Us open 10-ball tournament--fargo ratings analysis

mikepage

mikepage

AzB Silver Member
Silver Member
I'll reply to this post with the Fargo Ratings of all the players in the US Open 10-ball championships that just finished in Vegas. But first let me explain briefly what they mean. If someone got far into the tournament by playing easier opponents or winning many close matches, then they're going to have a lower rating than other players who finish the same. Likewise, a player who goes two-and-out losing close matches to high-rated players can end up rated pretty highly.

The player who wins the tournament does not always--in fact frequently does not--have the highest rating. And yes, that is the case here.

It's the DIFFERENCE between two player's Fargo Ratings that matters

With a Rating Difference of 17 points, an expected score is 9-8 in favor of the higher rated player. Then other rating differences mean the following:

17 points 9-8
36 points 9-7
58 points 9-6
85 points 9-5
117 points 9-4
158 points 9-3
217 points 9-2
317 points 9-1

These ratings take into account ONLY THIS TOURNAMENT.

Also, these ratings treat every game the same. So if a player is particularly good or particularly poor at winning an occasional key game, that won't show up here.

The ratings I report here are the SELF CONSISTENT Fargo Ratings. So if each player had this rating going INTO the tournament, the rating update procedure in the video below would suggest no changes after the tournament.

This video talks about 8-ball, but it's the same idea for 10-ball.

http://www.youtube.com/watch?v=CRWECaLnaxY
 
thanks!

Mike,

Thanks for posting. No real surprises among the top finishers, their Fargo ratings were in pretty much the same order as how I thought they played. Interesting to see some of the folks with poor Fargo ratings for this one tournament though.

Hu
 
Neat data, Mike. Thanks for doing that.

Here are some comments regarding the ratings for the women:

  • Ga Young Kim was the only woman with a Fargo Rating (rank of 51) in the top half of the field.
  • For 8 of the 11 women (including Kim), their Fargo Ratings rankings fell within the band where they placed in the tournament. For example, Ouschan's 105 Fargo Rating rank was within the 97-128 band where she actually placed.
  • Two of the women -- Cha and Little -- had Fargo Ratings rankings (102 and 69, respectively) that were worse than their actual finish (65-96 and 49-64, respectively).
  • One of the women -- Baretta -- had a Fargo Rating ranking (83) that was better than her actual finish (97-128).
  • By their Fargo Ratings, Shane Van Boening would be expected to beat Kelly Fisher 9-1. I imagine she would be happy to play him for big money at that game.
 
Last edited:
Interesting for one tourney. Keep a running count for the major tourneys or analyze some old data from the last couple years and see how these numbers average out.
 
Questions for Mike (or anyone else).

The interesting thing about these numbers to me is that the difference of ratings between some of the name players is quite big. The relative standings of the players isn't too surprising, but these numbers suggest some expected results that are quite surprising. Take Shane Van Boening against players like Deuel, Hohmann, Schmidt, or Kiamco and the expected result would be around 9-5 or 9-4.

Now, based on the form they were in this tournament, that's a fair prediction, and that's what Fargo Ratings give us. But I'm sure SVB wouldn't have given 4 or 5 games on the wire in a race to 9 had any of these guys asked for a challenge match during the tournament or right after it.

So my question is: do you think that we are (or just me) biased to think that these match-ups are closer than they really are? Or is it more an issue about the form they were in this particular tournament? And if we forget the form they happened to be in this particular tournament, do you think that the number of games played gives a result accurate enough to make these sort of predictions?

I do trust that the system is mathematically sound and gives the best estimate based on these particular results. But I'm also sure that most people will find these expected results quite far from what they think would have been a fair handicap. (Had they played a handicapped match.)
 
Last edited:
JarnoV said:
[...]
So my question is: do you think that we are (or just me) biased to think that these match-ups are closer than they really are? Or is it more an issue about the form they were in this particular tournament? And if we forget the form they happened to be in this particular tournament, do you think that the number of games played gives a result accurate enough to make these sort of predictions?
Click to expand...


I think it's both.

On the one hand, these are definitely far "spread out" relative to what the true long-termed ratings would be. That is, if I add three or four more tournaments to the mix, we might expect Shane to come down from 875 to 820, Corey to stay the same at 781, and Jasmin to move up over 700.

Imagine we cloned Sarah Rousey so that there were 128 identical copies of her that faced off in a tournament. A quarter of the Sarah's would go two and out, and some of those would lose 9-5 and 9-3 or something. One of those Sarah's would win the tournament. So even though we know these 128 players have the same rating in the long haul, the single tournament would produce a big artificial spread.

If you or anybody entered our system locally, we wouldn't do this self consistent procedure. Instead we would do the update procedure described in the video, which is more stable and wouldn't have people move so much after a few matches.

I do trust that the system is mathematically sound and gives the best estimate based on these particular results. But I'm also sure that most people will find these expected results quite far from what they think would have been a fair handicap. (Had they played a handicapped match.)
Click to expand...

Yes I agree.
 
mikepage said:
On the one hand, these are definitely far "spread out" relative to what the true long-termed ratings would be. That is, if I add three or four more tournaments to the mix, we might expect Shane to come down from 875 to 820, Corey to stay the same at 781, and Jasmin to move up over 700.
Click to expand...

On that note, would you be willing to compile a Fargo Rating list from more tournaments if someone else did the hard work of gathering the results into a format that you could easily use?

On a quick glance, at least World 10-Ball 2009 has complete records of the matches and Predator 10-Ball Tour too. Unfortunately it seems like the Mezz Hard Times 10-Ball tournament doesn't have complete records. Maybe I could try to contact someone to see if they're available in some format.

Since I don't expect anyone else to volunteer, that someone would probably be me. :-)
 
JarnoV said:
On that note, would you be willing to compile a Fargo Rating list from more tournaments if someone else did the hard work of gathering the results into a format that you could easily use?

On a quick glance, at least World 10-Ball 2009 has complete records of the matches and Predator 10-Ball Tour too. Unfortunately it seems like the Mezz Hard Times 10-Ball tournament doesn't have complete records. Maybe I could try to contact someone to see if they're available in some format.

Since I don't expect anyone else to volunteer, that someone would probably be me. :-)
Click to expand...

That sounds good. I'll send you the format later today.
 
AtLarge said:
[...]By their Fargo Ratings, Shane Van Boening would be expected to beat Kelly Fisher 9-1. I imagine she would be happy to play him for big money at that game.
Click to expand...

The rating gap between them surely is bigger than it should be, mostly because Kelly didn't play that many games and probably underperformed on the games she did play.

But there's another subtlety I've been glossing over.

If that rating difference was real, then that means SVB would win games at 9 times the rate of Kelly, or have a 0.9 chance of winning each game. This is not the same thing as 9-1 being a fair match. So I shouldn't have stated it that way.

Saying that after 10 games the most lkely score is 9-1 is not the same thing as 9-1 being a fair matchup. To win the 9-1 matchup, SVB would need to win 9 games in a row. His chance of doing that -assuming his chance of winning a single game is 0.9--is 38.7%.

Many people don't know this subtlety. Say you have someone for whom you regularly spot 8 games to 16, and after playing a large number of sets you split them. Could you give that person 3 on the wire to 6? or 2 on the wire to 4? or 1 on the wire to 2?. Often the better player rejects this just because of the volatility of the short sets but assumes that if the weaker player agreed to 100 sets playing 2-1, things would be even. But they wouldn't. The weaker player has an advantage.
 
mikepage said:
The rating gap between them surely is bigger than it should be, mostly because Kelly didn't play that many games and probably underperformed on the games she did play.

But there's another subtlety I've been glossing over.

If that rating difference was real, then that means SVB would win games at 9 times the rate of Kelly, or have a 0.9 chance of winning each game. This is not the same thing as 9-1 being a fair match. So I shouldn't have stated it that way.

Saying that after 10 games the most lkely score is 9-1 is not the same thing as 9-1 being a fair matchup. To win the 9-1 matchup, SVB would need to win 9 games in a row. His chance of doing that -assuming his chance of winning a single game is 0.9--is 38.7%.

Many people don't know this subtlety. Say you have someone for whom you regularly spot 8 games to 16, and after playing a large number of sets you split them. Could you give that person 3 on the wire to 6? or 2 on the wire to 4? or 1 on the wire to 2?. Often the better player rejects this just because of the volatility of the short sets but assumes that if the weaker player agreed to 100 sets playing 2-1, things would be even. But they wouldn't. The weaker player has an advantage.
Click to expand...

If after 10 games, the most likely score would be 9-1, then wouldn't a 10-2 race be ideal? If both players perform as the ratings predict, the match would be expected to go hill-hill. Of course, in a hill-hill match the stronger player would be favored to win the final game, but isn't that as it should be? Applying a handicap so that the match is expected to go hill-hill seems like the fairest way to do it, to me.

-Andrew
 
JarnoV said:
You forgot to send this?

I've got the world 10b 2009 almost ready now. ;-)
Click to expand...

Whoa --I'm really sorry.
Yes I forgot, and then I went out of town for a few days.

PM me an email address, and I'll send you a file & some instructions
 
Here are Fargo Ratings for both WPC 10-ball 2009 and US Open 10-ball tournament. I've included here only the players that have a robustness over 100, but here's a link to a list that contains all players that have robustness over 50.

Comments?

(Also, if anyone knows other recent big 10-ball tournaments that have complete records of matches available I'd be interested to know.)

PHP: 
Name               Rating Robustness
Francisco Bustamante  850 139
Hi-Wen Lo             838 120
Shane Van Boening     832 195
Dennis Orcollo        826 165
Mika Immonen          816 224
Lee Van Corteza       813 268
Darren Appleton       811 183
Manny Chau            811 103
Rodney Morris         800 106
Edwin Montal          799 105
Thomas Engert         785 153
Charlie Williams      783 150
Ralf Souquet          782 151
Roberto Gomez         780 131
Tony Drago            779 113
David Alcaide         770 145
Dennis Hatch          770 118
Warren Kiamco         767 141
Po-Cheng Kuo          751 113
Thorsten Hohmann      750 118
Marcus Chamat         743 154
Corey Deuel           737 113
Imran Majid           723 117
Oscar Dominguez       722 108
Tyler Edey            717 108
Chris Melling         717 101
Johnny Archer         712 142
Toru Kuribayashi      707 115
Daryl Peach           695 129
Shaun Wilkie          692 109
Huidji See            678 109
Ben Nunan             677 122
 
JarnoV said:
Here are Fargo Ratings for both WPC 10-ball 2009 and US Open 10-ball tournament. I've included here only the players that have a robustness over 100, but here's a link to a list that contains all players that have robustness over 50.

Comments?

(Also, if anyone knows other recent big 10-ball tournaments that have complete records of matches available I'd be interested to know.)

PHP: 
Name               Rating Robustness
Francisco Bustamante  850 139
Hi-Wen Lo             838 120
Shane Van Boening     832 195
Dennis Orcollo        826 165
Mika Immonen          816 224
Lee Van Corteza       813 268
Darren Appleton       811 183
Manny Chau            811 103
Rodney Morris         800 106
Edwin Montal          799 105
Thomas Engert         785 153
Charlie Williams      783 150
Ralf Souquet          782 151
Roberto Gomez         780 131
Tony Drago            779 113
David Alcaide         770 145
Dennis Hatch          770 118
Warren Kiamco         767 141
Po-Cheng Kuo          751 113
Thorsten Hohmann      750 118
Marcus Chamat         743 154
Corey Deuel           737 113
Imran Majid           723 117
Oscar Dominguez       722 108
Tyler Edey            717 108
Chris Melling         717 101
Johnny Archer         712 142
Toru Kuribayashi      707 115
Daryl Peach           695 129
Shaun Wilkie          692 109
Huidji See            678 109
Ben Nunan             677 122
Click to expand...
Lol skinny li'l Lee Van Highest number in "Robustness"
 
You must log in or register to reply here.
Back
Top