[hrothgar]

I've been playing in a number of day longs over the past month or two and have observed what appears to be a ridiculous amount of variance in both my own scores as well as those of some of my friends.

One day I am coming in in the top 5, the next I am well below average, the next I am in the middle of the pact.
The same holds true for the scores of a number of other folks here on the forums.

While I do believe that I have "on days" and "off days", the sheer magnitude of the swings is rather disturbing.
I believe that this is related to a combination of

• Relatively short tournaments (12 boards or less)
  
• Insufficient numbers of comparisons

Simply put, if a board is only being played 12 or 16 times, there's way too much luck in the results.

Given that the BBO platform is being used to run events like the ACBL's National Level individual )and all the other ACBL events, I really think that some work is need to increase the number of comparisons for each board...

[ArtK78]

In the "major" events that you refer to, there are a significant number of comparisons. And the length of the tournament is much longer than 12 boards.

As for the number of comparisons, sometimes that is small even in significant events. In my district's NAOP Flight "A" finals (Day 2, half the field surviving), there were 7 tables, resulting in a 6 top. While the field was strong, there is certainly a luck element when there are only 7 comparisons.

[pescetom]

If a duplicate bridge session typically consists of between 24 and 28 boards in total, there is a good mathematical reason, not just because people want to play for 3 hours.

[barmar]

pescetom, on 2019-January-15, 16:16, said:

If a duplicate bridge session typically consists of between 24 and 28 boards in total, there is a good mathematical reason, not just because people want to play for 3 hours.

Unfortunately, online players are generally not as interested in devoting as much time. People typically play online for quick diversions, they don't want to devote several hours for a whole session like that. That's why we had to create speedballs. Even 90 minutes for 12 boards was too long for many players. BBO used to have both speedballs and "slowballs", the latter got very poor attendance.

But the number of comparisons is generally considered adequate. In f2f bridge, it's very common to use sections with 13 tables, and this is considered adequate. BBO uses 14-15 tables for most sections. Lots of f2f bridge clubs have small games with only 5-8 tables; while this definitely increases variance, the results are still considered meaningful, and usually run true to expectations (my regular bridge club is like this, and the best pairs win pretty regularly). Larger, more important tournaments do matchpointing across multiple sections to improve accuracy, but the difference is rarely more than a percent or two. And it's not uncommon for the final day of a multi-day event like Blue Ribbon Pairs to reduce the field to just 1 or 2 sections.

[hrothgar]

barmar, on 2019-January-15, 16:37, said:

But the number of comparisons is generally considered adequate. In f2f bridge, it's very common to use sections with 13 tables, and this is considered adequate. BBO uses 14-15 tables for most sections. Lots of f2f bridge clubs have small games with only 5-8 tables; while this definitely increases variance, the results are still considered meaningful, and usually run true to expectations (my regular bridge club is like this, and the best pairs win pretty regularly). Larger, more important tournaments do matchpointing across multiple sections to improve accuracy, but the difference is rarely more than a percent or two. And it's not uncommon for the final day of a multi-day event like Blue Ribbon Pairs to reduce the field to just 1 or 2 sections.

The fact that F2F bridge clubs have constraints wrt the number of attendees is hardly an excuse to deliberately implement a poor solution.
(Unless, of course, dispensing master points via a lottery is viewed as a feature rather than a bug)

I'll track my numbers in the ACBL MP online over the next week or two and show the sheer amount of variance involve...
Its quite ridiculous.

[barmar]

hrothgar, on 2019-January-16, 06:34, said:

The fact that F2F bridge clubs have constraints wrt the number of attendees is hardly an excuse to deliberately implement a poor solution.
Its quite ridiculous.

But I didn't only compare to f2f clubs, I also compared to f2f tournaments, which don't have such constraints. I've played in many NABC events where the Bridgemate showed our section award at the end of the round, and then later I got the across-the-field result. It rarely changes by more than 2%, and usually less.

Of course, individual boards will show more variance, but when you average over 26 boards the final result doesn't change too much.