[blackshoe]

The recent thread on double dummy analysis brought this to mind. Basically, the question is how best to evaluate what you did with a bunch of hands, and where you should concentrate your efforts at improving. I was taught, in going over hands after a game, to look at on which hands I thought there was a bidding problem, which a play problem, which a problem with the opening lead, and which a problem with our defense. Count the number of hands for each of these four things, and whichever is the most, concentrate on that first. In addition, look at the 20 or 25% of hands on which you did most poorly, and try to determine why that happened. Sometimes it's your partnership's fault, and sometimes it's not. Sometimes it's a combination of factors. Especially in the local clubs around here, which seem to have a very high variance in the results (measured against double-dummy analysis, anyway). Someday I suppose I'll build a spreadsheet to track all this stuff. So far I just look at individual sessions.

Anyone have any other insights or suggestions?

[FrancesHinden]

Nothing beats finding a good player to take you through the post mortem. (By 'good' I mean noticeably better than you are - whatever your standard.)

[blackshoe]

FrancesHinden, on 2014-June-25, 13:02, said:

Nothing beats finding a good player to take you through the post mortem. (By 'good' I mean noticeably better than you are - whatever your standard.)

Oh, I agree with that!

[Mbodell]

blackshoe, on 2014-June-25, 12:57, said:

Anyone have any other insights or suggestions?

I went nearly 6 months while I was playing fairly frequently tracking in a spreadsheet for each hand I played:

The contract, which direction I sat, my partner, the number of tricks, the matchpoint score, and the strength of field (using powerratings, this usually had to wait a month and occasionally was missing).

I was interested in which partnerships and contracts I was getting the best matchpoint results and what not. In general I think filtering by 4 situations makes a lot of sense:

1. You declare the hand
2. You defend and are on opening lead
3. You defend and are not on opening lead
4. You are dummy

(going in order from perceived most control to least control). I think looking at each of these is interesting. If your partner is a lot different skilled than you, you will notice 1 and 4 are very different. The least skilled partner I was playing with had me averaging about 60% of the matchpoints when declaring, with her averaging between 25-30% of the matchpoints when she was declaring. Conversely there was a partner with whom I was playing where when I declared we got just over 50% of the matchpoints but when she declared we got between 65-70% of the matchpoints.

If your number 1 number is a lot worse than 4, concentrate on the declarer play (bridge master time).

If your 2 is a lot worse than 3, concentrate on how you make opening leads.

If your 2&3 is less than 1&4 concentrate on defense, signalling, and other defensive actions.

If your 1&4 is less than 2&3 concentrate on your constructive (and competitive) bidding.

I was surprised to find that across all my partners combined, including different bidding systems and quality of partners, that we got about the same matchpoints for 1&4 as for 2&3. We get about 2/3 of the matchpoints for "the average of any positive raw score" and about 1/3 of the matchpoints for "the average of any negative score" (even though obviously some good sac negative scores are great results and some positive slam in a part score are terrible scores), but that our side defend or our side declare was about equal.

When declaring, unsurprisingly, we score best for being in NT.

[blackshoe]

Hm. Good ideas, mbodell, thanks!

[SteveMoe]

Excel spreadsheets and PR files downloaded from a .txt PR file can speed your data collection and analysis....here's a dated file that shows one way to do this. I have some spreadsheets somewhere....

http://www.cincybrid...artnerships.pdf

Kudos to Mbodell for enlightened analytics.

BTW - the more one automates the process, the less one learns....

[jallerton]

Mbodell, on 2014-June-25, 15:38, said:

I was interested in which partnerships and contracts I was getting the best matchpoint results and what not. In general I think filtering by 4 situations makes a lot of sense:

1. You declare the hand
2. You defend and are on opening lead
3. You defend and are not on opening lead
4. You are dummy

(going in order from perceived most control to least control). I think looking at each of these is interesting. If your partner is a lot different skilled than you, you will notice 1 and 4 are very different. The least skilled partner I was playing with had me averaging about 60% of the matchpoints when declaring, with her averaging between 25-30% of the matchpoints when she was declaring. Conversely there was a partner with whom I was playing where when I declared we got just over 50% of the matchpoints but when she declared we got between 65-70% of the matchpoints.

If your number 1 number is a lot worse than 4, concentrate on the declarer play (bridge master time).

If your 2 is a lot worse than 3, concentrate on how you make opening leads.

If your 2&3 is less than 1&4 concentrate on defense, signalling, and other defensive actions.

If your 1&4 is less than 2&3 concentrate on your constructive (and competitive) bidding.

I was surprised to find that across all my partners combined, including different bidding systems and quality of partners, that we got about the same matchpoints for 1&4 as for 2&3. We get about 2/3 of the matchpoints for "the average of any positive raw score" and about 1/3 of the matchpoints for "the average of any negative score" (even though obviously some good sac negative scores are great results and some positive slam in a part score are terrible scores), but that our side defend or our side declare was about equal.

When declaring, unsurprisingly, we score best for being in NT.

Some clubs uses scoring software (Pianola) which automatically supplies players with the %score in your four scenarios. There are so many random factors in bridge that I think you'd need a fairly large sample size before the results become statistically significant.

I seem to recall that someone did some analysis of lots of hands played on Vugraph. This appeared to show that Nunes (#2 rated player in the world) was a far better declarer than Fantoni (#1 rated player in the world). Or maybe it was the other way round, but the stated difference between the two seemed highly implausible to me.

[paulg]

jallerton, on 2014-June-29, 14:57, said:

I seem to recall that someone did some analysis of lots of hands played on Vugraph.

Richard Pavlicek's site:

Major events data: http://www.rpbridge.net/rpme.htm
Partnership stats from major events http://www.rpbridge.net/9ya1.htm