info
Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss
Tournament bridge players inadequate vs. meaningful stats – current example
74 views
Skip to first unread message
Douglas
Nov 8, 2021, 12:22:44 PM11/8/21
to
There is a Bridgewinners thread with a CommonGame oddity posted recently, and a number of comments which are so off the wall, but conventional current thought, I cannot but speak.
The oddity commented on is nine consecutive deals where West is dealt an average of 16 HCPs.
My opening comment is “so what.” While it is unusual, but it has only a beginning significance to an inquisitive bridge playing pair. You do remember contract bridge is a partnership game?
For instance, if partner was dealt 4, or less, HCP’s in seven or eight of those boards, I would feel a little picked on in hindsight if I were E/W.
The hand record for that evening had 36 total boards. It happens E/W were playing significantly better HCP cards that evening, to the tune of almost 3 standard deviations. That translates to 17 double dummy games and slams possible for them by my count.
But again “so what.” Almost everyone gets to play the same boards. It is who gets the best results from that playing who gets the rewards. At least, that’s the theory.
My guess from these results is that the CommonGame is still using an older version of BridgeComposer to deal its boards. Its latest version’s (5.89) help file reveals (finally) what an inadequate dealing algorithm it used to use. If CommonGame updated to version 5.89, which supposedly uses BigDeal now, they would align with most of the rest of the tournament bridge world. At least the non-cherry picking part. And would probably create less future oddities.
Douglas
The oddity commented on is nine consecutive deals where West is dealt an average of 16 HCPs.
My opening comment is “so what.” While it is unusual, but it has only a beginning significance to an inquisitive bridge playing pair. You do remember contract bridge is a partnership game?
For instance, if partner was dealt 4, or less, HCP’s in seven or eight of those boards, I would feel a little picked on in hindsight if I were E/W.
The hand record for that evening had 36 total boards. It happens E/W were playing significantly better HCP cards that evening, to the tune of almost 3 standard deviations. That translates to 17 double dummy games and slams possible for them by my count.
But again “so what.” Almost everyone gets to play the same boards. It is who gets the best results from that playing who gets the rewards. At least, that’s the theory.
My guess from these results is that the CommonGame is still using an older version of BridgeComposer to deal its boards. Its latest version’s (5.89) help file reveals (finally) what an inadequate dealing algorithm it used to use. If CommonGame updated to version 5.89, which supposedly uses BigDeal now, they would align with most of the rest of the tournament bridge world. At least the non-cherry picking part. And would probably create less future oddities.
Douglas
James Dow Allen
Nov 9, 2021, 2:13:41 AM11/9/21
to
On Tuesday, November 9, 2021 at 12:22:44 AM UTC+7, Douglas wrote:
> There is a Bridgewinners thread with a CommonGame oddity posted recently, and a number of comments which are so off the wall, but conventional current thought, I cannot but speak.
>
> The oddity commented on is nine consecutive deals where West is dealt an average of 16 HCPs.
Given 32-board sessions, the chance that any of the players will have a nine-consecutive-hand streak
> There is a Bridgewinners thread with a CommonGame oddity posted recently, and a number of comments which are so off the wall, but conventional current thought, I cannot but speak.
>
> The oddity commented on is nine consecutive deals where West is dealt an average of 16 HCPs.
with an average HCP exceeding 15.0 is over 1.13%.
This might be a once-monthly occurrence at the rate some play bridge!
(Exceeding 15.5 hcp average is about 0.38%.) And the stated result is "cherry-picked" in a sense:
one could also get unlikely results over 8 consecutive hands, or over 10 hands, and so on.
This was found simulating 2 million such sessions. I used the Fisher-Yates shuffle, and a PRNG
better than the Mersenne Twister.
BTW: Douglas, Did you get my e-mail? I asked, per your offer, for any information you might
wish to share.
> Douglas
Cheers,
Gmail @ Jamesdowallen
0 new messages