Join Bridge Winners
Why are the Instant Matchpoint Results consistently, hilariously wrong?

I played in the Instant Matchpoint game Weds night (for the first time in a while). There is always some grousing, of course. But Larry Cohen pretty much throws up his hands. And, in skimming through old booklets (available on the web), he does that each year.

It appears (and this may be an assumption by me) that the scores are given to him, based on "over 500 plays of each hand" that happened a long time ago. So long ago that the getting to a routine game (by modern standards) gets you a great score. Looking over the results from my local club

  • On Board 5 -- EACH AND EVERY TABLE had 650. Perfectly flat. This was worth 62 points to NS. That's mildly annoying, but not outrageous. There were plenty of new players (and experienced ones like myself, who make mistakes and should know better), so I'm rather surprised that not one pair botched the bidding or play.
  • Look at Board 6 -- The results range are +140 or +170 (with a +150 in the mix). So, the +140s get ... 70% For an average minus. What?
  • Lest you think I'm just picking on these boards, the total score for the 9 N/Ss were 12754, and the total scores for the EW were 11546. Which means that N/S averaged over a full board more than E/W. I'd be interested to see if that's typical (our club, having 9 full, played boards 1-27 only). I think that's related to the hand distribution, as I only declared two hands.
  • In general, there are too many extreme scores. Board 1 has hands all over the place (as you'd expect with a 3 loser hand that may have to enter at the five level). Let me just give the matchpoint awards 97,90,90,90, 47,14,14,14. That's like look at a traveller and going "12, 11,11,11,6,1,1,1,0" If you just took those MPs as scores, we'd have "8,5,5,5,3,1,1,1" which would translate back as percentages to "100,62,62,62,38,12,12,12." So, the '14' results make sense. (Those who shot 6D and got set are duly punished). Perhaps someone can do a statistical outline to see how much.

But, despite all the underlying complaints, my main issue is this -- What's the point of giving the hands to Larry Cohen (or any expert) if you aren't going to allow him to correct the scores based on changes in bidding and standards in the last 50 years? Let him say "Yes, when this was played only the most aggressive bidders would get to this cold game, but now most people open this hand and get there" I didn't count, but on at least 5 of the hands Larry basically said "The scoring table screws the normal result."

We let that happen why, exactly? For purity of historical records?

If you can't bear to generate new deals, why not take the results and ask experts to create reasonable travelers. Maybe set up a computer program to play each hand 100 times. Sometimes with "Flight A" players, sometimes with "Flight B." (Sometimes A NS vs B EW or vice versa). Some set to "Aggressive" and some "passive". You could even use multiple programs. Mix the historical records with the experts imagined section and the computer results. Those should smooth out things to prevent the "I did worse than anyone else in my club and still got 70% on this board."

There are issues, but would this really be any worse?

Of course, the ideal is something like the Common game, but where every club's results are submitted to the ACBL and we have the most massive cross matchpointing in the history of the game. That would even be worth a day or two.

"So, what did you do on board 3?"

"Oh, our opponents doubled us in 4H, making 5"

"Was that a top?"

"No, we only got 1285, and top was 1297."

 

 

10 Comments
Getting Comments... loading...
.

Bottom Home Top