Fargo Ratings and table size

[sixpack]

If you have three items in the test and two of them are variables, you can not say the results of the third item tested are 100% correct unless you test ALL options for the variables, in this case only half of the data is there.

To complete this they would need to test the 10B on a 7' and a 9B on a 9'. Do I think the difference would be much different? No. Would I put my money on the results shown? Also no.

Simply put, there are too many variables to determine the outcome here. Assumptions, yes. Correlation, sure. Causation, no.

I understand your point and you are not wrong.

However, the assertion that FargoRate has made in the past is that it holds true for all combinations of equipment and games (except 1-pocket).

In this case the assertion is that table size doesn’t matter - which already assumes game does not matter. From that perspective it’s another datapoint that shows that FR does correctly predict results between players BNB o matter what equipment or game they are playing.

You are correct in that it doesn’t specifically isolate data for table size only. But that is the data we had to work with for this example.

FR has done other table size comparisons using the same games but different table sizes - but the players were different.

This example had the advantage of comparing the exact same players playing on different equipment - even though they are playing different games on each.

There is a lot of information available about how FR performs in different situations and it holds up surprisingly well. (To me, at least)

I saw this example as more of a fun confirmation than anything else. Like I said above, another data point that is interesting and shows FR is remarkably consistent.

 

[hang-the-9]

I would respect the results more if both games were the same, I simply can't see how you can compare two completely different games on different tables?

That's like taking an automatic racecar on one track and then taking the same model with a stick shift on a different track and trying to figure out which one's better?

I think it's a matter of using human intuition when trying to deal with math. Overall, a 700 player will beat a 650 rated player most of the time in ANY game, on ANY table. Exceptions don't enter here, if you know 2 guys that can beat a better player on a bar table most of the time, that does not matter to statistics that measure thousands of players. The only difference may be that the 700 player will play like a 750 on easy equipment but that also means the 650 will play like a 700, etc.. so everyone is still as equally likely to win or loose the same game. Meaning both players run more balls per turn on the table but the win/loss ratio will not change.

 

[Chili Palmer]

I think it's a matter of using human intuition when trying to deal with math. Overall, a 700 player will beat a 650 rated player most of the time in ANY game, on ANY table. Exceptions don't enter here, if you know 2 guys that can beat a better player on a bar table most of the time, that does not matter to statistics that measure thousands of players. The only difference may be that the 700 player will play like a 750 on easy equipment but that also means the 650 will play like a 700, etc.. so everyone is still as equally likely to win or loose the same game. Meaning both players run more balls per turn on the table but the win/loss ratio will not change.

Agreed, just stating facts

 

[hang-the-9]

Agreed, just stating facts

Well technically the Fargo ratings and the stats are the facts you are forming opinions based on your ideas of what should be LOL
It's like those optical illusions when they compare colors, shapes, sizes and things are way off what your brain is telling you.
A few times I was pretty upset about some tournament handicaps when the games started since I could see the player shoot well, then I played them and won the matches fairly easily, my ideas of their skills formed from a short time did not match their real ability to actually win and finish games based on their long term skills.

Same thing happens with Fargo results and ratings, and why Mike is posting these results, even though people still work to find issues. Statistics do have issues, but in the long run, over a large population, they are correct. Casinos base their whole business on it and do pretty well. Just because someone is "good" at gambling does not mean that over a long amount of time with a large amount of people they still get what they want no matter the blips in the numbers. Which is also why profiling by police and other agencies is used. Sure there are a billion "nice" Muslims, but if you look at the statistics, they show the numbers. It's not nice, but it is based on hard numbers.

 

[Poolplaya9]

If you have three items in the test and two of them are variables, you can not say the results of the third item tested are 100% correct unless you test ALL options for the variables, in this case only half of the data is there.

To complete this they would need to test the 10B on a 7' and a 9B on a 9'. Do I think the difference would be much different? No. Would I put my money on the results shown? Also no.

Simply put, there are too many variables to determine the outcome here. Assumptions, yes. Correlation, sure. Causation, no.

What makes you think he hasn't done all that testing?

And what makes you think your intuition, or even your perception based on a few thousand games you have seen, when you weren't even keeping close track of things, is somehow more reliable than many millions of games of data that were analyzed very closely?

 

[Chili Palmer]

What makes you think he hasn't done all that testing?

And what makes you think your intuition, or even your perception based on a few thousand games you have seen, when you weren't even keeping close track of things, is somehow more reliable than many millions of games of data that were analyzed very closely?

What makes you think you can buck statistics?

I've said it before and I will say it one last time - I know what the results say and I can't dispute them but, technically speaking you cannot say the results are 100% reliable.

End of story.

 

[skogstokig]

i know of the crossing of players between table sizes as a statistical ground for fargorate, but it's also true that it is, if only by two feet, a different game. how small would you go and still call it reasonable to include small table matches? 6ft table? 5ft table? carrom table?

 

[BasementDweller]

Variance is the last gripe I have against FargoRate. Well, it's really not even a gripe against FR but more so against how it's utilized. I see quite a few tourneys near me that use the 100% handicapping, I think that's what is labeled "hot" by FR.

I've never been and never will be a fan of 100% handicapping. I suppose that's just on principle. But I digress...

Anyway variance....so I haven't completely thought this through but what bothers me with FR is "it seems like" I too often get beaten by lesser players playing above their FR, while better players with higher robustness tend to play much closer to their's. I would think that's just the nature of the beast but it sure is frustrating.

I wonder if Mike has put anything out on variance??? An example would be playing a 550 rated player, that somehow plays a short set nearly flawlessly. It doesn't seem that uncommon for a 550 to play anywhere from 450 all the way to 650 or higher in a short set. Especially if they are only going to 3 or 4. Basically, when this happens -- the better player doesn't stand much of a chance. As a player with a 650 skill set isn't going to vary as much.

I know I don't have this completely sorted out as the better players still seem to win handicapped events, it's just the helpless losses that are frustrating. This is really more of a general observation I suppose. I guess it boils down to -- if handicaps are accurate, I would rather play better players than lesser ones.

 

[mikepage]

Variance is the last gripe I have against FargoRate. Well, it's really not even a gripe against FR but more so against how it's utilized. I see quite a few tourneys near me that use the 100% handicapping, I think that's what is labeled "hot" by FR.

I've never been and never will be a fan of 100% handicapping. I suppose that's just on principle. But I digress...

HOT matches are not really 100% handicapping. It just means no individual match is tilted in favor of the lower-rated player. On average, the higher-rated player has an advantage--perhaps 55%/45% if it is like races to 5 or so. This means for, say a 32-player single elimination tournament (for which you must win 5 matches to win), the highest-rated entrant has about 3 times the chance of winning the event as does the lowest-rated entrant.

Anyway variance....so I haven't completely thought this through but what bothers me with FR is "it seems like" I too often get beaten by lesser players playing above their FR, while better players with higher robustness tend to play much closer to their's. I would think that's just the nature of the beast but it sure is frustrating.

I recommend googling "confirmation bias."

I wonder if Mike has put anything out on variance??? An example would be playing a 550 rated player, that somehow plays a short set nearly flawlessly. It doesn't seem that uncommon for a 550 to play anywhere from 450 all the way to 650 or higher in a short set. Especially if they are only going to 3 or 4. Basically, when this happens -- the better player doesn't stand much of a chance. As a player with a 650 skill set isn't going to vary as much.

There IS this...

 

[mikepage]

i know of the crossing of players between table sizes as a statistical ground for fargorate, but it's also true that it is, if only by two feet, a different game. how small would you go and still call it reasonable to include small table matches? 6ft table? 5ft table? carrom table?

Can you wash a car with dirty water?

Your first answer is probably "of course not."

But then you think if you have a dusty or muddy car, you can most definitely make it cleaner with a bucket of dirty water. More generally, whether your bucket of water can make the car cleaner depends both on how dirty the water is and also on how dirty the car is.

If we already have many thousands of games between a particular two players on 9-foot tables and we are confident they are 60 points apart within a handful of points, then adding the 5-foot table data is like cleaning an already pretty clean car with dirty water. It would make our comparison of those two players worse.

But here is a way to think of a TON of 5-foot table data. There are several million games in our system played against opponents we don't really know. A giant dose of 5-foot-table data can take a massive number of players out there in the big grid who we previously had no idea how they played and al least separate them into pro-level, good amateur, weak amateur, and beginner. If we can successfully do that, Fargo Ratings get much better overall. In other words, we are now and will be for a while sufficiently data starved that buckets of dirty water improve things.

 

[Chili Palmer]

Variance is the last gripe I have against FargoRate. Well, it's really not even a gripe against FR but more so against how it's utilized. I see quite a few tourneys near me that use the 100% handicapping, I think that's what is labeled "hot" by FR.

I've never been and never will be a fan of 100% handicapping. I suppose that's just on principle. But I digress...

Anyway variance....so I haven't completely thought this through but what bothers me with FR is "it seems like" I too often get beaten by lesser players playing above their FR, while better players with higher robustness tend to play much closer to their's. I would think that's just the nature of the beast but it sure is frustrating.

I wonder if Mike has put anything out on variance??? An example would be playing a 550 rated player, that somehow plays a short set nearly flawlessly. It doesn't seem that uncommon for a 550 to play anywhere from 450 all the way to 650 or higher in a short set. Especially if they are only going to 3 or 4. Basically, when this happens -- the better player doesn't stand much of a chance. As a player with a 650 skill set isn't going to vary as much.

I know I don't have this completely sorted out as the better players still seem to win handicapped events, it's just the helpless losses that are frustrating. This is really more of a general observation I suppose. I guess it boils down to -- if handicaps are accurate, I would rather play better players than lesser ones.

Mike has posted here before that players can range from 50 points below to 50 points above their current FR rating but your statement makes me wonder - is that true for high level players also? Does a 750 FR really bounce from 700 to 800? I wouldn't think so, but it's entirely possible on any given day I suppose.

 

[BasementDweller]

HOT matches are not really 100% handicapping. It just means no individual match is tilted in favor of the lower-rated player. On average, the higher-rated player has an advantage--perhaps 55%/45% if it is like races to 5 or so. This means for, say a 32-player single elimination tournament (for which you must win 5 matches to win), the highest-rated entrant has about 3 times the chance of winning the event as does the lowest-rated entrant.




I recommend googling "confirmation bias."


There IS this...

Thanks

While we are googling, I recommend you Google "patronize".

So I guess you're saying that all skill levels have similar performance ranges, at least when it comes to racks won & lost. I guess that's a bit of a surprise.

I guess I'll slide the variance I see into the short race format that lesser players benefit from, instead of it being solely performative. So yes I'll lose fairly handicapped matches to either good or bad players, but it just "feels" worse when it's a shorter match. So if I lose 7-4 to a better player that outplays me, this doesn't compare to losing 3-3 to someone who plays two good racks and I hang up a ball and give them a 3rd game.

In effect, when a lesser player wins a short match in this fashion, they are playing exactly the same as an 800 level player would. They play two mistake free racks and you give them another. When this happens the better player cannot win. Even if you eliminate the errors of the better player, there will still be times the better player just has zero chance to win. Rarely, can the lesser player play their best and lose, but the inverse is not true. I guess that's the part of handicapping that is unavoidable and even adds a bit of excitement to the match in a sense.

So I see I have no real gripes with FargoRate, just a few overall complaints about handicapping in general. Now don't get me started on ball spots.

 

[Poolplaya9]

I guess I'll slide the variance I see into the short race format that lesser players benefit from, instead of it being solely performative. So yes I'll lose fairly handicapped matches to either good or bad players, but it just "feels" worse when it's a shorter match. So if I lose 7-4 to a better player that outplays me, this doesn't compare to losing 3-3 to someone who plays two good racks and I hang up a ball and give them a 3rd game.

I think you are correct in your assessment that this is more a function of short races rather than a disparity in performance ranges.

Keep in mind that it works both ways though. Unless you are always the best player those shorter races are also allowing you to beat the players that are better than you times when you would not have beat them had the race been longer too. It sure is easy to recognize and note and lament the times when things like that work against us, and even easier to fail to recognize and give the same weight to the times when those same things were a big benefit to us. Works this way with races lengths, works this way with rolls, etc.

 

[BasementDweller]

I think you are correct in your assessment that this is more a function of short races rather than a disparity in performance ranges.

Keep in mind that it works both ways though. Unless you are always the best player those shorter races are also allowing you to beat the players that are better than you times when you would not have beat them had the race been longer too. It sure is easy to recognize and note and lament the times when things like that work against us, and even easier to fail to recognize and give the same weight to the times when those same things were a big benefit to us. Works this way with races lengths, works this way with rolls, etc.

Sort of true but it tends to be just the weaker players that get the really short races. Unless you're in an area where there's a lot of say, 750 and above players. You'd have to be in a real hotbed for that ro be the case though.

 

[Fatboy]

Exactly. While you might be a '700' on a bar table - if it were an absolute rating - the guy who is a '700' on the 9' table would typically be a 750 on the bar table. I used to love both sides of this. I'd gamble with people on a bar table and they'd excuse losing to me by saying I was a 'bar table specialist' and they'd love to get me on the big table. I played better relative to most people on the big table than the bar table. So I'd usually beat them there too.

Then I'd be playing in a pool hall and play someone on a 9' table and they would say "I'm not used to these big tables, let's play on a bar box." And I'm no slouch on a bar box either.

The point is, they compare their bar table ball pocketing to my big table ball pocketing and think they can beat me.

It's the same thing I've heard throughout the years when people say "My offense is a 5 but my knowledge of the game is a 7".

Maybe, especially people who were great players at one time. But it doesn't really matter, if you can't execute the shots knowledge only gets you so far. Especially when the opponent probably has similar knowledge. Most people that say that really don't have superior knowledge IMO.

That’s correct, knowledge is all fine and dandy-but you have to execute. Not playing, old age robs all of us of that at one time or another as does lack of talent.

Great knowledge helps commentary not winning.

Flip side is all the firepower and ability to execute and not know how to put it to work the best way. That’s bad as well.

To play solid pool you must have both, one without the other just won’t cut it. That’s why there’s a ever changing list of who the top players are and nobody ever stays on top for long. A handful of years is pretty good. Seems long but is it really?

This applies to lots of things other than pool

Interesting thread

Fatboy

 

[skogstokig]

Can you wash a car with dirty water?

Your first answer is probably "of course not."

But then you think if you have a dusty or muddy car, you can most definitely make it cleaner with a bucket of dirty water. More generally, whether your bucket of water can make the car cleaner depends both on how dirty the water is and also on how dirty the car is.

If we already have many thousands of games between a particular two players on 9-foot tables and we are confident they are 60 points apart within a handful of points, then adding the 5-foot table data is like cleaning an already pretty clean car with dirty water. It would make our comparison of those two players worse.

But here is a way to think of a TON of 5-foot table data. There are several million games in our system played against opponents we don't really know. A giant dose of 5-foot-table data can take a massive number of players out there in the big grid who we previously had no idea how they played and al least separate them into pro-level, good amateur, weak amateur, and beginner. If we can successfully do that, Fargo Ratings get much better overall. In other words, we are now and will be for a while sufficiently data starved that buckets of dirty water improve things.

just to clarify, barbox matches is the dirty water in this analogy?

 

[mikepage]

just to clarify, barbox matches is the dirty water in this analogy?

Well yeah, sort of.

 

[Poolplaya9]

Sort of true but it tends to be just the weaker players that get the really short races. Unless you're in an area where there's a lot of say, 750 and above players. You'd have to be in a real hotbed for that ro be the case though.

I'm having trouble understanding what you were trying to say here, but if you were trying to say that the higher you are ranked in the pecking order of the field the less often you are going to end up on side benefiting from the shorter races, then I agree with you. But to find a bright side consider the alternative. With say races to 9 or even 7 instead of races to 3 or 4, half of those players worse than you would stop coming because now they have zero chance, and others would stop coming because that is too much of their day to devote to pool, or it would keep them out too late. Many of those player vacancies probably wouldn't be filled with new players and so now your tournaments have less players and less prize money, and the new blood the longer races did manage to attract would tend to be the better players, likely often better than you. Now your tournaments have less people, less prize money, keep you out late and/or take up a lot of your day, and are significantly harder for you to win, so it got even worse for you in just about every way.

Even if you would still prefer that to what you have now, you have to remember that the business has to do what is best for them, which usually means more concern for quantity of players rather than the quality of the players. The more people they can get in there, the more money they stand to make, and shorter races often tend to bring the bigger fields in your typical local weekly's. Plus the lesser skilled players are the ones who tend to spend more money anyway and are often preferred for that reason too. At least right now you have dependable tournaments where you get decent fields, it doesn't take up too much of your day or keep you out too late, and you still have a pretty good chance of doing well. Sounds half decent to me, and its the bright side in any case.

 

[DCMike]

Well yeah, sort of.

Regarding the consistency myth video: Very informative video! I find it very interesting and would not have thought that to be the case (at first). But now I think it makes sense. I think part of the push-back from people is that the type of “Fargo consistency” talked about in the video doesn’t fit well with traditional ideas of consistency. But what really is "consistency"? Take these two statements:

(1) Higher rated players tend to be more consistent than lower rated players.

(2) Higher rated players have the same Fargo performance variability as lower rated players.

Some may believe that only one of these statements can be true. Though I believe both can be true as long as we differentiate between the two ideas of "consistent" If fact, (2) may imply (1) if a suitable definition of "consistency" in statement (1) is pinned down.

Intuitively:

Take 2 players: Player 1 (Fargo 800) and Player 2 (Fargo 400), and have them each play many games against the ghost or some common third player. Give both players the same detrimental nudge to their “consistency” by making sure each misses an extra 1 out of 6 balls. (Something like: before every shot, a die is thrown and if it shows 6 spots, then the next shot is a miss, by definition,) In other words, give both players the same absolute change to “consistency”. (This means i am defining consistency in terms of a rate of missing balls, an arguable point.)

Using “Fargorate performance” as the measuring stick, we should expect a more catastrophic effect on the 800 player than on the 400 player, due to differing relative effects this nudge has on each player’s (vastly different) natural error rates. (Whether or not this seems intuitive, a very crude math analysis shows this is expected. It is as though, with the 400 level player, Fargorate views this nudge with a naked eye, but with the 800 level player, views it through a 16x magnification lens, figuatively making a mountain out of a molehill.)

Therefore if (1) were false, we should expect you (Mike Page) to be reporting that highly ranked players have more variability in Fargorate performance, because of the "making of mountains out of molehills" effect . But you are not. You are reporting that variability is constant, and this would support (1).

Crude math seems to support this as well. If we imagine Fargo rating as determined by playing games against a 500 level opponent, then Fargo rating is proportional to the logarithm of the odds of losing a game to this opponent. Deviations in Fargo performance would be (approximately) proportional to observed deviation in these odds, divided by the odds. For a 400 level player you'd have a quotient with denominator 2. For a 800 level player you'd have a quotient with denominator 0.125. So if these two quotients show the same variability (as the data seems to indicate), then the numerators must have differing variances, with the one corresponding to the 800 level player being less.

The best part of what the video suggest is that is permits that consistency can have some grounding by being quantifiable and empirically justified. In my mind it's ok to accept that (1) and (2) are compatible with each other as as it is understood the two notions of consistency refer to different things, with the traditional notion of consistency referring to something vague and subjective. Vague, subjective and even incorrect ideas can have lots of traction simply because they've been around a long time.

 

[BasementDweller]

I'm having trouble understanding what you were trying to say here, but if you were trying to say that the higher you are ranked in the pecking order of the field the less often you are going to end up on side benefiting from the shorter races, then I agree with you. But to find a bright side consider the alternative. With say races to 9 or even 7 instead of races to 3 or 4, half of those players worse than you would stop coming because now they have zero chance, and others would stop coming because that is too much of their day to devote to pool, or it would keep them out too late. Many of those player vacancies probably wouldn't be filled with new players and so now your tournaments have less players and less prize money, and the new blood the longer races did manage to attract would tend to be the better players, likely often better than you. Now your tournaments have less people, less prize money, keep you out late and/or take up a lot of your day, and are significantly harder for you to win, so it got even worse for you in just about every way.

Even if you would still prefer that to what you have now, you have to remember that the business has to do what is best for them, which usually means more concern for quantity of players rather than the quality of the players. The more people they can get in there, the more money they stand to make, and shorter races often tend to bring the bigger fields in your typical local weekly's. Plus the lesser skilled players are the ones who tend to spend more money anyway and are often preferred for that reason too. At least right now you have dependable tournaments where you get decent fields, it doesn't take up too much of your day or keep you out too late, and you still have a pretty good chance of doing well. Sounds half decent to me, and its the bright side in any case.

You were following me just fine about the short races. Once you get above 600, you will usually have to win a few games.

I don't disagree with anything you said in the rest of your post. I'm not one of those guys that's opposed to handicapped events. There's a time and a place for both open and handicapped tourneys, and I love what Mike and his team have done with FargoRate.

I guess I'm just pointing out my one lament about handicapped tourneys.

I'm really making it sound like I'm out playing all the time here, which may be the biggest joke in the whole thread as I rarely get a chance to play. When I do, I'm a threat to take out anyone in the room, including myself

It will be interesting to see how the dust settles in a few years as far as all these tourneys go. I pay more attention to tourney postings on Facebook and in my neck of the woods it looks like we are starting to see smaller game spots with more FargoRate caps for entries. While I feel bad for the stronger players, I like this approach as I think having really "fair" handicaps is a bit overestimated. So just giving a game or two here and there is enough to give lesser players a chance. It need not be a 8-3 race or some such ratio.