BBO Discussion Forums: Time to analyse ZAR Points - BBO Discussion Forums

Jump to content

  • 2 Pages +
  • 1
  • 2
  • You cannot start a new topic
  • You cannot reply to this topic

Time to analyse ZAR Points Just another gimmick?

#21 User is offline   32519 

  • Insane 2-Diamond Bidder
  • PipPipPipPipPipPip
  • Group: Advanced Members
  • Posts: 1,454
  • Joined: 2010-December-22
  • Gender:Male
  • Location:Mpumalanga, South Africa
  • Interests:Books, bridge, philately

Posted 2012-April-09, 03:06

In addition to counting traditional High Card Points, other conditions affect the ability to make tricks, including Environmental Factors.

Alongside Environmental Factors, we also have the X Factor:
Positive interpersonal factors that differentiate top bridge players, including: focus, killer instincts, and tenacity to never give up.
1

#22 User is offline   Dark Widow 

  • Pip
  • Group: Members
  • Posts: 9
  • Joined: 2012-April-10
  • Gender:Not Telling

Posted 2012-April-10, 02:20

View PostStatto, on 2012-April-03, 17:48, said:

Linear regression analysis over 100,000 random deals gives the following values for correlation between the combined partnership count and the number of tricks made double dummy in the best suit fit:

KnR - 0.866
Zar - 0.865
Milton - 0.809

How does the popular 4.5/3/1.5/1 with 5/3/1 method fare?
What about if a queen counts 1.5 when unsupported but 2 with a higher honour?
Are you making any adjustments for honours in short suits, especially singletons? Are the correlations better if you subtract somewhere between a half point and a whole point for such singleton honours?
0

#23 User is offline   gwnn 

  • Csaba the Hutt
  • PipPipPipPipPipPipPipPipPipPip
  • Group: Advanced Members
  • Posts: 11,505
  • Joined: 2006-June-16
  • Gender:Male
  • Location:Enschede, the Netherlands
  • Interests:matching LaTeX delimiters :(

Posted 2012-April-10, 02:28

View Postsfi, on 2012-April-08, 03:26, said:

If you really want another method, I offer the 'warm fluffy' coefficient. Say you have a decision to make. You will know whether it's maximum or minimum and those decisions are easy. If it's in the middle you look at the hand and count the number of 'warm fluffies' you get when looking at it. If it's enough, you take the more aggressive action; if not you take the more conservative one. Put another way, work on your instincts and learn to trust them.

I still prefer the Bowles method of evaluation (people interested will have to google it or bbfsearch it).
... and I can prove it with my usual, flawless logic.
      George Carlin
0

#24 User is offline   sfi 

  • PipPipPipPipPip
  • Group: Full Members
  • Posts: 522
  • Joined: 2009-May-18
  • Location:Oz

Posted 2012-April-10, 12:10

View Postgwnn, on 2012-April-10, 02:28, said:

I still prefer the Bowles method of evaluation (people interested will have to google it or bbfsearch it).


Also a good approach. But my method gives a number, which means it's statistically more valid.
0

#25 User is offline   chasetb 

  • PipPipPipPipPip
  • Group: Full Members
  • Posts: 756
  • Joined: 2009-December-20
  • Gender:Male
  • Location:Podunk, backwater USA

Posted 2012-April-10, 12:46

I have 3 more approaches to share:

Winning Trick Count (WTC) - basically this is like LTC (original), but it counts winners instead of losers. Occasionally, you get different evaluations. I've modified it, and my 'new' version is pretty similar to NLTC. Here's the link (A .pdf I made of it only runs 23 pages, so if you want this information to print out but like to save paper, PM me and I'll send my file to you) - http://www.harlowbri....uk/winning.pdf

Four Aces - This was by a superstar American team led by Oswald Jacoby in the 30s. I don't remember as much as I'd like, but I remember an average hand was 6.5 points, and any hand 9.5 points or better had to open. Of course, modern day bridge would open all 9.0 hands, most 8.5 hands, and distributional 8.0 hands. The scale is 3/2/1/0.5, but for NT hands there is an additional factor, the formula is 0.5 x (# of honors - 7) and then add it in. My gut feeling and some very basic paper-and-pencil analysis says that the honors thing is extremely useful, especially once you have 20+ HCP balanced hands.

Bissell - This is somewhat similar to Work, considering that you take the points the hand evaluates to and divide by 3 to get the expected # of tricks. I've slightly modified it as well, but only to correct for a few cases where it overevaluated. It looks to be valuable only for suit contracts. Aces start at 3, Kings start at 2, and Queens start at 1. Depending on the suit, lower honors gain points. Some example to show what I mean - AK is 6, AQ is 5, AQJ is 7, AQT is 6, KQ is 4, KQJ is 6, and KJT is 4. 4-card suits gain a point, 5-card suits add 2 more (for 3 distribution points total), and for every card above 5, add 3 more points per card. Original Bissell is that for 5+ card suits, you add 4 points for every card over 4, but then you have to remember the most you should have per suit is 3 x (# of cards). I'll admit it might be better to start adding 4 points per card once you get to 7+ cards, but I don't care enough to look into it.
"It's not enough to win the tricks that belong to you. Try also for some that belong to the opponents."

"Learn from the mistakes of others. You won't live long enough to make them all yourself."

"One advantage of bad bidding is that you get practice at playing atrocious contracts."

-Alfred Sheinwold
0

#26 User is offline   Statto 

  • PipPipPipPipPip
  • Group: Full Members
  • Posts: 636
  • Joined: 2011-December-01
  • Gender:Not Telling
  • Location:UK
  • Interests:Bridge, Poker, but not in conflation.
    Statistics, but not massaged by the media.

Posted 2012-April-10, 16:27

View Postsfi, on 2012-April-08, 03:26, said:

It may be of academic interest ...

It's not just of academic interest. Any bridge player who wants to improve should be looking at hands after the event to see where they could have bid or played better, but to avoid bias should also look those where they bid and played really well, and give themselves a pat on the back :D

Looking at other methods of hand evaluation in various contexts is not of purely academic interest. You can see whether Zar or Banzai points would have helped you find the game that quite a few others did, or was found against your team-mates. Not really for a reason to use them, but to see what value they have, and whether there is something there which can help improve your judgement :)
A perfection of means, and confusion of aims, seems to be our main problem Albert Einstein
0

#27 User is offline   Statto 

  • PipPipPipPipPip
  • Group: Full Members
  • Posts: 636
  • Joined: 2011-December-01
  • Gender:Not Telling
  • Location:UK
  • Interests:Bridge, Poker, but not in conflation.
    Statistics, but not massaged by the media.

Posted 2012-April-10, 16:33

View PostDark Widow, on 2012-April-10, 02:20, said:

How does the popular 4.5/3/1.5/1 with 5/3/1 method fare?
What about if a queen counts 1.5 when unsupported but 2 with a higher honour?
Are you making any adjustments for honours in short suits, especially singletons? Are the correlations better if you subtract somewhere between a half point and a whole point for such singleton honours?

I will try and do this if I have time. But it may be better if I spend the time to tidy up the Java code and put it online somewhere so you can play around yourselves. Send me a PM to nag me if you're interested B-)
A perfection of means, and confusion of aims, seems to be our main problem Albert Einstein
0

#28 User is offline   weejonnie 

  • PipPipPip
  • Group: Full Members
  • Posts: 72
  • Joined: 2012-April-11

Posted 2012-April-14, 17:38

View PostStatto, on 2012-April-10, 16:33, said:

I will try and do this if I have time. But it may be better if I spend the time to tidy up the Java code and put it online somewhere so you can play around yourselves. Send me a PM to nag me if you're interested B-)


I wish people would remember the golden rule of bridge.

"Bridge is NOT about points. Bridge is about TRICKS"

Apply that adage with a bit of common sense and you'll do far better than just implementing an algorithm. The only problem is that you have to think a bit.
0

#29 User is offline   Statto 

  • PipPipPipPipPip
  • Group: Full Members
  • Posts: 636
  • Joined: 2011-December-01
  • Gender:Not Telling
  • Location:UK
  • Interests:Bridge, Poker, but not in conflation.
    Statistics, but not massaged by the media.

Posted 2012-April-15, 21:31

View Postweejonnie, on 2012-April-14, 17:38, said:

Apply that adage with a bit of common sense and you'll do far better than just implementing an algorithm. The only problem is that you have to think a bit.

Don't worry, I do. But it is also useful to study how algorithms work (or don't) in order to improve one's own judgement, if you are that way inclined. And the GIB programmers have to stick to algorithms, or devise a NN learning system.

I mean, you surely already use an algorithm to decide whether to make an opening bid. Something about Milton points, I gather, but I never got the hang of it. I just look at my hand and decide if it feels like it's worth an opening bid B-)
A perfection of means, and confusion of aims, seems to be our main problem Albert Einstein
0

Share this topic:


  • 2 Pages +
  • 1
  • 2
  • You cannot start a new topic
  • You cannot reply to this topic

1 User(s) are reading this topic
0 members, 1 guests, 0 anonymous users