BBO Discussion Forums: The decimal places in IMP results - BBO Discussion Forums

Jump to content

Page 1 of 1
  • You cannot start a new topic
  • You cannot reply to this topic

The decimal places in IMP results for the 1000 and first time

#1 User is offline   Miron 

  • PipPipPipPip
  • Group: Full Members
  • Posts: 358
  • Joined: 2006-January-30
  • Location:Praha, Czech Republic

Posted 2006-July-10, 06:33

Hi,
this was surely answered but I haven't found it (could this be in Bridge Library section?):

How are calculated the IMPs at BBO (especially the decimal places)?
0

#2 User is offline   helene_t 

  • The Abbess
  • PipPipPipPipPipPipPipPipPipPipPip
  • Group: Advanced Members
  • Posts: 17,401
  • Joined: 2004-April-22
  • Gender:Female
  • Location:Odense, Denmark
  • Interests:History, languages

Posted 2006-July-10, 07:01

Your score is compared to that of all other pairs that sit in the same direction. So for each other table you get some number of IMPs. The average of all those IMPs is your chross-IMP score. This is quite standard.
The world would be such a happy place, if only everyone played Acol :) --- TramTicket
0

#3 User is offline   Miron 

  • PipPipPipPip
  • Group: Full Members
  • Posts: 358
  • Joined: 2006-January-30
  • Location:Praha, Czech Republic

Posted 2006-July-10, 07:21

But the decimal places?

The table which converts total points difference (your against average) is for full IMPs (20-40 is 1 imp etc.). How can you get 1.3 IMP?
0

#4 User is offline   inquiry 

  • PipPipPipPipPipPipPipPipPipPip
  • Group: Admin
  • Posts: 14,566
  • Joined: 2003-February-13
  • Gender:Male
  • Location:Amelia Island, FL
  • Interests:Bridge, what else?

Posted 2006-July-10, 07:24

A short example, three tables.

Quote

  1 3 HX   W SA  Player1   6     800     13.00         
  2 3D     N  H6  Player2   12     170     -2.50         
  3 3N     N  H3  Player3   8     -100    -10.50 


Player1, +800, Player 2 +170 = player1 up 630 or +12 imps, player 2 is -12.00 imps

Player1 +800. Player3 -100 = player1 up 900 or +14 imps, player3 = -14 imps

Player2 +170, player3 -100 = player2 up 270 or 7 imps, player 1 minus 7 imps.

Player 1 +12 +14 = 26
Player2 -12 + 7 = -5 = -5/2 = -2.5
Player3 -14 -7 = =21 = -21/2 = -10.5

Notice divide by the number of hands each partnership played to get average imp score.
--Ben--

#5 User is offline   FrancesHinden 

  • Limit bidder
  • PipPipPipPipPipPipPipPipPip
  • Group: Advanced Members
  • Posts: 8,483
  • Joined: 2004-November-02
  • Gender:Female
  • Location:England
  • Interests:Bridge, classical music, skiing... but I spend more time earning a living than doing any of those

Posted 2006-July-10, 07:35

You are getting the order of what happens (averaging and imping) the wrong way round.

The program calculates your imp result against every other table's result who has played that board. If the board has been played 6 times, you will have five imp results - think of it as scoring with five different team-mates sitting the other way. That gives you a total imp result (called the 'cross-imp') which can be quite a big number. To make the number easier to understand, and to allow like-for-like comparisons when there are different numbers of tables involved, this total imp number is divided by the number of comparisons to give you an average cross-imp for the board.

To take a simple example, suppose you score +620, one other pair score +620 and everyone else has made +170 on your cards. If the board has been played 10 times you will score +10 imps 8 times and 0 imps once, averaging +80/9 or +8.9 imps times as you have 9 comparisons. If the board has been played 20 times you will score +10 imps 18 times and 0 imps once, averaging +9.5 imps.

There is an alternative method of scoring called 'Butler' imps which is close to what you describe: in that form of scoring your result is compared to the average of all the other results and that difference imp'd (sometimes not all the other results are used). In practice both methods get very similar answers over a big enough set of boards, but cross-imps are generally preferred now (Butler imps were easier to calculate in the days before computers did our scoring for us).
0

#6 User is offline   Miron 

  • PipPipPipPip
  • Group: Full Members
  • Posts: 358
  • Joined: 2006-January-30
  • Location:Praha, Czech Republic

Posted 2006-July-10, 10:20

Thanks for explaining!

Now i understand it. BBO uses cross-imps divided with number of tables the board is played. (Hope this is true :P )
0

#7 User is offline   Trumpace 

  • Hideous Rabbit
  • PipPipPipPipPipPip
  • Group: Advanced Members
  • Posts: 1,040
  • Joined: 2005-January-22
  • Gender:Male

Posted 2006-July-10, 10:57

Miron, on Jul 10 2006, 11:20 AM, said:

Thanks for explaining!

Now i understand it. BBO uses cross-imps divided with number of tables the board is played. (Hope this is true :lol: )

If I understood Fred correctly, it is cross-imps divided by number of comparisons, number of comparisons is 1 less than the number of tables.
0

#8 User is offline   inquiry 

  • PipPipPipPipPipPipPipPipPipPip
  • Group: Admin
  • Posts: 14,566
  • Joined: 2003-February-13
  • Gender:Male
  • Location:Amelia Island, FL
  • Interests:Bridge, what else?

Posted 2006-July-10, 11:43

Miron, on Jul 10 2006, 11:20 AM, said:

Thanks for explaining!

Now i understand it. BBO uses cross-imps divided with number of tables the board is played. (Hope this is true :lol: )

As trumpace points out, the calculation is divided by the number of comparisons. Note for the three replay example I gave, the division was by two.. That is, you add up all your imps you earned compared one at a time, then divide by the number of comparisions you did to get your average result.
--Ben--

Page 1 of 1
  • 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