Normalizing averages

[hrothgar]

I disagree. Throwing out extreme data (2sd+ from the mean) will result in regression to the mean.

 

Regression towards the mean has a specific meaning.

It involves adding samples rather than discarding observations.

 

Please feel free to try again...

[mgoetze]

The imp table is designed around common bridge swings such as 480 vs 980, 170 vs 620 etc.

 

A possible solution would be to compare against the median rather than the mean.

[barmar]

A possible solution would be to compare against the median rather than the mean.

This might even be a good use for the least-used type of average, the mode, i.e. the most common result.

[Quartic]

A possible solution would be to compare against the median rather than the mean.

 

This might even be a good use for the least-used type of average, the mode, i.e. the most common result.

 

The problem here is when there are two normal, but significantly different results on the board, say +620 and +170.

[Wayne_LV]

What if?

 

Rather than generating hands using a random number generator, use hands that have been played in ACBL tournaments and/or club games. Compare results, not with the BBO field, but with the field that played in the ACBL game. I know that would involve substantial data entry but could this not be done with volunteers. To cross verify accuracy, assign a set of boards to 2 diff volunteers and then cross check the boards to make sure they match. If they don't match hand for hand, board for board, discard that set of board.

 

Another possiblility would be to use BBO records for hand hands played in BBO ACBL tournamnents for comparison and scoring in the "club" games. The likelyhood of someone playing a board in a tournament and getting the same board in a club game is next to nil, and if it did happen who would remember it?

 

I agree that the current method is not good, especially on weakends (pun intended) when weak fields are the norm.

[RunemPard]

I see a lot of people saying "start a team match...", yet, I never see any team matches to join. If I attempt to make a team match, no one ever joins to begin with.

 

I would like to see a better team match system.

[barmar]

What if?

 

Rather than generating hands using a random number generator, use hands that have been played in ACBL tournaments and/or club games. Compare results, not with the BBO field, but with the field that played in the ACBL game.

I think this would violate the bridge law that says that hands must be dealt randomly. You're allowed to ignore this law for special events, like Instant Matchpoint tournaments, but normal games should be random dealing.

[phil_20686]

I see a lot of people saying "start a team match...", yet, I never see any team matches to join. If I attempt to make a team match, no one ever joins to begin with.

 

I would like to see a better team match system.

 

You need to use the windows client to see lobby chat. There are always people trying to start (BBO) expert team games.

[helene_t]

Play matchpoints instead, that is slightly more roboust.