FARGO Ratings

[JC]

How will the players be handled at BCA nationals who have low robustness numbers and skewed ratings?

I have a member of my league team with 50 games considered and a rating of 751. I am a 606 with over 500 games in and play a fair bit better than this guy. I would guess him to be about a 565 or so. Will it be a problem getting in the proper division to compete? Who makes the call? How can he convince them of his real level or does he just stay home until it's sorted out?

JC

 

[mikepage]

Mike,

Is there a bug in the program? I tried SVB vs Orcollo in a race to 99 and it gave me odds of infinity.

It worked with games in the 80s, but bot higher.

Gideon

We are aware of that. The quick fix is for you to do a race to 50 and then double everything.

What is happening is to compute SVB's chance of winning a race to 11, we need to compute
the chance that SVB wins 11 to 0
the chance that SVB wins 11 to 1
The chance that SVB wins 11 to 2
...
the chance that SVB wins 11 to 10

and add them up.

For each of these scores, we must count the number of ways it can happen. For the 11-0 score there is only one way (SVB wins 11 games in a row).
SSSSSSSSSSS

For the 11-1 score, there are 10 ways

DSSSSSSSSSSS (Dennis win's the first game)
SDSSSSSSSSSS (Dennis wins the second game)
SSDSSSSSSSSS
etc.

For the 11-6 score there are 462 ways

For long races the numbers get really big. There are ways around this; it just hasn't been a priority...

 

[ronscuba]

Yes, that's right. My guess is you were formerly classified as "open," and your wife as "advanced."

We entered 1 BCA national tournament in the team trophy division in 2014. Knocked out in 3rd round. I thought the trophy division was the lowest for coed teams. We play in a BCA league in NYC.

 

[Corwyn_8]

For the 11-6 score there are 462 ways

For long races the numbers get really big. There are ways around this; it just hasn't been a priority...

I don't want to tell you your business, but are you cacheing those choose function results?

Thank you kindly.

 

[watchez]

By the way - Jon Brown is a 735 -- which only puts him 24 points behind a Top 100 Pro in the world. Pretty good for the guy that won the Open singles last year.

So that leads me to two questions -- does this prove the Fargorate is viable and would have stopped Jon Brown from playing in the Open Singles last year? (And if you know him well too bad you didn't notice he was in the event to tip Mark and his staff off to block it.....just like Lancd Schofield can be found on many league player master lists.) Anyways......

Or does this show that Jon Brown has a huge inflated rating cause of his one year success in the open and really isn't a player deemed to have the skill level 24 points below a Top 100 male pro?

Can you respond to this here - maybe you didn't go back to the other thread.

 

[Jeff G. Martin]

By the way - Jon Brown is a 735 -- which only puts him 24 points behind a Top 100 Pro in the world. Pretty good for the guy that won the Open singles last year.

So that leads me to two questions -- does this prove the Fargorate is viable and would have stopped Jon Brown from playing in the Open Singles last year? (And if you know him well too bad you didn't notice he was in the event to tip Mark and his staff off to block it.....just like Lancd Schofield can be found on many league player master lists.) Anyways......

Or does this show that Jon Brown has a huge inflated rating cause of his one year success in the open and really isn't a player deemed to have the skill level 24 points below a Top 100 male pro?

Yes Jon's rating would be inflated from taking first. He only has 179 games in FargoRate, and it's safe to say that 70-80 of them are from Nationals.

 

[watchez]

Yes Jon's rating would be inflated from taking first. He only has 179 games in FargoRate, and it's safe to say that 70-80 of them are from Nationals.

If 200 games is the threshold to have a solid FargoRate rating, than 179 is close enough. Not much is gonna change if he played another 21 games. Especially if those 70-80 games were from the BCAPL nationals against like, Open player competition. The guy that won Open Singles is a hair away from being one of the Top 100 Male players in the world.

 

[Jeff G. Martin]

If 200 games is the threshold to have a solid FargoRate rating, than 179 is close enough. Not much is gonna change if he played another 21 games. Especially if those 70-80 games were from the BCAPL nationals against like, Open player competition. The guy that won Open Singles is a hair away from being one of the Top 100 Male players in the world.

But prior to the tournament, he would have only had around 100 games, thus very preliminary. With every match win his rating would go up quicker than if he started this tournament with 200 games. If he did in fact have only 100 games logged, then proceeded to win 10-12 matches straight, his rating would most likely be bloated from this tournament.

 

[watchez]

But prior to the tournament, he would have only had around 100 games, thus very preliminary. With every match win his rating would go up quicker than if he started this tournament with 200 games. If he did in fact have only 100 games logged, then proceeded to win 10-12 matches straight, his rating would most likely be bloated from this tournament.

He beat OPEN players --- he didn't beat anyone that was Top 1000 or Top 10,000 for that matter in the world, let alone Top 100 which is basically where he is ranked now.

Beyond that, I keep reading 200 games is the threshold. So it's the threshold unless you have a really good tournament??????? Point being there shouldn't be any quantification to it.

 

[Jeff G. Martin]

He beat OPEN players --- he didn't beat anyone that was Top 1000 or Top 10,000 for that matter in the world, let alone Top 100 which is basically where he is ranked now.

Beyond that, I keep reading 200 games is the threshold. So it's the threshold unless you have a really good tournament??????? Point being there shouldn't be any quantification to it.

Yes, 200 games is the point where you become established. And yes, if you have a really good tournaments during those 200 games, your rating might not be 100% accurate. If you take your best month of tournaments and base your rating solely off of that, it's most likely not an accurate representation of your normal speed.

I'm just answering your previous question of whether or not his rating can be largely inflated because of his success in the open division. If he only has 200 games, and nearly half of them were from one event during which he never lost a match, its probably not accurate yet. Add to that, it's very likely that almost all of his opponents were not established themselves. (Lance Schofield has so many games because a lot of them are from a Wisconsin rating system whose data was shared with FargoRate. But I would bet that others don't have very many. Or alternatively, if they were established with good ratings, he would gain even more points in those matches.)

 

[mikepage]

Yes, 200 games is the point where you become established. And yes, if you have a really good tournaments during those 200 games, your rating might not be 100% accurate. If you take your best month of tournaments and base your rating solely off of that, it's most likely not an accurate representation of your normal speed.

I'm just answering your previous question of whether or not his rating can be largely inflated because of his success in the open division. If he only has 200 games, and nearly half of them were from one event during which he never lost a match, its probably not accurate yet. Add to that, it's very likely that almost all of his opponents were not established themselves. (Lance Schofield has so many games because a lot of them are from a Wisconsin rating system whose data was shared with FargoRate. But I would bet that others don't have very many. Or alternatively, if they were established with good ratings, he would gain even more points in those matches.)

Thanks Jeff.

I have a number of comments.

Jon is at 735 with 179 games. I would not be surprised to see him settle in 20-40 points lower with more games, but who knows.

75 of his 179 games are from BCAPL Vegas this past summer where he didn't lose a match. His other hundred or so are all from 2009 and 2010, and because we weight older games lower than more recent games, those hundred act more like 25 games statistically. So he is not as close to 200 "current" games as it might look.

Jeff mentioned he might have played a lot against unrated or poorly rated opponents. While this is certainly a consideration, I think people might be surprised at how widespread our data collection has been. Below are Jon's matches in this tournament along with his opponent's rating and robustness. You will see opponents from many different states and provinces with established ratings. Lance S. has 529 games, 359 from our data collection and 170 shared with us from the Wisconsin effort.

There is a player from the Ukraine, Artem K., who played in a few Eurotour events and got to a rating of about 800 with 203 games. Now with 260 games or so he is down to 760. I think in that case about half his games were against poorly-rated opponents. So yes, that 203 games wasn't really 200 games against established opponents.

 

[BRussell]

There is a player from the Ukraine, Artem K., who played in a few Eurotour events and got to a rating of about 800 with 203 games. Now with 260 games or so he is down to 760.

It seems to me that poorly established ratings ought to regress to the mean a lot more - or in bayesian terms the prior probability ought to be weighted more than it appears to be now, if it is at all.

I noticed last week that there was a female player who had a rating in the 900s, but now she's off the list. I assume it was this same issue, where playing a small number of matches and doing exceptionally well produces an extreme value that doesn't make any sense (or it could have just been a simple error).

I know you're trying to do this by giving the ratings different colors, but I also think you could regress them by, say, 1/3 towards the average rating when they're not established.

 

[mikepage]

It seems to me that poorly established ratings ought to regress to the mean a lot more - or in bayesian terms the prior probability ought to be weighted more than it appears to be now, if it is at all.

It is not.

I noticed last week that there was a female player who had a rating in the 900s, but now she's off the list. I assume it was this same issue, where playing a small number of matches and doing exceptionally well produces an extreme value that doesn't make any sense (or it could have just been a simple error).

We had an error importing league data that we have since fixed.

 

[watchez]

Jon is just one example and i chose him cause he won the Open singles last year.
I don't care if he has old games, new games or no games. He shouldn't be ranked so high to be a top 100 male player in the world.

Anyways - what this tells me is that 200 games should not be the published threshold. I am a 571, haven't won more than two matches in singles the past two years and have a higher rating than people that beat me. But I am damned cause I have supported the BCAPL and played every year since 1998 while some new player with limited games in the system will get a walk thru.

 

[mikepage]

Jon is just one example and i chose him cause he won the Open singles last year.
I don't care if he has old games, new games or no games. He shouldn't be ranked so high to be a top 100 male player in the world.

Jon's rating is not yet established, and even if it was it is not at the "top 100 male player in the world" level.

Anyways - what this tells me is that 200 games should not be the published threshold. I am a 571, haven't won more than two matches in singles the past two years and have a higher rating than people that beat me. But I am damned cause I have supported the BCAPL and played every year since 1998 while some new player with limited games in the system will get a walk thru.

You are a 571 with 188 games. That's close to the middle of the "open" division in the past.

And guess what? In the last 5 years, you've played 25 matches in the OPEN; you've won 13 and you've lost 12.

What we're working very hard to do is to insure players like you do not run into brick walls. In 2014, your two loses were 1-5 to Mark Tafoya (NM), who is rated 670 with 859 games and 2-5 to Chad Behnke (IA) who is rated 696 with 215 games.

We cannot guarantee everything will be perfect; On the contrary we can guarantee it won't be perfect. But we are very confident the skill separation in the divisions will be much better this year than it has every been, and that it will get even better in the future.

More data makes us better. Players have an opportunity to enjoy a fantastic tournament AND to be part of the solution here. Come give it a try.

 

[watchez]

Jon's rating is not yet established, and even if it was it is not at the "top 100 male player in the world" level.




You are a 571 with 188 games. That's close to the middle of the "open" division in the past.

And guess what? In the last 5 years, you've played 25 matches in the OPEN; you've won 13 and you've lost 12.

What we're working very hard to do is to insure players like you do not run into brick walls. In 2014, your two loses were 1-5 to Mark Tafoya (NM), who is rated 670 with 859 games and 2-5 to Chad Behnke (IA) who is rated 696 with 215 games.

We cannot guarantee everything will be perfect; On the contrary we can guarantee it won't be perfect. But we are very confident the skill separation in the divisions will be much better this year than it has every been, and that it will get even better in the future.

More data makes us better. Players have an opportunity to enjoy a fantastic tournament AND to be part of the solution here. Come give it a try.

Yeah I tried my best to beat Behnke by safety breaking him but he was too polished having to play all his brothers over the years.

Can I assume you deem a top of the Open division to be a 600 rated player? I make this assumption based on the Team threshold being 3000 (3000 divided by 5 = 600)? If that is so, I won't mind playing in a singles event with players that truly are 600 or lower. I would mind playing against a Top 100 male player in the world.

 

[pogmothoin]

There is no BCAPL/USAPL close to me. I play in NAPA and the APA. So of course there will be no data for me or the many other players in my area.

My main point of interest is how Fargo Rate can be used to handicap local tournaments. It seems that over time this will be the fairest and most consistent way to rate players from different rooms.

Is there a way a local TD can enter data or forward data for inclusion in your system? Are there any plans to offer that ability to room owners or TD's? Thanks.

 

[tucson9ball]

Yeah I tried my best to beat Behnke by safety breaking him but he was too polished having to play all his brothers over the years.

Can I assume you deem a top of the Open division to be a 600 rated player? I make this assumption based on the Team threshold being 3000 (3000 divided by 5 = 600)? If that is so, I won't mind playing in a singles event with players that truly are 600 or lower. I would mind playing against a Top 100 male player in the world.

With the top 15% of entries playing Masters and the next 35% in advanced, I would bet a Fargo of 600 would probably be the Max for the Open.
Our local guys have been discussing the same topic for a month or so.
I'm going to send in my entry for singles and let my Fargo of 666 give em hell! LOL

 

[cleary]

Could someone tell me what matches are being analyzed to form these numbers? Is there a place we can view these databases?

 

[chevybob20]

There is no BCAPL/USAPL close to me. I play in NAPA and the APA. So of course there will be no data for me or the many other players in my area.

My main point of interest is how Fargo Rate can be used to handicap local tournaments. It seems that over time this will be the fairest and most consistent way to rate players from different rooms.

Is there a way a local TD can enter data or forward data for inclusion in your system? Are there any plans to offer that ability to room owners or TD's? Thanks.

Have you checked to see if your name is in the system? I did and was surprised that my name was in the system because I too have never played in a BCA match or in a BCA league. It didn't list a rating yet just my name.

I have only played in the APA league system and haven't played in that league for years. However, I do participate in the US Amateur Championships that is owned by the APA.

So, the APA must be sharing info.