New Rating System

[mikepage]

Like i said, it could take a while. Looking at some of the ratings in my area, there's no way some of them could find action based on the fargo matchup, including some even races.

It's a little like seeing a billion-piece jigsaw puzzle come together. As the pieces fly into place, you start getting some colors and some boundaries, but you can't really make out what you're looking at. Then as more pieces fly into place you start to make out some patterns and then you recognize objects, etc...

So in an emerging area like OR, there are many people who have games against opponents who are poorly rated. So as more data comes in for those opponents and the algorithm has a better idea of what to make of your wins and loses against those people, YOUR rating becomes more accurate. So as more data comes in your rating gets better even if we don't get in more data for you.

So basically we understand what you are saying, and we don't sweat it: more data will fix it...

 

[mikepage]

I can believe that Mike. One problem I see is tracking identities. It seems like it would be easy to have different names, like Mike Page and Michael J. Page and M. John Page that you could use depending on where you're playing.

lol... We have this problem even WITHOUT people doing it intentionally.

 

[ShortBusRuss]

I don't know what CSI-sanctioned league you are talking about, but here in the Denver area, the (BCA) leagues are so competitive, that every single game counts. Our team has fielded a weak roster a week or two this session, and we are behind about 5 rounds, and playing the first place team tomorrow. We are planning on bringing all our killers and attempting to blank the team.

Any league that structures things in such a way that it would actually be advantageous to dump games is at the low end of the amateur spectrum, and to a certain extent, those player won't me much of a factor in any tournament they play. Harsh, but true.

But, to clarify my stance. I have played rated chess in the past, and I will say this.. The big chess tournaments have divisions that are rating-dependent. Other than those, there is no real point to playing rated chess games. But people do.. A LOT.

Why? Because we want to measure ourselves against those at our own level, and against those above us. It is why the rank amateur brags that they "only got beat with 2 balls on the table this time!!" It's why a better player might be proud that he got to 3 games on Shane Van Boening in the USBTC 8 ball championship...

At no time in the past has there been any realistic measurement in pool outside of overall tournament/gambling results that would definitively show whether a player had improved, and if so, by how much... Fargorate is that system. The more I hear about it, the more closely it seems to mimic the USCF (United States Chess Federation) rating system.

Fargorate could actually be just the thing that is a catalyst to get more overall participation in the game. Imagine a "most improved" trophy being given out in a Fargorate-measured league system. That could be won by anyone, no matter the rating.

I see any number of ways this could be used by room owners or pool promoters...

Short Bus Russ - C Player (Fargorate ~300?)

I half agree. Easier to track, yes. More accurate, debatable. In the CSI-sanctioned leagues around here, it is considered good strategy to lose the rest of the games once the winner for the night has been decided. FOR BOTH TEAMS. So you have both players in each of those games trying to lose, and thinking that's just strategy. Feed that into a "games won" system that creates relative ratings for everyone in the world, and see how accurate it becomes.

No system based strictly on measured data will work in a league setting where finishing way better than the rest of the teams in the league might put your team at a "disadvantage" with higher ratings. A league rating system MUST have a subjective component to be accurate, or the cheaters will always win. In fact, the absolute best rating system is completely subjective - one guy who sees every match played by every player and has sole authority to set the ratings. That system doesn't scale well, though.

I don't agree with the notion of lumping league results and tournament results together. I'm not saying Fargo can't generate good results, just that it's only as good as the data upon which it's based, a statement that's true for ANY rating system (including the LAST supposedly "most accurate" system used by CSI).

 

[ShortBusRuss]

An interesting thing happens when people start believing the ratings. I can tell you this from a lot of experience. Note that in the above scenario, the players had the respect of their peers. Their peers KNEW they played better than their assigned speed... When people around you start believing the assigned speed, things change A LOT.

I get contacted every week by pro players, by short-stop-speed players, by regional amateur players. EVERY SINGLE TIME the player seems to want to make sure he or she has gotten credit for the good tournaments or the good matches.... "You have me as a 660 when I KNOW I play at 700 speed.... What tournaments do you have for me ?.... "You have this person above me, and I don't believe that's right." "I double-dipped Sylver Ochoa back in 2009. Do you have that? This happens over and over again. I have yet, ever, to have someone wanting to make sure we got in the tournament where they went two and out.

Around here players are very interested in which players are 10 or 20 points higher than them... And Bob who is a 472 really wants to go above Pete, who is a 478. I ask people to give it some time to become entrenched in your area before becoming too attached to a view on this issue.

This is exactly the behavior I see in chess. When the statistics are all there, and you can pull up your rating breakdown to show your friends that you beat John Morra at Derby City Classic when you got in dead stroke for an hour, that is really worth something.

Being able to look back at your own results and definitively see that you played at this level in 2015, and now you play at this level in 2020, and oh, btw, you are averaging 3 more games per set against player X. Holy ish, I HAVE improved! I better get to practicing!!

The desire for incremental improvement becomes MUCH stronger when you know exactly what everyone around you is rated. Pool is too entrenched in the whole "gambling/tournaments is the only way to measure yourself"..

Guess what? It's not.

Short Bus Russ - C Player (Fargorate ~300?)

 

[robsnotes4u]

This is exactly the behavior I see in chess. When the statistics are all there, and you can pull up your rating breakdown to show your friends that you beat John Morra at Derby City Classic when you got in dead stroke for an hour, that is really worth something.



Being able to look back at your own results and definitively see that you played at this level in 2015, and now you play at this level in 2020, and oh, btw, you are averaging 3 more games per set against player X. Holy ish, I HAVE improved! I better get to practicing!!



The desire for incremental improvement becomes MUCH stronger when you know exactly what everyone around you is rated. Pool is too entrenched in the whole "gambling/tournaments is the only way to measure yourself"..



Guess what? It's not.



Short Bus Russ - C Player (Fargorate ~300?)

Great post!!


Sent from my iPad using Tapatalk

 

[Paul Schofield]

I get contacted every week by pro players, by short-stop-speed players, by regional amateur players. EVERY SINGLE TIME the player seems to want to make sure he or she has gotten credit for the good tournaments or the good matches.... "You have me as a 660 when I KNOW I play at 700 speed.... What tournaments do you have for me ?.... "You have this person above me, and I don't believe that's right." "I double-dipped Sylver Ochoa back in 2009. Do you have that? This happens over and over again. I have yet, ever, to have someone wanting to make sure we got in the tournament where they went two and out.

Around here players are very interested in which players are 10 or 20 points higher than them... And Bob who is a 472 really wants to go above Pete, who is a 478. I ask people to give it some time to become entrenched in your area before becoming too attached to a view on this issue.

I do believe you. I also think that your experiences are anecdotal. What you are describing is just not the culture of our game.

 

[Paul Schofield]

Many smart people, with good intentions, have tried, in vain, to prevent sandbagging by choosing the “right” statistics, the “right” mathematics and the “right” match-making methods. The best handicapping system possible cannot eliminate the incentive to sandbag when winning the prize depends on playing better than your rating indicates you should. And, the best system of policing (disqualifying) sandbaggers is likely to punish players for honest, rapid improvement. The statistics and mathematics used have little or nothing to do with whether or not people sandbag. It’s about the incentives.

I wrote this some time ago when a partner and I started working on a system we will introduce soon. Hope you'll like it.

Regards,
Don

Don, I can't agree with you more. In the worst way, I would like all prize money removed from my leagues. The most prestigious amateur events in all of sports have no cash prizes.

 

[SWRLE]

Consider some absolute measures of performance in pool...

---what fraction of games do you win against the 9-ball ghost?
---what is you straight-pool high-run after 10 attempts?
---what fraction of 8-ball games do you run out?
---what is your pot percentage on a spot shot?

and so forth...

Now contrast this to relative performance: John wins two thirds of the games against Joe, so John is 100 points above Joe.

At this point, John and Joe could be 800 and 700, or 600 and 400, or 300 and 200. We don't know, and we have no reference for either player to the absolute measures above.

John and Joe's ratings can shift up and down by any amount and still represent the data that John wins two thirds of the games against Joe.

Now add in Bill and George and Sue and Mary and Bob and Sergio and Mohammed and Xiaofeng.

What happens is the formerly arbitrary become more entrenched and resistant
to shifting. We can fix the scale in any way we want. Suppose we say the average of the ratings of the top 100 players in the world does not change.

A consequence of this is every absolute measure mentioned above has a fixed one-to-one correspondence with a rating...

So perhaps beating the 9-Ball ghost on a standard gold crown is a Fargo rating of 687.

Perhaps running out 20% of the 8-Ball racks on a Valley is a Fargo Rating of 617.

It no longer matters that the ratings are not DEFINED by absolute performance; what matters id the ratings can ASSOCIATED with absolute performance--and that, importantly, means a 630 in Erie PA is a 630 in Fargo ND.

I assume you meant 600 and 500. Two questions: on what data or facts is the assertion that a 100 difference means the higher rated player will win twice as many games as the lower rated player? Also, what evidence supports that ratio at all skill levels? In other words, if a 600 has a 2/3 probability of winning every game against a 500, does a 900 really have a 2/3 probabilty of winning every game against a 800? What data supports this assumption?

 

[Bob Jewett]

... Two questions: on what data or facts is the assertion that a 100 difference means the higher rated player will win twice as many games as the lower rated player? Also, what evidence supports that ratio at all skill levels? In other words, if a 600 has a 2/3 probability of winning every game against a 500, does a 900 really have a 2/3 probabilty of winning every game against a 800? What data supports this assumption?

As for the 100 points giving a 2:1 win ratio, that is by definition. It is not an assumption. That is the fundamental definition of what the ratings mean. The real follow-on question is whether two players 200 points apart have a 4:1 win ratio as expected. The data from the NPL -- a handicapped 9-ball league in the SF Bay Area -- indicates that they do.

 

[Corwyn_8]

As for the 100 points giving a 2:1 win ratio, that is by definition. It is not an assumption. That is the fundamental definition of what the ratings mean. The real follow-on question is whether two players 200 points apart have a 4:1 win ratio as expected. The data from the NPL -- a handicapped 9-ball league in the SF Bay Area -- indicates that they do.

Which I find quite an interesting result. I can't think of any inherent reason why that should follow, even if we had a perfect understanding of player speed.

Thank you kindly.

 

[Corwyn_8]

Outside the pool world, Facebook is free. Google is free. Yahoo mail is free. Etc. If those companies charged the end user for their service, I feel they would have never made it. Now, all those companies make a sh1t ton of money, of course. But the consumer does not directly pay any fees.

I think you are under a misapprehension about who the consumers of facebook are. If you have a facebook page, you aren't the consumer, you are the product. The consumers are the advertisers who are paying facebook (for eyeballs basically).

Thank You Kindly.

 

[mikepage]

Which I find quite an interesting result. I can't think of any inherent reason why that should follow, even if we had a perfect understanding of player speed.

Thank you kindly.

I agree. There are some underlying assumptions--like that we don't have paper-rock-scissor situations--that we can only really test by looking at the results.

Here are some of those interesting results. We can do a small case study on ME. I have a rating of 632. That rating is based on 5500 games played against nearly 400 different opponents, with no more than a few percent of my games against any particular opponent.

There are six players (all of whom themselves have thousands of games against many opponents determining their ratings) against whom I've played at least 100 games total--generally several tournament matches and perhaps some league games.... I'll show the expected versus actual against those, and then I'll include my teammate Linda Zsedeney to get a bigger rating range



Larry Wentz (500) expectation, 174 to 69: actual, 179 to 64
Rory Hendrickson (725) expectation, 40 to 76; actual 40 to 76
Dustin Hansen (576) expectation 69 to 46; actual 79 to 36
Bruce Wheeler (610) expectation 62 to 53; actual 67 to 48
Tim Blinn (576) expectation 67 to 45; actual 64 to 48
Gary Dandurand (517) expectation 72 to 32; actual 75 to 29
Linda Zsedeney (434) expectation 55 to 15; actual 54 to 16

These are all close with the exception of games against Dustin Hansen, who has moved up about 50 points since most of those games were played.

Now if you look at Rory (you've already seen he beats me 76 to 40), we can see how he does against a few others. He is a 725, and he's played 38 games against Justin Bergman (795) and 43 games against Dennis Orcollo (813)

Bergman (795) expectation 14 to 24; actual 14 to 24
Orcollo (813) expectation 15 to 28; actual 14 to 29

 

[bwally]

I agree. There are some underlying assumptions--like that we don't have paper-rock-scissor situations--that we can only really test by looking at the results.

Here are some of those interesting results. We can do a small case study on ME. I have a rating of 632. That rating is based on 5500 games played against nearly 400 different opponents, with no more than a few percent of my games against any particular opponent.

There are six players (all of whom themselves have thousands of games against many opponents determining their ratings) against whom I've played at least 100 games total--generally several tournament matches and perhaps some league games.... I'll show the expected versus actual against those, and then I'll include my teammate Linda Zsedeney to get a bigger rating range



Larry Wentz (500) expectation, 174 to 69: actual, 179 to 64
Rory Hendrickson (725) expectation, 40 to 76; actual 40 to 76
Dustin Hansen (576) expectation 69 to 46; actual 79 to 36
Bruce Wheeler (610) expectation 62 to 53; actual 67 to 48
Tim Blinn (576) expectation 67 to 45; actual 64 to 48
Gary Dandurand (517) expectation 72 to 32; actual 75 to 29
Linda Zsedeney (434) expectation 55 to 15; actual 54 to 16

These are all close with the exception of games against Dustin Hansen, who has moved up about 50 points since most of those games were played.

Now if you look at Rory (you've already seen he beats me 76 to 40), we can see how he does against a few others. He is a 725, and he's played 38 games against Justin Bergman (795) and 43 games against Dennis Orcollo (813)

Bergman (795) expectation 14 to 24; actual 14 to 24
Orcollo (813) expectation 15 to 28; actual 14 to 29

Wow, those are extremely accurate. Impressive Mike! :thumbup:

 

[robsnotes4u]

I agree. There are some underlying assumptions--like that we don't have paper-rock-scissor situations--that we can only really test by looking at the results.



Here are some of those interesting results. We can do a small case study on ME. I have a rating of 632. That rating is based on 5500 games played against nearly 400 different opponents, with no more than a few percent of my games against any particular opponent.



There are six players (all of whom themselves have thousands of games against many opponents determining their ratings) against whom I've played at least 100 games total--generally several tournament matches and perhaps some league games.... I'll show the expected versus actual against those, and then I'll include my teammate Linda Zsedeney to get a bigger rating range







Larry Wentz (500) expectation, 174 to 69: actual, 179 to 64

Rory Hendrickson (725) expectation, 40 to 76; actual 40 to 76

Dustin Hansen (576) expectation 69 to 46; actual 79 to 36

Bruce Wheeler (610) expectation 62 to 53; actual 67 to 48

Tim Blinn (576) expectation 67 to 45; actual 64 to 48

Gary Dandurand (517) expectation 72 to 32; actual 75 to 29

Linda Zsedeney (434) expectation 55 to 15; actual 54 to 16



These are all close with the exception of games against Dustin Hansen, who has moved up about 50 points since most of those games were played.



Now if you look at Rory (you've already seen he beats me 76 to 40), we can see how he does against a few others. He is a 725, and he's played 38 games against Justin Bergman (795) and 43 games against Dennis Orcollo (813)



Bergman (795) expectation 14 to 24; actual 14 to 24

Orcollo (813) expectation 15 to 28; actual 14 to 29

That is great real world information


Sent from my iPad using Tapatalk

 

[Bob Jewett]

Which I find quite an interesting result. I can't think of any inherent reason why that should follow, even if we had a perfect understanding of player speed.

Thank you kindly.

Actually, there is a fairly simple and reasonable model that would explain the ratio results. Suppose that the underlying factor was how many games on average a player wins per turn at the table. If that number is more or less independent of who they are playing, then the result is that the ratios work. I would write out the full proof but the margins on this page do not give enough space.

 

[Paul Schofield]

Mike, For the most part, everyone in my room knows how everyone else plays. This should not surprise anyone. Let's say that Fargo Rate comes along and confirms what players know. How does this increase casual play? The lifeblood of the game is not leagues. It is not tournaments. It is an abundance of brisk casual play by a lot of players.

 

[mikepage]

Mike, For the most part, everyone in my room knows how everyone else plays. This should not surprise anyone. Let's say that Fargo Rate comes along and confirms what players know. How does this increase casual play? The lifeblood of the game is not leagues. It is not tournaments. It is an abundance of brisk casual play by a lot of players.

If I figure this out, I will be sure to let you know.

 

[bwally]

Mike, For the most part, everyone in my room knows how everyone else plays. This should not surprise anyone. Let's say that Fargo Rate comes along and confirms what players know. How does this increase casual play? The lifeblood of the game is not leagues. It is not tournaments. It is an abundance of brisk casual play by a lot of players.

I will try to take an angle on this. It could give an individual a measurable target to try to improve on. Example, when I first looked up my Fargo Rating it was 602, in the past 4 weeks it has dropped under 600 (without me playing a game). It will drive to practice hard so I get above the 600 mark. Then the next target will be 650 and so on.

Of course this is just my opinion and we will have to see what the results from Fargo will be. My guess is that it will be great. :thumbup:

 

[robsnotes4u]

I will try to take an angle on this. It could give an individual a measurable target to try to improve on. Example, when I first looked up my Fargo Rating it was 602, in the past 4 weeks it has dropped under 600 (without me playing a game). It will drive to practice hard so I get above the 600 mark. Then the next target will be 650 and so on.



Of course this is just my opinion and we will have to see what the results from Fargo will be. My guess is that it will be great. :thumbup:

Fargo Rate can be to pool like Fitbit is to Fitness


Sent from my iPhone using Tapatalk

 

[Paul Schofield]

If I figure this out, I will be sure to let you know.

I am counting on you. Pool needs to book a winner.