Life In 19x19

Huge upset tonight in the first round of the Championship elimination... #1 seeded Ryan Li fell to #8 seeded Matthew Burrall. That makes them 1-1, and the best of three playoff in the first round will be tomorrow.

Meanwhile, Jeremy, Ben and Eric bested their opponents with 2-0 and advance to round two.

Only three games tomorrow morning, as both Jeremy and Yuan will have to wait to see who their next opponent will be between Ryan and Matthew.


and a rousing discussion after the game about AGA scoring.

Does that mean there was a disput?

hyperpape wrote:Does that mean there was a disput?

no. only in the online kibitz.

I see Ryan Li is through.

The finals are listed as best-of-3 or best-of-5. I understand the results of the round-robin are brought forward, but how would a best-of-5 arise?

pwaldron wrote:I see Ryan Li is through.

The finals are listed as best-of-3 or best-of-5. I understand the results of the round-robin are brought forward, but how would a best-of-5 arise?

Depends on how many time slots are available. If there are four slots then a best of 5 is played, one game being carried over from the round robin. If the knockout takes to many rounds to complete and there not time enough for four more games then a best of three is played.

Well, Ryan Li just won through to the finals. I am not sure that was the best result for him. Now he has to play a best of 5 with one game up. Had he lost and won the third game when it would have been a best of 3 and one game up.

While he was not guaranteed to win a third game in this semifinal, and so had to play to win, he has actually made it harder for himself to win the final, even though he is probably considered to be the favorite as he is already one up.

How will the lower places de decided?

Matti wrote:How will the lower places de decided?

Look at the results grid: http://www.usgo.org/aga-professional-results-2014

DrStraw wrote:Well, Ryan Li just won through to the finals. I am not sure that was the best result for him. Now he has to play a best of 5 with one game up. Had he lost and won the third game when it would have been a best of 3 and one game up.

While he was not guaranteed to win a third game in this semifinal, and so had to play to win, he has actually made it harder for himself to win the final, even though he is probably considered to be the favorite as he is already one up.

I was wondering the same thing this morning as I got dressed and I sat down to do the math when I had tea. I had thought that the best-of-five must automatically be better for Ryan, thinking that the probability the better player is more likely to win a best-of-n series as n increases and figuring that winning the first game of any series must be good.

As it turns out, it's not the case for realistic situations. If the probability of Ryan beating Eric in an individual game is less than 65% then he is better off with the best-of-three option. There is a crossover where the best-of-five is better: if Ryan's probability of winning a single game is greater than 65% then he's better with the longer series. Of course in that scenario he wins the series more than 90% of the time, so it's not really relevant.

This isn't to say that Ryan isn't the favourite. He looks to have something like a 60-80% chance of winning a best-of-five assuming he's at least as strong as Eric. With the best-of-three, though, he started at 75% and went up from there. Based on ratings it looks like Ryan has about a 70% chance of winning.

Ryan Li 1p

Congrats to Ryan Li 1p!

What does 'WD' mean? Did Zhou withdraw from the tournament?

Yes, due to "family emergency"

xed_over wrote:Ryan Li 1p

Thanks, xed_over for your work keeping us up to date!

pwaldron wrote:...
I was wondering the same thing this morning as I got dressed and I sat down to do the math when I had tea. I had thought that the best-of-five must automatically be better for Ryan, thinking that the probability the better player is more likely to win a best-of-n series as n increases and figuring that winning the first game of any series must be good.

As it turns out, it's not the case for realistic situations. If the probability of Ryan beating Eric in an individual game is less than 65% then he is better off with the best-of-three option. There is a crossover where the best-of-five is better: if Ryan's probability of winning a single game is greater than 65% then he's better with the longer series. ...

Two things.

1) My math is 66 and two thirds percent as the crossover point, not 65 percent. (actually, it is Excel's math using its negative binomial function assuming I did it right)


2) Maybe this is obvious, but I think your first point about Ryan being better off as N increases is eventually true, due to the law of large numbers (assuming his probability of winning a given game is >.5).