 mp

Self-inflicted damage is now 4 mp – (–6 mp) = 10 mp. Adjustment minus self-inflicted damage is –10 mp, but obviously we are not going to give them less than their –6 mp from the Actual frequency table. Now, this is also strange, because both the +500 obtained at a different table and the +300 for our heroes amount to –6 mp.

I fail to find a consistent and fair method for dealing with this. Is it simply impossible to devise such a method, or am I just missing something?



Any views?

[aguahombre]

You should stop opening your emails.

[mfa1010]

In Denmark we do like this, but I don't know if this is homebrew or adopted standard from elsewhere.

For OS and the rest of the field, the matchpoint table is the one that includes the adjusted score for OS (680), so:


Board MP table:

Quote

1 x +1430 = +6 mp
5 x +680 = 0 mp
1 x +500 = –6 mp

For NOS the starting point is also this (0 mp). But then we deduct what the serious error cost.

In example 1 (50% 500, 50% 800) the serious error cost 5,5 mp as correctly calculated by the emailer.
[50% from -6 to -5, and 50% from -6 to +4]

So NOS get 0 - 5,5 = -5,5mp

In example 2 (100% 800) the serious error cost 10 mp.
[From -6 to +4]

Since the serious error cost more than the adjustment of 6 mp [from -6 to 0], NOS keep the result of -6 mp, which they would have got if they had entered the table result (300) in the mp table. They can't lose more.

[Trinidad]

bluejak, on 2014-April-21, 08:34, said:

The TD adjusts the +300 to +680, but N/S committed a serious error, without which they Could have scored either +800 or +500, each with a probability of 50%.

This is wrong.

There are three possibilities:

Either the 500/800 decision is in the hands of NS, or it is in the hands of EW, or (I don't know how) it is some random act (e.g. coming from outer space).

1. It is in the hands of NS
If NS not have made a serious error and would have taken the right view, they would have gotten +800. If NS would not have made a serious error and would have taken the wrong view, they would have gotten +500. Taking a wrong view is not a serious error. Therefore, it is 100% +500.

2. It is in the hands of EW
This is only possible if NS blew 1.5 tricks: The serious error cost a trick and (e.g.) declarer could now avoid a 50-50 guess. Again, having an opponent guess correctly is not a serious error on your side. And again, the serious error is 100% +500.

3. Random act
Random acts from outside are not serious errors. You get the picture: 100% +500.

So, to convert all the tables to normal, non-negative, European MPs, we get (frequency, score, NS MPs, EW MPs, Danish NS MPs, Danish EW MP, this table in bold):

Actual table result
1 x 1430 12  0  6 -6
4 x  680  7  5  1 -1
1 x  500  2 10 -4  4
1 x  300  0 12 -6  6

Table result without the serious error by NS
1 x 1430 12  0  6 -6
4 x  680  7  5  1 -1
2 x  500  1 11 -5  5

Therefore, the serious error cost NS 1 MP. Later, we will subtract 1 MP.

Spoiler

If you really insist on weighting the score and have a contribution for NS 800:
1 x 1430 12  0  6 -6
1 x  800 10  2  4 -4
4 x  680  5  7 -1  -1
1 x  500  1 11 -5  5
The serious error cost 8 MP. 50%x 1+ 50%x8= 4.5 MP. This means that later we would subtract 4.5 MP instead of 1.

EW:
1 x 1430 12  0  6 -6
5 x  680  6  6  0  0
1 x  500  0 12 -6  6

EW score 6 MPs
NS:
1 x 1430 12  0  6 -6
5 x  680  6  6  0  0
1 x  500  0 12 -6  6

NS at this table score 6 (adjusted result) - 1 (serious error) = 5 (or in Danish MPs: 0-1=-1).

Rik

[mfa1010]

mfa1010, on 2014-April-21, 11:52, said:

In Denmark we do like this, but I don't know if this is homebrew or adopted standard from elsewhere.

I see now that the topic has the subtitle "Denmark".
We do however have pretty clear (although complicated) regulations to handle this.

Trinidad, on 2014-April-21, 12:02, said:

Quote

The TD adjusts the +300 to +680, but N/S committed a serious error, without which they Could have scored either +800 or +500, each with a probability of 50%.

This is wrong.

Under Danish regulations a weighted score or even an artificial score is explicitely allowed here as the hypothetical score for NOS without their serious error.
Presumably to be used when the number of tricks is hard to judge.

[bluejak]

mfa1010, on 2014-April-21, 13:02, said:

I see now that the topic has the subtitle "Denmark".
We do however have pretty clear (although complicated) regulations to handle this.

Are you present at the Danish Festival in July? I shall be there, and the person who wrote to me, so perhaps we could discuss it together.

[mfa1010]

bluejak, on 2014-April-23, 11:35, said:

Are you present at the Danish Festival in July? I shall be there, and the person who wrote to me, so perhaps we could discuss it together.

Yes, but in the two weekends only (fri-sun).

[jallerton]

In England, this type of situation is covered in The White Book, I believe.

[FrancesHinden]

Trinidad, on 2014-April-21, 12:02, said:

This is wrong.

etc

Rik

I don't think you have read the original post carefully.

The NS result at the table was +300.
(let's say they defended 6D doubled, NV)

The TD originally adjusted the NS score to +680, and EW get -680
(let's say that is 5S+1 and the EW 6D bid was illegal for some reason, maybe the bidder had UI suggesting he save)

However, during the defence to 6Dx found at the table, NS committed a serious error
(let's say that North led out of turn at some point early in the hand, and the resulting major penalty card let declarer enforce a hugely advantageous lead penalty)

If that serious error had not occurred, the defence would have gone differently, but the TD doesn't know exactly whether NS would have got +500 or +800 - perhaps North would have had a 50/50 guess at trick 11.