Not trying to be a smartaleckor anything but i am not aware of any handicapping system that cares who wins or loses a match.:grin:
FargoRate Custodian - Keeping the Data Clean
- Thread starter CSI Media
- Start date
Not trying to be a smartaleckor anything but i am not aware of any handicapping system that cares who wins or loses a match.:grin:
With 1250 matches so far in apa and being raised ...and lowered 4 times in the last year between a 5 and 6 I have a lil experience how the system works.
All I am saying is that your opponents mishaps ...like an early 8 and your lucky 8 on the break should not influence your handicap. Neither one of those Incidents gives an indication of your playing level and therefore should not cause you to be raised.
More later....headed to work
To date, the accu-stats rating system remains the best one I've seen. Of course it was more time consuming because someone has to track every game of every match and input the results, which is exactly what Pat does when streaming his matches. Combining the results of a series of matches will give you a realistic idea of the playing ability of a player. Mike Sigel consistently scored the highest of all the pros the two years that Pat applied accu-stats in all the tourneys. He would train people how to do it and have someone working each individual match. It just became too exhaustive an endeavor to continue.
Trust me I understand what you mean.
I think games that involve the other player scratching on the 8, hitting it in early or putting it in the wrong pocket should simply not be included in the stats at all. However they should be tracked as there are people that use that to their advantage in trying to manipulate things.
8 on the breaks should certainly count for 2 reasons:
1.) It is a skill and they are not common, especially among lower level players.
2.) It is a win and better players do it more often so it needs to factor in. Considering 1 above, it is not going to have a major impact on anyone's rating whether they are a 2 or a 6.
Last edited:
When you are looking at HUNDREDS and THOUSANDS of games per player as data entry points then any rolls, scratches on the 8, and flukes will average out among everyone.
The big thing here is that the games you win and lose and to whom consistently are what matters. Everything that is involved in being at that consistent level IS what makes you the player you are.
Sure, you COULD break it down into the EXACT mistakes you make, the exact amount of unforced errors, the exact amount of times you scratched on the eight or made it out of turn, but the end result would be exactly the same when compared to everyone else. Namely those who are BETTER than you make less mistakes than you do.
That's literally the bottom line and it's the same no matter what level you are rated at. Compare a 728 player with an 828 player and it's 100% true that over a significant span of games the 728 WILL make more unforced errors, a lot more of them. Sure the 728 CAN play perfect and error free sometimes, he can run out a race to five a decent amount of time, he can freewheel over lesser players appearing to play nothing but perfect. But when he plays the 828 the errors he makes and gets away with end up costing him as the 828 punishes them with flawless play.
The knowledge, the ability to execute, the nerves....all of this plays into where you are as a player.
It's ALL contained in the performance you turn in day after day and week after week.
What do we tell each other all the time about how to get better? We say to get better you have to play better players until you can beat them. You have to "pay your dues" and get pounded on by better players until you learn what they know and can use it to beat them. And until you can you are ALWAYS rated lower than them by everyone who knows you and them. No one sits around and spends a lot of time dissecting WHY you are still not at that level yet....they just know that you aren't.
And this is the essence of Fargo Ratings. It doesn't care WHY you lost or won a single game. One single game isn't enough to form any reliable basis about what level you are as a player. It doesn't care why YOU lost 324 games and won 220 games out of the 524 games it has on record that you played.
It only cares that based on that record COUPLED with the records of everyone you played and everyone they played and so on that it can ACCURATELY predict how you will do against anyone else with an established rating. As long as it can do that within a standard expected deviation then it's enough to say that the rating is accurate.
So if player A is a really strong shot maker and this is why he wins a lot even though he plays bad patterns and Player B is a really strong safety player who needs to have simple run outs which he gets often by playing safe....both of them could be rated equally despite very different styles. How they got to their rating doesn't matter.
Player C is a really strong shot maker AND a really strong safety player...so more than likely he will be rated above A and B because of the fact that he has the same level of skill on both aspects that they each individually have. Thus over the course of hundreds of games that combination of strengths results in having a higher win percentage against stronger opponents than A and B have.
The other night I played a 598 and everyone local is under the impression that this person should have a much higher Fargo rating than that. I am a 610. Fargo predicts that the races should be relatively even with me having a slight edge in longer races.
Race to 15 I lost the first set 15-8. I won sets two and three 15-13. Fargo predicts that the longer we play I will continue to have a slight edge but it cannot tell you WHY I have that edge. What do I do SLIGHTLY better than my opponent that allows me to have that edge? Is it ONE thing or five things?
What does he do worse than me that allows him to be at a slight disadvantage? Is it one thing or five things?
If we had stopped at one set everyone would have concluded that Fargo Ratings were WRONG for either him or me, either he was way too low or I was way too high. But you can't make these determinations based on ONE SET. You need more data, more games to check the predictions against.
And this then is the rub....can Fargo be manipulated by getting in a lot of losing games? Yes, as long as there are local weekly tournaments which report to Fargo then players CAN get a lot of losing games into the system. In chess as well players often sandbag by losing to lower rated players to keep their ratings lower for the express purpose of being able to enter into rated events which they can easily win as the actual best player in the field.
This is why there is AI that tracks the trends to determine if a player seems to be performing significantly different than at previous times. In other words WHEN performance is outside the mean predicted outcome then the system needs to try to figure out the WHY? Because IF the system is known to be accurate for HONEST players who try their best based on the data then whenever a player is performing well outside their zone there has to be a reason. Sandbagging? Illness? Lessons? Extra practice? Drugs? A freaky run of scratching on the 8 27 games in a row?
Obviously when a player is on the rise no one accuses them of sandbagging. So if I player performs awesome on cocaine and has four great tournaments racking up 200 wins against good players then he probably will do worse when he isn't on drugs. So is he sandbagging on the way down? No, he is reaching his average between drug enhanced performance and sober performance.
But if he loses 200 games in a row then what is it? And this is what data analysis and AI tries to figure out and assign less weight to. It's not a perfect science but at the end of the day Fargo Ratings are science based and not opinion based and THAT is what makes all the difference.
Yes there will be some people who are underrated and some might be overrated. But the vast majority will be accurately rated and the community at large CAN and SHOULD help to weed out those who are deliberately trying to game the system in order to make the system even better for everyone else. We have lived with ratings and handicapped events and leagues and tournaments and tours for 30 years now. They are not going away.
Getting away from opinion-based ratings though is worth it. Finally having a way to know with pretty good confidence that two players from different parts of the world are actually even is worth a lot towards the goal of not only keeping pool alive but also to grow it in my opinion.
The big thing here is that the games you win and lose and to whom consistently are what matters. Everything that is involved in being at that consistent level IS what makes you the player you are.
Sure, you COULD break it down into the EXACT mistakes you make, the exact amount of unforced errors, the exact amount of times you scratched on the eight or made it out of turn, but the end result would be exactly the same when compared to everyone else. Namely those who are BETTER than you make less mistakes than you do.
That's literally the bottom line and it's the same no matter what level you are rated at. Compare a 728 player with an 828 player and it's 100% true that over a significant span of games the 728 WILL make more unforced errors, a lot more of them. Sure the 728 CAN play perfect and error free sometimes, he can run out a race to five a decent amount of time, he can freewheel over lesser players appearing to play nothing but perfect. But when he plays the 828 the errors he makes and gets away with end up costing him as the 828 punishes them with flawless play.
The knowledge, the ability to execute, the nerves....all of this plays into where you are as a player.
It's ALL contained in the performance you turn in day after day and week after week.
What do we tell each other all the time about how to get better? We say to get better you have to play better players until you can beat them. You have to "pay your dues" and get pounded on by better players until you learn what they know and can use it to beat them. And until you can you are ALWAYS rated lower than them by everyone who knows you and them. No one sits around and spends a lot of time dissecting WHY you are still not at that level yet....they just know that you aren't.
And this is the essence of Fargo Ratings. It doesn't care WHY you lost or won a single game. One single game isn't enough to form any reliable basis about what level you are as a player. It doesn't care why YOU lost 324 games and won 220 games out of the 524 games it has on record that you played.
It only cares that based on that record COUPLED with the records of everyone you played and everyone they played and so on that it can ACCURATELY predict how you will do against anyone else with an established rating. As long as it can do that within a standard expected deviation then it's enough to say that the rating is accurate.
So if player A is a really strong shot maker and this is why he wins a lot even though he plays bad patterns and Player B is a really strong safety player who needs to have simple run outs which he gets often by playing safe....both of them could be rated equally despite very different styles. How they got to their rating doesn't matter.
Player C is a really strong shot maker AND a really strong safety player...so more than likely he will be rated above A and B because of the fact that he has the same level of skill on both aspects that they each individually have. Thus over the course of hundreds of games that combination of strengths results in having a higher win percentage against stronger opponents than A and B have.
The other night I played a 598 and everyone local is under the impression that this person should have a much higher Fargo rating than that. I am a 610. Fargo predicts that the races should be relatively even with me having a slight edge in longer races.
Race to 15 I lost the first set 15-8. I won sets two and three 15-13. Fargo predicts that the longer we play I will continue to have a slight edge but it cannot tell you WHY I have that edge. What do I do SLIGHTLY better than my opponent that allows me to have that edge? Is it ONE thing or five things?
What does he do worse than me that allows him to be at a slight disadvantage? Is it one thing or five things?
If we had stopped at one set everyone would have concluded that Fargo Ratings were WRONG for either him or me, either he was way too low or I was way too high. But you can't make these determinations based on ONE SET. You need more data, more games to check the predictions against.
And this then is the rub....can Fargo be manipulated by getting in a lot of losing games? Yes, as long as there are local weekly tournaments which report to Fargo then players CAN get a lot of losing games into the system. In chess as well players often sandbag by losing to lower rated players to keep their ratings lower for the express purpose of being able to enter into rated events which they can easily win as the actual best player in the field.
This is why there is AI that tracks the trends to determine if a player seems to be performing significantly different than at previous times. In other words WHEN performance is outside the mean predicted outcome then the system needs to try to figure out the WHY? Because IF the system is known to be accurate for HONEST players who try their best based on the data then whenever a player is performing well outside their zone there has to be a reason. Sandbagging? Illness? Lessons? Extra practice? Drugs? A freaky run of scratching on the 8 27 games in a row?
Obviously when a player is on the rise no one accuses them of sandbagging. So if I player performs awesome on cocaine and has four great tournaments racking up 200 wins against good players then he probably will do worse when he isn't on drugs. So is he sandbagging on the way down? No, he is reaching his average between drug enhanced performance and sober performance.
But if he loses 200 games in a row then what is it? And this is what data analysis and AI tries to figure out and assign less weight to. It's not a perfect science but at the end of the day Fargo Ratings are science based and not opinion based and THAT is what makes all the difference.
Yes there will be some people who are underrated and some might be overrated. But the vast majority will be accurately rated and the community at large CAN and SHOULD help to weed out those who are deliberately trying to game the system in order to make the system even better for everyone else. We have lived with ratings and handicapped events and leagues and tournaments and tours for 30 years now. They are not going away.
Getting away from opinion-based ratings though is worth it. Finally having a way to know with pretty good confidence that two players from different parts of the world are actually even is worth a lot towards the goal of not only keeping pool alive but also to grow it in my opinion.
JC
Coos Cues
Not sure why my rating has dropped without leaving home when I'm still the top player in our small poorly connected league and my robustness remains the same. Which makes me think our league stats are not being considered. Yet some in our league have changed considerably without leaving home.
Does this mean people I have played in the past outside of here are losing ground also? Under performing against the well connected pool?
Some things about Fargo seem contradictory to me.
JC
Does this mean people I have played in the past outside of here are losing ground also? Under performing against the well connected pool?
Some things about Fargo seem contradictory to me.
JC
There are two primary likely causes if your rating changes without you playing. Both mean FargoRate is getting a better handle on how you play.
(1) More historical games have been added for you--perhaps past league games. (I note you said your robustness stayed the same so this is not the case for you).
(2) The second one is also important. Suppose a bunch of your games that are already in there were actually played against opponents who themselves have little information in the system. Your rating is based on your wins and losses against opponents of known rating, and if FargoRate doesn't know how these opponents play, it is not going to use that information to rate you. Nevertheless, FargoRate keeps those games in the back of its mind. As your former opponents play more and get better rated themselves, FargoRate starts paying more attention to your games against them.
Other than some one-off situations where we were able to get in some past league data, the mechanism to get league data in is through FargoRate LMS--our new league management system that is connected to our database. That is out there now for USA Pool League (match play) and is being used in some test locations for round-robin formats
JC
Coos Cues
Thanks Mike I guess that makes sense. No wonder I want to quit pool. Everyone I beat wasn't as good as I thought they were:smile:
Seriously though, when will we be able to replace league sys with the LMS software for our BCA league so that Fargo Ratings can start being more meaningful? What are the bugs holding it up? Which is also I believe holding up Western BCA from using it to rate our regional events which is going to be a turning point in grouping players correctly IMO.
If you need more to beta test it, we're up for it here.
JC