Bowling Average
When
I bowled, no one ever asked me what the total number of pins I knocked down
was, but a lot of bowlers wanted to know what my average was. Master Points is the total number of
pins. Power Ratings is the bowling
average.
What is a Pair Rating
Pair Ratings are the strongest rating for a partnership. It is the pair’s average game percentage adjusted by their degree of difficulty (DOD). DOD is the difference between the pair’s average competition, the other N/S pairs if they are playing N/S, and the average pair. ?? – 50. In an Easy Bridge game this might be -10, (40 – 50), and in the Blue Ribbon Pairs it might be +15, (65 – 50). You need at least 12 sessions with a partner to have a Pair Rating with that partner.
What is a Power Rating
You and your favorite partner have just entered the bridge player’s twilight zone. When you open your eyes you see you and your partner are seated N/S. You look around the room and see that the other N/S players, your competition, are players of average skill. DOD = 0. When you look across the table at your favorite partner, you see yourself.
Power Rating is your average game percentage when your partner is yourself, and your DOD = 0. Power Rating is Self Rating doubled.
Everyone
has good days and bad days. But
everyone has an average level of play.
This is your Self Rating.
Your average level of play and your partner(s) average level of play equal
your average game percentage and your DOD.
Self + Part = G% + DOD. When I find Self Ratings for every
player that I have data for, about 200,000 players, I prove the Self Ratings
are correct by using this formula.
This formula must prove correct 200,000 times, (once for each player),
or else I don’t have the correct Self Ratings. While everyone has a Self Rating, (I
need this to calculate DOD), only players who meet the minimum requirements are
considered rated and can be found in the Power Rating leader tables. You need 11 games with rated partners
other than your favorite partner, (the rated partner you have the most games
with).
Example of Self Rating calculation
I play 11 games with John Doe. We average 50%. I play 11 games with Jane Doe. We average 55%. What does this tell me about my power rating? Nothing, but it does tell me that Jane Doe has a higher self rating than John. Jane has a self rating 5 points higher than John. Jane and John play 11 games with each other and they average 60%. Now we all become rated players. To become a rated player you need 12 games with other rated players. 11 of those games need to be with rated players other than your favorite rated partner, the rated partner you play the most games with. The only way to become a rated player with only 12 games is to play one game with each of 12 different rated players. The self ratings are calculated as follows:
This example assumes that all opponents are dead average. DOD = 0.
I did not calculate the above ratings (27.5 32.5 & 22.5). I applied different self ratings to the formula, your self rating plus the average of your partner’s self ratings equals your average game percentage, until all the equations were correct.
Power ratings assume that everyone has an average level of play and over the long run their scores reflect that average level of play. But this is real world. Some players make more mistakes with some partners than with others. Power ratings can not tell if your partner is making more mistakes, you are making more mistakes, or a combination of both. I doubt that any rating system, including Master Points, and Lehman Ratings which are found on OKBridge, will ever figure that one out. When power ratings see bad results, it assigns lower ratings to both partners. You can see where this is happening by looking at your personal summary.
Under “Unit Power Ratings” select your unit. Then find your name in the yellow box at the top of the page. If your name is not listed, contact me at ckchampion@netzero.net, and I will be happy to add your name. After you select your name you will see your summary. The left side of the screen shows a summary of your partners during the past 24 months. Your games with the partners near the top of the page have exceeded your and your partner’s power ratings. Your average game percentage with that partner has been divided between you and your partner using a ratio equal to your and your partner’s overall power ratings. The partners near the bottom, well someone, you or your partner, has not been playing up to their average level of play. It is no prediction of your future games, just a recap of what has happened during the past 24 months. The right side of your summary is devoted to individual games played during the past 4 months.
Power ratings measures one attribute of each player, his or her expertise at the bridge table. Master Points measures 4 attributes. (1) a bridge player’s expertise. (2) a bridge player’s partner’s expertise. (3) a bridge player’s opponent’s expertise. (4) How many ACBL sanctioned games a bridge played has competed in. Power Ratings looks only at a bridge player’s average game percentage over the last two years. The number of games played is not considered. This system calculates the player’s opponent’s average rating for each and every game. The player’s game percentage is adjusted up or down according to how much the opponents are above or below average. This eliminates the opponent’s expertise. This system subtracts the partner’s rating from the player’s game percentage. This eliminates the partner’s expertise. The remaining percentage is the bridge player’s expertise, or "Power Rating".
Your game percentage (G%) for each and every game you play is divided into 3 parts. Your contribution/Self Rating (self). Your partners contribution/Self Rating (Part) Your opponents contribution/ Self Rating (DOD). The sum of the opponents sitting your way, N/S if you are sitting N/S, is averaged and the amount above or below the average pair (50) is your degree of difficulty (DOD).
Self = G% + DOD - Part or Self + Part - DOD = G%
This
is the basic formula used to prove everyone’s rating. The program makes educated guesses then
uses this formula to see if each guess/Self Rating is too high or too low.
Cycle 1: The program starts with a rating of 25 for all players. All
DOD’s will be 0 and all partner ratings will be 25. Each player will get
a new rating of Self = G% + 0 - 25. If your average game is 52% then your
rating is 27. Since the new rating is different from the old rating, 25, then
it is not correct and a new cycle needs to be run.
Cycle 2: The next cycle starts with 27 for you as well as new ratings for all
other players. DOD’s for all games will be calculated and used to
calculate new ratings. Let’s say your average DOD is 50.25 (.25 above
average), average partner’s rating is 26, your average game is 52%. Self
= 52 + .25 - 26. Your rating is now 26.25 which does not match your previous
rating of 27. Not correct and a new cycle using a rating of 26.25 is run. Many Cycles: Typically 12,000 cycles are
needed to arrive at correct ratings for all players.
Adjustments are made to recognize the difficulty of a partnership’s opponents. Since you are competing against the players sitting in your direction, N/S for example, the ratings for all the other N/S players are averaged together, to calculate how much higher or lower than average your N/S opponent is. This percent is then added or subtracted from your game percentage. If your opponents are experienced, your game percentage is adjusted upward. If your opponents are less experienced, well your game percentage adjusts down. Self + Part = G% + DOD.
When
I suspended KO Power Ratings, I tried replacing them with actual KO
results. This did not work because
comparing the W/L record of teams in bracket 1 with teams in other brackets is
like comparing apples and oranges.
Instead I chose to prove/disprove the validity of Power Ratings. If teams flagged as underrated had W/L
records both above and below 50%, then Power Ratings would fail as often as
they worked. The teams shown here
are selected based on underrate (
KO’s should be bracketed so that all teams in a bracket have about the same chance of winning. This test is designed to show which teams win far more often than the average team, and which teams virtually never win their bracket.
Master Point
to Power Ratings conversion chart
This chart is the average Power Ratings for players in these Master Point ranges. If you want the average Power Rating for 5000 Master Points, it is the low end of the Power Rating range corresponding to 5000 to 10000 Master Points. For 7500 Master Points, the average Power Rating is the exact middle of the corresponding Power Rating range.
Games
Stratified by Power Ratings
The
game files include Strats based on Pair Ratings first, Power Ratings second,
Master Points last. At the end of
each line, the first character is the Strat assigned at game time. The second character is the Strat
assigned by Power Ratings. The last
number is the actual rating used to calculate the Power Rating Strat. Strat A is the top 3rd of the
field. B middle 3rd. C
bottom 3rd. If a Pair
Rating exists for that pair, it is used.
Otherwise each player’s Power Rating is used. If a player has neither, then a Power
Rating is computed using the Master Point to Power
Rating conversion chart.
What is the purpose of Power
Ratings?
It is a rating system based on a bridge player’s ability today.
Bridge events should be bracketed/stratified so that all teams/pairs within a bracket/strat have close to the same skill level and chance of winning.
When events are bracketed/stratified using Master Points, young hot shots (good players with few Master Points) are matched up against opponents who are not their equals. Seasoned veterans (players with lots of Master Points who have lost a few steps) are matched up against stronger opponents. Power Ratings identifies these players and assigns them to brackets/strats that they can compete in. I got the tearms “young hot shots” and “seasoned veterans” from ACBL members discussing the short comings of rating players based on Master Points.
Power Ratings brackets all tournament events played each month. Power Ratings tracks the teams that are not bracketed correctly by Master Points. Teams that are bracketed to low have won/loss records greater than 50%. Teams that are bracketed to high have won/loss records less than 50%. All teams that Power Ratings bracket 2 brackets lower than Master Points are on the -2 line.
Here are examples of young hot shots, KO teams that should be in a higher bracket.
We need to keep Master Points as rewards for winning events. However using Master Points to push players into ever higher brackets/strats is a disincentive.
Another use for Power
Ratings
Carry over calculations for multi day events. The reason we do this is because the field on the last day is stronger than the field on the first day. Carryovers are weighted so that high scores on the last day count more.
I think the way carryovers are calculated is not based on the strength of the advancing field. The carryovers are adjusted so that the spread between first and last is 2 ½ boards. If that is the actual spread, then the actual scores are carried over. If the spread is 10 boards then ¼ of the scores are carried over.
Power Ratings DOD should be added to all scores to reflect the true strength of the field each day. This will also eliminate the penalty of drawing a difficult section the first day.
· You need ACBLscore. Almost all clubs use this to score their games.
This looks daunting, but once you learn it, it will seem as easy as creating the monthly reports for the ACBL.
1) start ACBLscor
2) From the menu (top line) select "Utilities"
3) Select "Backup/Restore"
4) Select "2 Backup Game Files"
5) Backup Location:
Type in "c:\acblscor\gamefile"
or
hit the "browse" button and select a back-up location
(my favorite is desktop, usually at the top of the list)
6) Enter game files to backup. Type in "1107??" for July, 2011
the file name will be 1107xx.lzh, and found in
"c:\acblscor\gamefile" or on the "desktop"
Please attach this file to an e-mail and send to me at
ckchampion@netzero.net
I
can use up to two years worth of games, but I will use any amount of games you send.
I update each month. The best time to send me your new games is the same time
you e_mail your reports to ACBL.
You are done, but if you need more help with browse buttons and attachments:
Browse button:
1) From ACBLscor, after step 4 "2 Backup Game Files"
click on the browse button
Desktop is at the top. Click on it.
or
You can expand a different location by clicking on the icon
to the left of the location. You can continue expanding locations
within each location by clicking on the icon to the left of
the sub location. For example "ACBLSCOR" and then "GAMEFILE".
When you find the location you want, click on it, not the icon to
the left, and click on the "OK" button.
2) from your email account "attaching the backup gamefile"
select attach or attachments
you should see a browse button. Click on it.
you should see a list of locations on the left.
a) if you see "Desktop" and that is where you put the backup gamefile:
click on it and find "1107XX.lzh" in the list on the right side.
click on it and click on the "open" button or the "OK" button.
click on "attach" and "finish" or whatever you have to do to
complete the attachment process in your email account.
send email to me.
b) if you do not see "Desktop" or you put the backup gamefile
somewhere else click on the "Folders" button
Click on the icon to the left of "Desktop"
Click on the icon to the left of "Computer"
Click on the icon to the left of your hard drive "C:"
Click on the icon to the left of your high level folder "ACBLSCOR"
when you get to the last folder "GAMEFILE" click on it, not the icon
to the left, and find "1107XX.LZH" in the list of files on the right.
Click on the "open" or "OK" button.
click on "attach" and "finish" or whatever you have to do to
complete the attachment process in your email account.
send email to me.
Any questions please contact me.
ckchampion@netzero.net
Chris