Web page that calculates APA 8-ball rating from win/loss record

[tomker]

When I was an SL5, I beat a SL6 4-3. So that makes me a SL7?

According to the handicap table, a 7 is expected to win 55% of his games against a 6. (It's a 5-4 race.)

So if you won 57% of your games against a 6, then yes, you played like a 7 for that match.

Of course this only takes into account how many games you've won and lost, but that's also the only concrete information we have from the APA regarding expected performance of various skill levels. They give us the handicap table which tells us the odds of rated players beating other rated players. We don't have any other concrete information about how the rating system works.

 

[tomker]

...

There IS a variable in the system that is automated, but I won't go into explanation of how it works. My LO has been truthful, but I would not be surprised if there are overrides in the system.

I am concerned that most of this thread is now posts about how the actual rating system works, and not about my web page or the math behind it.

I hope nobody thinks I started this thread to get into that discussion.

 

[DAVE_M]

I am concerned that most of this thread is now posts about how the actual rating system works, and not about my web page or the math behind it.

I hope nobody thinks I started this thread to get into that discussion.

I hate to be the bearer of bad news, but AzB tends to get off topic often.

 

[Cornerman]

I hate to be the bearer of bad news, but AzB tends to get off topic often.

LOL!!!

Freddie <~~~ let's talk about TAP and VNEA now

 

[shewmake]

If there is a link and I got one via pm I wouldn't complain

 

[TCIndepMo]

The chart has nothing to do with probable results that are anticipated from the various match ups. The chart was ONLY designed to speed up play each night so the fifth point would end at a reasonable hour. It does that. Mission accomplished.

Crunch the numbers all you want. It is still the LOs responsibility to input the weekly data, review all SLs and changes and, if warranted, override the computer number to lower or raise a SL to what the LO believes is fair to each player AND THEIR OPPONENT. Every area has a "players grapevne" which, while certainly not always accurate, is just another tool available to the LO What does your local grapevine think about you? I know what it thinks about me - been hearing it for years. LOL.

The SL calculated by the software is the beginning of the process, not the end.

And for those about to flame me with personal nightmare stories of LO malfeasance (or whatever) please be aware - in no way is it to my benefit to manipulate who wins a match or goes to Vegas. I write the same check to whomever wins, etc. The LO who plays favorites is soon out of business. But for the sandbagger eventually ID'ed by the LO we know it is so convenient to come up here and tell your one side of the story with you always being the "good guy".

If the APA is so swamped with lousy LOs and sandbaggers why is it the biggest amateur league on the planet? ... been in operation since 1979? . pay the most for an amateur membership? And on, and on, and on, etc.

I love this job.

LO since 1980.

 

[tomker]

The chart has nothing to do with probable results that are anticipated from the various match ups. The chart was ONLY designed to speed up play each night so the fifth point would end at a reasonable hour. It does that. Mission accomplished.
...

Of course the APA handicap chart tells you how likely it is for one rated player to beat another rated player at one game.

The entire point of the handicap system and chart is to give us fair matches, which means both players have to have a 50% chance of winning.

That translates directly to each player having a certain chance of winning any particular game, otherwise the odds aren't 50-50 and it's an unfair race.

So if you're going to claim that the APA handicap chart can't tell us probable results that's the same as saying the handicap system is unfair.

 

[justadub]

Of course the APA handicap chart tells you how likely it is for one rated player to beat another rated player at one game.

The entire point of the handicap system and chart is to give us fair matches, which means both players have to have a 50% chance of winning.

That translates directly to each player having a certain chance of winning any particular game, otherwise the odds aren't 50-50 and it's an unfair race.

So if you're going to claim that the APA handicap chart can't tell us probable results that's the same as saying the handicap system is unfair.

This argument comes up quite frequently in these threads....I don't think that any handicap system is designed to arrive at a true 50-50 outcome. And I don't know how it could.

In the APA's case, in my opinion, the handicap is designed to make an attempt at getting the match close. Meaning that the lesser player has a decent chance when playing the better player. I do not, for one minute, believe that when applied correctly....does it give the lesser player an even chance. A coin flip.

Better players are better players, for any number of reasons. Most of the time, the better player should win, if they play to their ability.

What the handicap does is to make the better player have to play his best to win. Just like he would have to if he played someone of equal skill. Now, if the lesser player plays over their head, they are very likely going to win. But that happens when players of equal skill play against each other, as well.

For an example, tonight, if the throws go the way I want them to, I (an SL5) am going to play an SL7. One that I have no business playing even. I don't really have any sort of chance to win, even with the two game handicap, unless I play WAY over my head, and he has an amazingly off night. (And I get lucky, win the lag and put the 8 in on the break.... then he uncharacteristically E-8's....and then I somehow win another one! ) Of course, these extremes don't happen terribly often, but it is an example of how the handicap doesn't remotely make it 50-50 between any two players.

The trick, of course, is to get the skill levels right, or as close to right as possible in a situation as large as this one. It's my observation that in our division, with between 100-150 players in any given session, it's pretty close. Are there some who aren't "right" in my opinion, yes. But usually by one skill level, and that usually gets corrected as the session goes along and the math catches up. And yes, once in a while, there is a player who is way off. Does that get frustrating, yup. But again, the math eventually catches up, and they are where they belong.

I like how it works, given the scope it has to account for. In our area, it works pretty much like its supposed to.

 

[railbird99]

Of course the APA handicap chart tells you how likely it is for one rated player to beat another rated player at one game.

The entire point of the handicap system and chart is to give us fair matches, which means both players have to have a 50% chance of winning.

That translates directly to each player having a certain chance of winning any particular game, otherwise the odds aren't 50-50 and it's an unfair race.

So if you're going to claim that the APA handicap chart can't tell us probable results that's the same as saying the handicap system is unfair.

It's pretty clear the APA handicap system is not perfect. In 8-ball, for example, you only have 6 handicap levels. You only have to play but so good to be a 7, and you could play twice as good as the lowest 7 and guess what, you're still a 7. So, since the handicap levels reach a ceiling and a floor, at those levels, more than likely the races are not "correct" since you can't accurately determine their true playing strength due to the limitations of the discrete handicap levels.

This is a neat exercise, but I think you should take into account innings just like the APA does. It's very simple, just use innings per win in 8-ball. This is the only way to determine how well you played. I'll explain:

Your method may work fairly well for the lower end of skill levels, because no one is capable of running out, and everyone has a chance to shoot and win every game. However, even 5's are fully capable of breaking and running a rack every now and then. Also there are 8 on the breaks, and early 8's, etc... With that in mind, say a 7 is playing a 5, and the 7 wins 5-2, but the 7 broke and ran 2 racks. Using your method, all you see is that the score is 5-2, even though when the 5 had a chance to shoot he was staying pretty even with the 7, 3-2. It says nothing about the strength of the 5, or how the 5 shot in those racks where he didn't get a chance to shoot. Two strong 7's could play each other, and one of the 7's breaks and runs 4 racks, and wins 5-1, but that in no way means the 7 that lost played like a 2 or a 3.

In essence, the score doesn't really say much about how well you shot, and it in no way helps determine the rating you are/will be. So if it doesn't do either of those two things, what does it do?

 

[APA Operator]

Of course the APA handicap chart tells you how likely it is for one rated player to beat another rated player at one game.

The entire point of the handicap system and chart is to give us fair matches, which means both players have to have a 50% chance of winning.

That translates directly to each player having a certain chance of winning any particular game, otherwise the odds aren't 50-50 and it's an unfair race.

So if you're going to claim that the APA handicap chart can't tell us probable results that's the same as saying the handicap system is unfair.

When you come up with a system that produces a 50-50 match between any two random players such that two teams can play five of those matches in a few hours, let me know and I'll GIVE you my APA franchise.

The point of the handicap system is to give everyone a CHANCE, but nobody says that's a 50-50 chance. The point of the race chart is to get the matches done in a reasonable amount of time. I don't need a race chart to tell you who will probably win a match between two players with different skill levels.

 

[KRJ]

Hi guys, long-time occasional lurker and first-time poster.

.

Well, just for kicks and giggles, start keeping track using another "formula" and see how it compares.

Try it the ACS was. In 8 ball, you get 10 points for the win, and an extra point for every ball your opponent did not make. The loser gets one point for each ball they made.

Thus, best score is a runout of 17-0. This really puts folks closer together. Because two 7's in APA could really be far apart. But in ACS, 2 guys might have the same win percentage, but one runs out more, thus collects more points, giving his opponent even less points. They are NOT rated the same.

Those 2 guys are NOT the same, and the ACS will show that because there is a broader spectrum to rate players. Thus, you really will find out the great 5's are and the weak 5's are, same with every level.

I would think that would be great info come playoff time

Thus, compare the APA system with your system and see how the players shake out.

 

[tomker]

... For an example, tonight, if the throws go the way I want them to, I (an SL5) am going to play an SL7. One that I have no business playing even. I don't really have any sort of chance to win ...

I understand that for practical reasons there must be minimum and maximum ratings, and that means there are corner cases where the handicap system can't be fair.

But, those are just corner cases. I assume that the majority of matches in your league are between players who aren't 2s or 7s.

If the handicap system isn't fair for those matches, and isn't even designed to be fair, then that's a disappointment. My understanding is that handicap systems in other sports like golf, bowling, and car racing give participants a fair (50-50) chance. I don't know why that wouldn't be the goal for APA matches too.

That being said, I just ran the numbers for my team this season and see that there have only been 10 upsets out of the 27 matches we've played between unequal skill levels. That's only 37% which is indeed not very close to 50%.

 

[tomker]

...
This is a neat exercise, but I think you should take into account innings just like the APA does. It's very simple, just use innings per win in 8-ball. This is the only way to determine how well you played. I'll explain:
...

If you can win games in 1-2 innings but you lose most of your games and matches, how well are you really playing? Sure, you might have the ability to play "pretty" sometimes, but the point of pool is usually to win games.

The reason I made my statistic/page is to generate a number based on games won/lost which is the only thing that ultimately matters in pool, in my opinion.

 

[tomker]

...
Try it the ACS was. In 8 ball, you get 10 points for the win, and an extra point for every ball your opponent did not make. The loser gets one point for each ball they made.
...

I've heard of systems that take into account the number of balls left on the table but I've never understood the idea behind that.

Sure, if you're just playing "bar pool" and hitting in as many balls as you can before somebody wins, then I'm sure it's a great system.

If you're playing with some idea of strategy then the number of balls somebody has on the table at any given time seems almost irrelevant?

 

[railbird99]

I understand that for practical reasons there must be minimum and maximum ratings, and that means there are corner cases where the handicap system can't be fair.

But, those are just corner cases. I assume that the majority of matches in your league are between players who aren't 2s or 7s.

If the handicap system isn't fair for those matches, and isn't even designed to be fair, then that's a disappointment. My understanding is that handicap systems in other sports like golf, bowling, and car racing give participants a fair (50-50) chance. I don't know why that wouldn't be the goal for APA matches too.

That being said, I just ran the numbers for my team this season and see that there have only been 10 upsets out of the 27 matches we've played between unequal skill levels. That's only 37% which is indeed not very close to 50%.

This has been stated several times already. There are two main reasons it can't be 50/50.

First, since there are only 6 discrete handicap levels in 8-ball, it means every single skill level is going to have players of different strengths (e.g. very rarely are you going to have two 3's that are exactly the same strength). Therefore, how can all matches be 50/50 if the race is the same even though playing strength varies within the same skill level.

Second, because matches have to be relatively short to get all matches done in one night, and the APA needs to allow for any two players to play against each other regardless of skill level, you are very limited in how much you can spread the amount of games between a single skill level difference (e.g. maybe a 5 playing a 4 should be more than one game difference to achieve 50/50, but that would mean a 4 playing a 6 would have to be even more, etc, matches would be way too long to complete in the allotted time).

If you want to compare pool to other sports, make sure you take into account the type of game they are playing. In 8-ball and 9-ball, there is only so many ways you can choose to handicap the game, without changing the game itself. For example, with golf and bowling, it's easy to handicap because each person gets an individual score that can have a lot of variance, so you can handicap it based on strokes/points. In those games you are really playing separately from your opponent, and you have an individual score. 8-ball can really only be handicapped by games, and that can't vary very much due to time constraints.

 

[tomker]

...
If you want to compare pool to other sports, make sure you take into account the type of game they are playing. In 8-ball and 9-ball, there is only so many ways you can choose to handicap the game, without changing the game itself. For example, with golf and bowling, it's easy to handicap because each person gets an individual score that can have a lot of variance, so you can handicap it based on strokes/points. In those games you are really playing separately from your opponent, and you have an individual score. 8-ball can really only be handicapped by games, and that can't vary very much due to time constraints.

Sorry if this is beating a dead horse. I haven't seen any other threads on the board about this subject but I didn't search for it specifically and I'm an infrequent reader. My apologies. I am happy to table the discussion at any point.

I understand that due to the low number of ratings and games, the races can't give you literally 50-50 odds out to 3 decimal places.

But if higher-rated players are going to have a 63% chance of winning then that's a pretty enormous advantage and one that seems correctable just by adding or subtracting an extra game to the race.

If the ELO rating system is used for pool then it IS possible to generate a handicap table, mathematically, that IS fair:

http://www.fargobilliards.com/pool-tournaments/fargo-ratings/

 

[railbird99]

I understand that due to the low number of ratings and games, the races can't give you literally 50-50 odds out to 3 decimal places.

But if higher-rated players are going to have a 63% chance of winning then that's a pretty enormous advantage and one that seems correctable just by adding or subtracting an extra game to the race.

You can't single out one match or one skill level combination (e.g. a 5 playing a 4) and say you can correct the difference by adding or subtracting a game just in that case.

If you determine that a 5 has a 63% chance to win a match against a 4 in a 4-3 race, and you adjust it to 5-3 or 4-2, now you have to adjust all of the other possible handicap differences all the way up to a 7 playing a 6, and if you make it 4-2 which is shorter, then what do you do for a 4 playing a 3 and 3 playing a 2, because you need to at least require the lower player to win 2 games. a 4-2 match between a 2 and 3 will take a while. In short, that small change messes up the entire handicap board.

Matches in my area already take 4 hours on average, and this is what I was trying to explain in the last post. One small change affects the entire handicap table, and causes a dramatic time difference in the overall match.

If the ELO rating system is used for pool then it IS possible to generate a handicap table, mathematically, that IS fair:

http://www.fargobilliards.com/pool-tournaments/fargo-ratings/

I have and always will be a proponent of using an ELO-based rating system for pool players. It let's all players know exactly how good they are relative to other players and it gives them a single number to track their progress as they improve.

However, as far as handicapping the 8-ball match based on ELO ratings, it still suffers from the same limitations, because there are only a limited number of discrete handicap levels. Yes, that system may work better than the APA because there are 3 more handicap levels, but you also have to realize that they only play 4 matches a night while the APA plays 5. Also, with 4 matches, it is possible to tie, where in the APA format, there is always a winner.

Furthermore, the ELO system only makes it more clear that there is no way to have a 50/50 match. Each handicap level includes a wide range of rating difference. So if I'm playing someone rated 34 points higher than me, it's still an even 5-5 race. You could argue that the difference is negligible, but it's still much more than one percent. You used 63% as being unreasonable, so where do you draw the line? It's all subjective.

 

[tomker]

... One small change affects the entire handicap table, and causes a dramatic time difference in the overall match.

Unfortunately we can't know this with any certainty because the APA rating system is secret, so we don't know the actual odds of one rating beating another, in one game or in a match. So we don't know what would have to change to make the handicap table more fair.

...
Furthermore, the ELO system only makes it more clear that there is no way to have a 50/50 match. Each handicap level includes a wide range of rating difference. So if I'm playing someone rated 34 points higher than me, it's still an even 5-5 race. You could argue that the difference is negligible, but it's still much more than one percent. You used 63% as being unreasonable, so where do you draw the line? It's all subjective.

Yes, point taken re: granularity of ratings and matches.

Certainly for a specific match, one player will necessarily have a few percent advantage vs. another player. I understand the practical concerns here.

But across an entire season, it seems reasonable to expect these advantages to balance each other out, and lower rated players should be able to beat higher rated players close to 50% of the time.

 

[KRJ]

I've heard of systems that take into account the number of balls left on the table but I've never understood the idea behind that.

Sure, if you're just playing "bar pool" and hitting in as many balls as you can before somebody wins, then I'm sure it's a great system.

If you're playing with some idea of strategy then the number of balls somebody has on the table at any given time seems almost irrelevant?

because the better players run out more, and leave more opponent balls on the table. Thus, wining 10- 7 or 11-6 is good, but the folks that are winning 17-0 or 16-1 or 15-2 are quite better players. Even though technically they would both be APA 7's, but in the real world they are far apart. The ACS can separate those based on using this system, and it's hard to "game" since you can't suddenly not run out, because you may lose the round.

I've yet to meet or find anyone claim they can sandbag in an ACS league... just not possible imho.

 

[jeffj2h]

Tomker,

Sounds like you have gathered a lot of stats so perhaps you could do a more thorough analysis than what I did 3 years ago.

I posted stats on AZ from one season of my APA 8-ball league. I concluded that the Equalizer system did a pretty good job of nudging players towards a 50% win rate. SL's of 2-3 had lower win rates because they are still figuring out which end of the cue is which. SL's of 6-7 had higher win rates because they were skilled enough to overcome the handicap. Middle SL's were close to 50% win.

My analysis is at
http://forums.azbilliards.com/showthread.php?t=271656

-Jeff