Do fargo ratings change fast enough?

Jude Rosenstock · Aug 14, 2019

I'm currently a league director here in Austin and we report directly through Fargo's League Management System so my players' FargoRatings are updated regularly. My league has about 80 players.

I think the best way to describe this is, robustness is key. Players with high robustness will act like stones in a stream. They won't move but everyone around them will. My robustness is reaching 1k now and players with Robustness under 200 will change dramatically after playing me.

Also, starter ratings are important. Players are given a starter rating which is a somewhat arbitrary process but if it's incredibly inaccurate, their FargoRating will change quickly. For me, my starter rating was 625 which was pretty close but a little low. As a result, my FargoRating has been as high as 705 but never lower than 625 and has been hovering around 650. Had I started higher or lower, my changes would have likely been more dramatic.

Some of my league players were given starter ratings that were completely wrong and that's impacted their overall score. As a league director, FargoRate gives some background data so you can investigate how players are actually trending. That gives me an opportunity to make an adjustment which really is only applicable when their robustness is low. Once they're established, the system completely takes over and their play determines any changes.

That said, I find the FargoRate team to be very responsive and diligent. They're working around the clock to insure the data is correct and applied in the best possible fashion. There's a lot of data and a lot of the scoring mechanisms are still done on paper so I would imagine they get a lot of data-entry work but as we progress as an industry, hopefully, this process will become more automated.

alphadog · Aug 14, 2019

Jude Rosenstock said:
I'm currently a league director here in Austin and we report directly through Fargo's League Management System so my players' FargoRatings are updated regularly. My league has about 80 players.

I think the best way to describe this is, robustness is key. Players with high robustness will act like stones in a stream. They won't move but everyone around them will. My robustness is reaching 1k now and players with Robustness under 200 will change dramatically after playing me.

Also, starter ratings are important. Players are given a starter rating which is a somewhat arbitrary process but if it's incredibly inaccurate, their FargoRating will change quickly. For me, my starter rating was 625 which was pretty close but a little low. As a result, my FargoRating has been as high as 705 but never lower than 625 and has been hovering around 650. Had I started higher or lower, my changes would have likely been more dramatic.

Some of my league players were given starter ratings that were completely wrong and that's impacted their overall score. As a league director, FargoRate gives some background data so you can investigate how players are actually trending. That gives me an opportunity to make an adjustment which really is only applicable when their robustness is low. Once their established, the system completely takes over and their play determines any changes.

That said, I find the FargoRate team to be very responsive and diligent. They're working around the clock to insure the data is correct and applied in the best possible fashion. There's a lot of data and a lot of the scoring mechanisms are still done on paper so I would imagine they get a lot of data-entry work but as we progress as an industry, hopefully, this process will become more automated.

Great reply. Thank you for taking the time to put a explanation of how you see fargorate applied to your league.

Jude Rosenstock · Aug 14, 2019

alphadog said:
Great reply. Thank you for taking the time to put a explanation of how you see fargorate applied to your league.

Thank you! And as an FYI, my league was in session last night. I entered scores immediately afterward and my FargoRating updated today (I just checked). After beating a player with a FargoRating of 570 by the score of 8-4 (hill-hill), my FargoRating increased by 1 point. My FargoRating typically doesn't change by more than a point or two after each match.

poolscholar · Aug 14, 2019

Jude Rosenstock said:
I'm currently a league director here in Austin and we report directly through Fargo's League Management System so my players' FargoRatings are updated regularly. My league has about 80 players.

I think the best way to describe this is, robustness is key. Players with high robustness will act like stones in a stream. They won't move but everyone around them will. My robustness is reaching 1k now and players with Robustness under 200 will change dramatically after playing me.

Also, starter ratings are important. Players are given a starter rating which is a somewhat arbitrary process but if it's incredibly inaccurate, their FargoRating will change quickly. For me, my starter rating was 625 which was pretty close but a little low. As a result, my FargoRating has been as high as 705 but never lower than 625 and has been hovering around 650. Had I started higher or lower, my changes would have likely been more dramatic.

Some of my league players were given starter ratings that were completely wrong and that's impacted their overall score. As a league director, FargoRate gives some background data so you can investigate how players are actually trending. That gives me an opportunity to make an adjustment which really is only applicable when their robustness is low. Once they're established, the system completely takes over and their play determines any changes.

That said, I find the FargoRate team to be very responsive and diligent. They're working around the clock to insure the data is correct and applied in the best possible fashion. There's a lot of data and a lot of the scoring mechanisms are still done on paper so I would imagine they get a lot of data-entry work but as we progress as an industry, hopefully, this process will become more automated.

Clearly new players are going to adapt quickly. Just wondering about well established players.

Bob Jewett · Aug 14, 2019

poolscholar said:
Is a 1 point difference between two people significant? Does that require a lot of games for both players?

To answer that question it's important to understand how statistics and probability work. Let's look at a simple case.

Suppose two players have identical abilities and they play each other 10,000 games. The most likely outcome is 5000-5000 but that is also very unlikely to happen. That's kind of a paradox. It would be a bad bet to flip a coin 10000 times and bet on exactly 5000 heads, even though that's the most likely outcome.

The variation above and below 5000 wins for either player is expected to be about 25. A score within 25 of 5000 will occur about 68% of the time. A score within 50 of 5000 (twice the variation) will happen 95% of the time.

So how does this translate to Fargorate for our two players? If they only played each other, a score of 4975-5025 would give them a Fargo difference of 1.44 and a score of 4950-5050 would give a Fargo difference of 2.88 points.

10,000 is a lot of games. If they played only 1000 games against each other, the likely errors would be about three times larger.

The great thing about FargoRate is that the games don't have to be against a particular player to find a (fairly) accurate rating difference. As long as they have good connections through other players, those games help improve the accuracy of their relative ratings.

Jude Rosenstock · Aug 14, 2019

Bob Jewett said:
To answer that question it's important to understand how statistics and probability work. Let's look at a simple case.

Suppose two players have identical abilities and they play each other 10,000 games. The most likely outcome is 5000-5000 but that is also very unlikely to happen. That's kind of a paradox. It would be a bad bet to flip a coin 10000 times and bet on exactly 5000 heads, even though that's the most likely outcome.

The variation above and below 5000 wins for either player is expected to be about 25. A score within 25 of 5000 will occur about 68% of the time. A score within 50 of 5000 (twice the variation) will happen 95% of the time.

So how does this translate to Fargorate for our two players? If they only played each other, a score of 4975-5025 would give them a Fargo difference of 1.44 and a score of 4950-5050 would give a Fargo difference of 2.88 points.

10,000 is a lot of games. If they played only 1000 games against each other, the likely errors would be about three times larger.

The great thing about FargoRate is that the games don't have to be against a particular player to find a (fairly) accurate rating difference. As long as they have good connections through other players, those games help improve the accuracy of their relative ratings.

I will often tell people to look at it in 25 point chunks. Within 25 points is about the same. At around 25 points, the better player is expected to win but it should be very close. At 50 points, the better player wins by a couple games. At 100 points, the better player is usually doubling the score of the weaker player.

That said, I also emphasize this is data and a set provides a very small sample size. Anomalies happen and are more likely to happen as the sets get shorter and shorter.

MattPoland · Aug 14, 2019

On this topic, I’ve certainly wondered whether Mike Dechaine, Rodney Morris or Johnny Archer are playing near their rating today. It’s hard to picture them as top 10 US players still given their circumstances and performance in the last couple years.

Sent from my iPhone using Tapatalk Pro

Jude Rosenstock · Aug 14, 2019

MattPoland said:
On this topic, I’ve certainly wondered whether Mike Dechaine, Rodney Morris or Johnny Archer are playing near their rating today. It’s hard to picture them as top 10 US players still given their circumstances and performance in the last couple years.

Sent from my iPhone using Tapatalk Pro

It's very interesting that you bring this up. Certainly, if you want to see a list of top players, recent and robust activity should matter and, as you suggest, if the top players are inactive, they've likely dropped. On the other hand, if I see a guy who has a 500 rating and no activity in the past couple of years, I assume he's gotten better.

alphadog · Aug 14, 2019

MattPoland said:
On this topic, I’ve certainly wondered whether Mike Dechaine, Rodney Morris or Johnny Archer are playing near their rating today. It’s hard to picture them as top 10 US players still given their circumstances and performance in the last couple years.

Sent from my iPhone using Tapatalk Pro

Mike and Rodney have still been winning.
Archer I dont know about.

poolscholar · Aug 15, 2019

I think its pretty clear that bad players have the most potential to go up quickly in rating.

On the other end, the elite players have the most potential to go down (but probably not quickly) because its very hard to maintain the highest level of play forever. I imagine Efren used to be an 800+ when he was young but now he randomly misses balls

Jude Rosenstock · Aug 15, 2019

poolscholar said:
I think its pretty clear that bad players have the most potential to go up quickly in rating.

On the other end, the elite players have the most potential to go down (but probably not quickly) because its very hard to maintain the highest level of play forever. I imagine Efren used to be an 800+ when he was young but now he randomly misses balls

Yes and no. There's a couple things in play here both in theory and in practice. Most elite players have lots of data because they compete more. That's going to be the driving force behind their stagnant rating. However, to your point, beginners will have a more dramatic learning curve. If someone is rated 400 and they don't know how to draw and one day, they figure it out, they could jump 50 points very quickly. That said, most beginners have very little data supporting their rating.

Do fargo ratings change fast enough?

Jude Rosenstock

AzB Silver Member

alphadog

AzB Silver Member

Jude Rosenstock

AzB Silver Member

poolscholar

AzB Silver Member

Bob Jewett

AZB Osmium Member

Jude Rosenstock

AzB Silver Member

MattPoland

AzB Silver Member

Jude Rosenstock

AzB Silver Member

alphadog

AzB Silver Member

poolscholar

AzB Silver Member

Jude Rosenstock

AzB Silver Member