[hrothgar]

Sucked in by the promise of a tee shirt that I I doubt that I can ever win and would not fit even if I did, I ventured into some of BBO's Daily tournaments this weekend...

I was shocked to see that one needs scores in the low 80s to win these events. I understand that these tournaments are something of a lottery, however, this really seems to be pushing things a bit far. It feels as if winning the event requires both

1. Playing hands really really well (nothing wrong with that)
2. Getting dealt hands that allow one to score a top (or alternatively, adopting a deliberately high variance strategy that increases the likelihood that you end up with a top/bottom

I have some concerns that the pendulum might have swung a bit far and the luck of th draw with respect to the hands that one is dealt might be outweighing other factors.

I was hoping that one of the BBO types might be able to provide me with some datasets that I could play with.

Ideally, I'd like a couple files in CSV format.

Each file would contain a matrix where columns show different boards that were played across a large number of tournaments.
Each row would show the %score that different players had on each board

From the looks of things, the tournaments have roughly 29 or so people playing each board (sometimes less). If a board gets played a smaller numbers of time, please pad with a "NaN"

Ideally, I'd like to have one file for the 8 board Daylong tournaments and a second file for the 12 board ACBL tournaments. (If you have the data available, getting a third file with the 24 board Pajama Nationals would be nice)

A large data set would be best (at least 1000 boards, 10K would be even better)

FWIW, here's what I would be interested in looking at

1. Is the variance in the board results the same across tournament types or different? My guess is that the variance will be significantly higher in the 8 board mini tournaments, but it would be nice to take a look at this. (Note: This belief is based on a variety of factors including a belief that people will be shooting more in short events, cheaper cost attracts weaker players, and the fact that the pajama nationals has multiple rounds... Actually, now that I think about it, having three different files for the pajama nationals would be even better)

2. I'd also like to use this information to run some monte carlo simulations and try to understand whether winning events reflects playing well versus getting dealt the right hands.

[barmar]

hrothgar, on 2019-April-20, 17:01, said:

Sucked in by the promise of a tee shirt that I I doubt that I can ever win and would not fit even if I did, I ventured into some of BBO's Daily tournaments this weekend...

Those aren't the tourneys we're using for this giveaway. You have to play in the 8-board Robot Duplicate tourneys, not the Daylong tourneys. These generally only have 15-30 players, and winning scores are not as extreme. You're also all playing the same 8 boards, not using deal pools like the daylongs.

But I'll see if I can send you the data you're looking for.

[hrothgar]

barmar, on 2019-April-20, 17:36, said:

Those aren't the tourneys we're using for this giveaway. You have to play in the 8-board Robot Duplicate tourneys, not the Daylong tourneys. These generally only have 15-30 players, and winning scores are not as extreme. You're also all playing the same 8 boards, not using deal pools like the daylongs.

But I'll see if I can send you the data you're looking for.

Thanks!

[steve2005]

The daylongs are funny you finally do well get 65% and your not even in top 100! lol

[uday]

Some daylongs , esp the Mon-Tue-Wed ones, have over 10,000 entrants. I'm guessing it takes more than 65% to get into the top 100 -- it's a short game ( 8 boards ) so volatility plays a strong role, I imagine.

[hrothgar]

uday, on 2019-April-21, 13:41, said:

Some daylongs , esp the Mon-Tue-Wed ones, have over 10,000 entrants. I'm guessing it takes more than 65% to get into the top 100 -- it's a short game ( 8 boards ) so volatility plays a strong role, I imagine.

Hi Uday

Thanks for that data point. One of the things that I am hoping to better understand is the amount of variance that is observed in individual boards.
In turn, I can use that to try and understand the distribution of results and just what a 4 sigma or 5 sigma event looks like.

[barmar]

uday, on 2019-April-21, 13:41, said:

Some daylongs , esp the Mon-Tue-Wed ones, have over 10,000 entrants. I'm guessing it takes more than 65% to get into the top 100 -- it's a short game ( 8 boards ) so volatility plays a strong role, I imagine.

Are you talking about the free daylongs? Those probably get more entrants early in the week because of the limit of 3 per week. So if most people play Monday, Tuesday, and Wednesday they won't be able to play any more that week, and the remaining days will get fewer entrants.

But we removed this limit for people who use the new HTML version, so we'll hopefully see more entries later in the week now.