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

[jeffj2h]

BTW, the APA Equalizer document everyone is talking about is, I believe, a copy of a fax of a rec.sport.billiard post from October 2003. That document describes the Equalizer system for 8-ball. Basically it was they take only games you have WON and average the "innings minus safeties" for those games. Then compare it to a table (e.g. a SL7 will have a innings minus safeties of 0.0 to 2.0). But the LO has discretion to move this up or down.

I have not seen an equivalent description of the 9-ball Equalizer system. But it would seem to make sense that they average your "innings minus safeties" and compare it to some number. I've heard a rumor the number is 23, but I have no idea if that's true. So if you win your 9-ball match in <23 innings minus safeties, you did better then your current SL. If you win your 9-ball match in >23 innings minus safeties, you did worse then your current Sl. Would not matter if you are a SL2 or SL9; they would always compare you to 23. I'm not sure how or if they look at match loses in 9-ball.

The APA has access to data from many thousands of matches. I imagine years ago they could have hired a mathematician to develop a system that is a lot more complex then the above.

Again, that 8-ball document is over 11 years old, and my 9-ball theory is my own musings.

 

[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

Cool stats and post. You definitely collected more data than I have. So far I just have score sheets for my own team for the current session.

I would definitely like to get score sheets from all the teams in my division but it seems awkward to ask all the captains for their score sheets (or cell phone pictures of them, anyway)--I don't want anybody to think I'm somehow trying to use the data against them. I wonder how you got all your data?

In terms of everybody averaging around 50%, I think that's partially the Equalizer system and partially just self-regulation. For any given match, a team captain will throw somebody he thinks has a good chance of winning without burning too many rating points, so it shouldn't be a huge surprise if everybody averages out to ~50% over time even if the rating system is kinda crappy, because there are opposing forces regulating things.

I think a better metric would be, what percentage of the time do lower rated players beat higher rated players, and vice versa. For the handicapping system to be fair, I think the percentages should work out to 50-50.

I posted that for the data I have, it's 37-63, which doesn't seem great, but that's only for 27 matches, so it's a pretty small sample size.

 

[jeffj2h]

I wonder how you got all your data?

The APA web site, members.poolplayers.com. From there I could get the SL, number of wins, number of losses, and total points for all 75 players in my division. The total points is the sum of the SL's a player beat. Knowing that and the number of matches a player won, I knew the average SL of the players they beat.

The data was for 13 weeks of play.

 

[railbird99]

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.

I'm not saying we know the odds of one rating beating another, that's what your post seems to be about. I'm saying IF you concluded that one player had way more than a 50% chance of winning, and you chose to adjust the race, the ripple effect of that would be changing the entire handicap board which would lead to matches exceeding time constraints. This is why you can't just adjust the games like you suggested.

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.

I'm not following you here. The percentages are never going to "balance out". It's a broken system.

If you want to know what a better solution is, without changing the game itself, let's go back to ELO ratings and chess. In chess there are no handicaps. What they do is match people up who are similar strength, so for example if there is a tournament you would have a class A with 1901-2100 players, class B with 1701-1900 players, etc. This way you never have a 3 playing a 7. In team matches, you would have board 1, board 2, and board 3, and board 1 would have the best player from each team, board 2 second best, etc. The way to fix this is not by trying to get 50/50 matches but to have everyone play people close to their skill level.

You could still have a 1700 player playing against a 1900 player even, but the incentive is that the if the 1700 player wins he will gain more ELO points than if the 1900 player wins, and won't lose many ELO points in the probable outcome where he loses.

 

[railbird99]

The APA web site, members.poolplayers.com. From there I could get the SL, number of wins, number of losses, and total points for all 75 players in my division. The total points is the sum of the SL's a player beat. Knowing that and the number of matches a player won, I knew the average SL of the players they beat.

The data was for 13 weeks of play.

Maybe your division is setup differently on their website, but I can only see wins/losses for members of my team, and just the SL's and matches played for the rest of the division.

 

[tomker]

I'm not saying we know the odds of one rating beating another, that's what your post seems to be about. I'm saying IF you concluded that one player had way more than a 50% chance of winning, and you chose to adjust the race, the ripple effect of that would be changing the entire handicap board which would lead to matches exceeding time constraints. This is why you can't just adjust the games like you suggested.

Sorry, I'm still not following you. Each cell in the handicap table represents a different matchup and can thus be considered individually, no? I don't see why changing one race means you have to change all the others, since they're not dependent on each other. And I don't see why any changes to the handicap table mean the matches will take longer. What if you reduce the number of games in the race?

I'm not following you here. The percentages are never going to "balance out". It's a broken system.

The supposition is that the races are currently unfair because the higher rated player is more likely to win. But the races could be changed so the lower rated player is more likely to win. Consider when a 4 plays a 5, it's a 3-4 race. Maybe the 5 is more likely to win that. But surely a 2-4 race would be in favor of the 4.

So the races could all be adjusted so that in half the cases the higher rated player is more likely to win and vice versa. So when you play players with different skill levels, half the time you have an advantage and the other half of the time you have a disadvantage and it all balances out.

 

[railbird99]

Sorry, I'm still not following you. Each cell in the handicap table represents a different matchup and can thus be considered individually, no? I don't see why changing one race means you have to change all the others, since they're not dependent on each other. And I don't see why any changes to the handicap table mean the matches will take longer. What if you reduce the number of games in the race?

Jeeze. I really thought I made myself clear. I'll try one more time.

The difference between a 5 and a 4 is the same as the difference between a 4 and a 3. If you determine that a 4-2 race is necessary to make a 5 playing a 4 fair, then you must also have a two game gap for all skill level differences of 1. So since a 4 playing a 3 is one skill level difference just like 5 and a 4, you must change that race from 3-2 to 4-2 as well. You can't make it 3-1, because you can't let any player win a whole match just with one game. Now that you have made a 4 playing a 3 4-2, a 2 playing a 3 must also be 4-2. The lower players also play the slowest, so that's a big difference in time for those matches. That's not all, now for skill level differences of 2, where it used to be a 2 game gap, will now have to be increased as well, so it's not the same as a 1 skill level difference.

Do you in any way get my point? It's all about the relative difference in skill levels. It makes no sense to have a 5 playing a 4 be a two game gap, but a 4 playing a 3 is only a one game difference.

The supposition is that the races are currently unfair because the higher rated player is more likely to win. But the races could be changed so the lower rated player is more likely to win. Consider when a 4 plays a 5, it's a 3-4 race. Maybe the 5 is more likely to win that. But surely a 2-4 race would be in favor of the 4.

So the races could all be adjusted so that in half the cases the higher rated player is more likely to win and vice versa. So when you play players with different skill levels, half the time you have an advantage and the other half of the time you have a disadvantage and it all balances out.

Let me get this straight. As is stands now, the lower player has a significantly better chance to beat a higher rated player than if they played even, it's just not 50/50, and your solution to this is to deliberately make all matches be unfair, to one player or the other to somehow make it overall 50/50? So you feel like people won't care that every individual match they play will be unfair towards someone? If you are really being serious, I want some of whatever you're on.

 

[jeffj2h]

Maybe your division is setup differently on their website, but I can only see wins/losses for members of my team, and just the SL's and matches played for the rest of the division.

The APA in my area (Plano, TX) adopted a new top-gun formula a few years ago. So I can no longer see the "total points" stat. However I can still click on a team name, see all their players, see SL's, number of matches played, and number of matches won.

 

[tomker]

Jeeze. I really thought I made myself clear. I'll try one more time.

The difference between a 5 and a 4 is the same as the difference between a 4 and a 3.

Ah, okay, here's the part of your argument that I wasn't getting and completely disagree with.

A 1 point difference in rating does not translate to a certain probability of winning a game. Look at the race between a 2 and a 3, it's 2-3. There's a 1 point rating gap but a HUGE difference in skill there. The stronger player is expected to win 50% more games. That percentage gradually decreases and you end up with a 6 vs. 7 match which is a 4-5 race and the stronger player is only expected to win 25% more games.

So it's clearly not a linear scale and you can't look at the absolute difference between ratings and make assumptions about how they must affect the handicap table.

Let me get this straight. As is stands now, the lower player has a significantly better chance to beat a higher rated player than if they played even, it's just not 50/50, and your solution to this is to deliberately make all matches be unfair, to one player or the other to somehow make it overall 50/50?

Actually I was agreeing with one of your points so your tone here is annoying.

Earlier you made the point that races can't be 100% fair (i.e., 50-50) because there isn't enough granularity of ratings or matches in a race, i.e., what if you have a strong 5 playing a weak 5 and they both have to win 4 games? That's not 50-50, and I agree, because of practical considerations.

So no, I'm not saying the APA should make all the races unfair deliberately. They just need to balance out the unfairness so it doesn't predominantly go in one direction.

 

[railbird99]

My tone is going to become increasingly more annoying the more I have to repeat myself. I know you think you're on to something but I have been dealing with tournaments and handicapping well before you started thinking about it, and you're not the first person to delve into the subject. I also happen to be a software developer and I have created software designed for this very purpose.

Ah, okay, here's the part of your argument that I wasn't getting and completely disagree with.

A 1 point difference in rating does not translate to a certain probability of winning a game. Look at the race between a 2 and a 3, it's 2-3. There's a 1 point rating gap but a HUGE difference in skill there. The stronger player is expected to win 50% more games. That percentage gradually decreases and you end up with a 6 vs. 7 match which is a 4-5 race and the stronger player is only expected to win 25% more games.

So it's clearly not a linear scale and you can't look at the absolute difference between ratings and make assumptions about how they must affect the handicap table.

Yes, it is linear. Allow me to explain:

The APA algorithm has changed over the years, but it is still based on basically the same metric, innings per win. Your skill level, even though it's a single number, really represents a range of this metric under the hood (just like each handicap level represents a range of ELO ratings in the league you mentioned). Let's say a 7 is anyone who averages 2 innings or less, per game that they win. Then a 6 could be 2-3.5 innings, a 5 would then be 3.5-5, etc. The one thing that remains constant in each skill level is the range, which in my example is 1.5 innings, except for the highest and lowest skill levels.

Therefore, the difference between 5/4, and then a 4/3 is exactly the same, and is linear. The reason the difference between a 2 and a 3 may be more is because it's an "edge case" as you put it, and you can't go below a 2, so a 2 may average 100 innings per win, and will still be a 2. That's a completely different flaw in the system.

So no, I'm not saying the APA should make all the races unfair deliberately. They just need to balance out the unfairness so it doesn't predominantly go in one direction.

Again, that's the whole issue. It is as much a function of how the game of 8-ball is played as it is of the handicapping system. There has to be compromises made somewhere, and since you can't achieve 50/50, the next best thing is just giving the lower player a chance to win. If it is such a big issue, try to throw players in such a way that they are always playing someone close to their skill level.

 

[tomker]

...
Yes, it is linear. Allow me to explain:

The APA algorithm has changed over the years, but it is still based on basically the same metric, innings per win.

Okay. You might be right. You sound very sure of this even though the Equalizer system is supposed to be secret, so who knows, but I won't argue with you about this.

The problem is that you've now just introduced a completely new dimension to the discussion, one that hasn't existed before. Maybe you were writing your posts with the assumption that I knew about this inning stuff already, which probably explains some of your frustration.

If the rating system is based on innings and the handicapping system is based on wins, then that's so fundamentally broken that it's probably pointless to continue discussing it.

You seem to be familiar with chess so to make a chess analogy, that would be almost exactly like adjusting somebody's Elo rating according to how many moves it takes him to win a game. That would be a mind-bendingly stupid way to do things and it wouldn't be any less stupid for pool.

 

[railbird99]

Okay. You might be right. You sound very sure of this even though the Equalizer system is supposed to be secret, so who knows, but I won't argue with you about this.

The problem is that you've now just introduced a completely new dimension to the discussion, one that hasn't existed before. Maybe you were writing your posts with the assumption that I knew about this inning stuff already, which probably explains some of your frustration.

If the rating system is based on innings and the handicapping system is based on wins, then that's so fundamentally broken that it's probably pointless to continue discussing it.

You seem to be familiar with chess so to make a chess analogy, that would be almost exactly like adjusting somebody's Elo rating according to how many moves it takes him to win a game. That would be a mind-bendingly stupid way to do things and it wouldn't be any less stupid for pool.

I'm sorry if I seemed a little irritated or arrogant, but yes I've been in the APA for a long time, and I actually studied the leak that came out back in the 90's, and I know from the inside that it has definitely changed, but not that much. It has always been widely known that it is mostly based on innings.

You seem intelligent, just not familiar with the APA and handicapping. If you really think about it and try to understand it, it may not be perfect, but it actually makes a lot of sense.

The better the player you are, the less amount of turns at the table is required, on average, to run all of the balls and win a game of 8-ball. If John beats Sally 3-0, you have no idea what skill level they are, just that it seems like John is much better than Sally. However, if you took into account that the match had an average of 20 innings per game, you know that John is probably not a 7.

Here's a little more detail:

Lets say two sevens get into a safe battle so the inning are really high. Well this is where the box on the scoresheet for defensive shots comes into play. Each defensive shot cancels out an inning so that you can tell how many turns it actually took the winner to win the game when trying to make shots.

On the flip side, if a 3 beats another 3, 2-0 in 2 innings, but one game was an 8 on the break, and the other game the opponent made the 8 early, that's taken into account, so that it doesn't look like the 3 ran out twice.

It's also fairly smart in the sense that it's innings per WIN, and not just innings per game. This helps with sandbagging. If someone chooses to miss on purpose to get their innings up, their opponent will likely capitalize and win the game. The innings from the games you lose don't factor in at all. So in order to be successful at sandbagging you would have to miss on purpose while also winning the game. It makes it more difficult.

If you only went by the score, sandbagging would be easier, as you could just lose games and it would throw your rating completely off.

As far as your comparison to chess, that's completely different. Chess is not about just trying to find the quickest way to get mate. You have positional players that just build up a strong defense and play long drawn out games, but are just as skilled as aggressive players that take risks. Sometimes a weak player will go for a quick mate without knowing if it will work or not, and let's say that it ends up working, but only because their opponent let it happen. A good player will see the same mate possibility, but also see that his opponent could possibly refute it, and he will then choose a longer line that he is able to see that ultimately leads to mate. Chess would be much more complicated to try to determine skill level based on the moves played in the game.

In conclusion, please use some critical thinking and actually do the research and think hard about things before you just jump to conclusions so quickly. People have been thinking about these issues for a long time.

 

[tomker]

...
You seem intelligent, just not familiar with the APA and handicapping. If you really think about it and try to understand it, it may not be perfect, but it actually makes a lot of sense.
... In conclusion, please use some critical thinking and actually do the research and think hard about things before you just jump to conclusions so quickly. People have been thinking about these issues for a long time.

I don't think any amount of deep thought is going to get around the fact that the point of pool is to win games/matches.

It's easy to measure who won a game/match. It's not like we're searching for Higgs bosons via particle backscatter. The winner is written down on the score sheet.

So if you want to design a handicapping system to even out wins and losses, it should obviously be based on wins and losses. If you want to instead base it primarily on a statistic that has a debatable degree of correlation (innings per win) then you have to have a pretty spectacular reason for doing so.

To refute your justifications for innings-per-win point by point:

1. I sort of disagree with the idea that innings-per-win indicates player strength. Clearly 2s and some 3s will rack up a ton of innings and 7s will generally have very few per game. But in between, I think it's debatable how useful of a statistic this is. It depends so much on player style. Are you the kind of player who goes for the out every shot, so if you win it's only in a couple innings, but you often lose? Or are you the kind of player who's very patient and strategic and plays a lot of 2-way shots and usually win?

2. I disagree that innings-per-win makes it difficult to sandbag. All you have to do to drive up your inning count is not try very hard on 2-way shots. That's a perfectly legitimate way to play and not dishonest at all and not sandbagging, but it will drive down your APA rating. Ridiculous.

3. If you go by win/loss record, you could update a rating based on the result of a match and not each individual game. So to sandbag you'd have to lose an entire match. That's a direct deterrent to sandbagging.

 

[railbird99]

1. I sort of disagree with the idea that innings-per-win indicates player strength. Clearly 2s and some 3s will rack up a ton of innings and 7s will generally have very few per game. But in between, I think it's debatable how useful of a statistic this is. It depends so much on player style. Are you the kind of player who goes for the out every shot, so if you win it's only in a couple innings, but you often lose? Or are you the kind of player who's very patient and strategic and plays a lot of 2-way shots and usually win?

I understand where you're coming from, and no, the APA's system is not perfect, but it does make sense, and works fairly well. I realize you feel like what you are saying makes perfect sense, but I can tell you from experience it just doesn't work like that.

First of all, if you are a player that goes for the out every time, great, on average, how many attempts at going for the out do you need to win a game? If your wins are only in a couple innings, but you lose a lot as a result of going for the out instead of playing smart, well your losses aren't as much of an indicator of your skill level as it is of your opponent's, who apparently is good enough to run out when you fail to get out. You are assuming that when this person messes up the out, the other player is instantly winning the game. They are likely to have another chance if playing a mid to low level player. If you have the capability of winning games in two innings, then it's only fair that you are higher skilled than people who are incapable of doing so.

You are placing a lot of emphasis on two-way shots. They don't come up as often as you seem to suggest. Taking smart shots that don't leave your opponent with much if you miss still gives your opponent another chance at the table, which means a chance to play safe or a chance to make a low percentage shot, which means they end up losing more often than players that have the ability to keep the innings lower. The higher skilled players limit the chances their opponent get to the table, period.

In addition, in order for you to run up innings with two-way shots, that means you are missing the majority of your two-way shots, which definitely is an indication of your skill.

2. I disagree that innings-per-win makes it difficult to sandbag. All you have to do to drive up your inning count is not try very hard on 2-way shots. That's a perfectly legitimate way to play and not dishonest at all and not sandbagging, but it will drive down your APA rating. Ridiculous.

All you are doing here is mentioning a loop hole. I already admitted sandbagging is still possible, I said innings per win makes it more difficult. You can't possibly eliminate all sandbagging. Furthermore, as I explained above, even if you play two-way shots, you are still giving your opponent a chance at the table to play safe or make a hard shot, which still means you have to risk losing the game to do this, and if you lose the game, you ran up the innings for nothing.

3. If you go by win/loss record, you could update a rating based on the result of a match and not each individual game. So to sandbag you'd have to lose an entire match. That's a direct deterrent to sandbagging.

I know that seems to make sense, but unfortunately losing matches is not a great deterrent to sandbagging. People are more than willing to lose a couple matches to go down. In fact, most people do this now not realizing that it doesn't really help them. Another reason the innings per win works well is because it's not advertised and is supposed to be "hidden".

 

[Skippy27]

No mathematical system is going to be perfect because you have the human factor keeping score and players trying to manipulate the system.

Keeping innings and safeties are a very important part of making sure players are rated how they should be. Better players shoot less innings especially when you factor in safeties. That is true of any skill match up. What needs to be watched are players that are sandbagging by purposely barely missing shots so they don't appear to be hitting safe, but none the less are and racking up the innings. When I am keeping score with 4-7 SL players I watch closely and anything I personally think they did on purpose, no matter how well they act, I put down as a safety. Sometimes their captains/scorekeeper will argue with me when they sign the sheet, but I dont care. You keep yours and I keep mine. Every team should be doing this and the ratings will be what they should be.

I am in St. Louis and the APA is acting as our LO, but before that the guy that originally owned our area would show up just about every league night to our bar and many others and watch people play. Not only in matches, but also on other practice tables. He would see how good people are and he would adjust ratings as he needed.

It is disappointing to watch a team in my division that has 2, possibly 3, players sandbagging pretty hard to avoid getting moved up. But the team is staying right in the rankings they need to so that they stay qualified for the next team Vegas qualifier. Nothing will fix that unless all the teams keep score against them correctly.

 

[tomker]

I understand where you're coming from, and no, the APA's system is not perfect, but it does make sense, and works fairly well. I realize you feel like what you are saying makes perfect sense, but I can tell you from experience it just doesn't work like that.

Eh, a rating system based on wins/losses is basically the Elo system, which works fine for much bigger sports (and bigger-stakes sports) like chess, baseball, basketball, football, etc. If your claim is that it "just doesn't work" for pool then I think the burden of proof is on you to explain your position.

So far you have just been arguing that innings-per-win is correlated to player strength which is presumably correlated to wins/losses. So your argument for using innings-per-win is that it has some correlation to the thing that we actually want to use. So why not just use that thing?

I think in an earlier post you said you were a proponent of the Elo system so I realize that you are being put in the awkward position of defending something you don't think is optimal. If you don't want to keep discussing this then I'll understand.

First of all, if you are a player that goes for the out every time, great, on average, how many attempts at going for the out do you need to win a game? If your wins are only in a couple innings, but you lose a lot as a result of going for the out instead of playing smart, well your losses aren't as much of an indicator of your skill level as it is of your opponent's, who apparently is good enough to run out when you fail to get out. ...

If you run down and mess up your out, it's not hard for a 5/6/7 to win the game. Maybe they'll have to play a couple Ds to win, maybe not. But it's not unbelievable for somebody who can win games in 1-2 innings to usually lose against 6s and 7s and be fairly rated as a 5. I know a couple people like this personally, so this isn't just a hypothetical.

You are placing a lot of emphasis on two-way shots. They don't come up as often as you seem to suggest. Taking smart shots that don't leave your opponent with much if you miss still gives your opponent another chance at the table,

I admit that this "lazy 2-way" strategy doesn't work against very good players who will manage to do something even in difficult situations but it can be very effective against anybody who isn't a stronger 6 or a 7. In fact, I do it myself fairly often so I know exactly how effective it can be. I don't do it to sandbag or run up my innings, it's just that I might not see a way to get out, so I'll be really lazy about making my shots while I try to figure out what to do. I will also try for low-percentage shots fairly often in such situations rather than hit all my ducks in and possibly be left with some problem balls and a losing situation.

I know that seems to make sense, but unfortunately losing matches is not a great deterrent to sandbagging. People are more than willing to lose a couple matches to go down. In fact, most people do this now not realizing that it doesn't really help them. Another reason the innings per win works well is because it's not advertised and is supposed to be "hidden".

I didn't know that innings-per-win was the main variable in the rating system but I think maybe a lot of people do. Luckily I don't think anybody in my local division is focused on sandbagging but when I go to other locations and to tournaments, there are tons of people who talk about padding their innings and get upset if you mark down Ds for them.

Since you said you're a software developer, surely you're familiar with the idea of "security through obscurity" and the reasons why it's a bad idea. The APA rating system is based on this faulty premise and people are hacking it just as they would a closed security system.

 

[tomker]

... What needs to be watched are players that are sandbagging by purposely barely missing shots so they don't appear to be hitting safe, but none the less are and racking up the innings. ...

There's a difference between purposely missing shots to keep your rating down and just not trying very hard to make certain shots. Or strategically choosing to take low-percentage shots.

 

[railbird99]

Eh, a rating system based on wins/losses is basically the Elo system, which works fine for much bigger sports (and bigger-stakes sports) like chess, baseball, basketball, football, etc. If your claim is that it "just doesn't work" for pool then I think the burden of proof is on you to explain your position.

So far you have just been arguing that innings-per-win is correlated to player strength which is presumably correlated to wins/losses. So your argument for using innings-per-win is that it has some correlation to the thing that we actually want to use. So why not just use that thing?

I think in an earlier post you said you were a proponent of the Elo system so I realize that you are being put in the awkward position of defending something you don't think is optimal. If you don't want to keep discussing this then I'll understand.

Yes the ELO system could be used for pool, and it could work very well, but that does not mean the APA's system is completely flawed. What you aren't realizing is that using the ELO system is not actually going to help with handicapping the races. The ELO system will not directly translate into a 50/50 race no more than the APA's sytem will. The games you mentioned that use the ELO system do not use it to handicap the game, just to determine who the best players/teams are relative to other players. The games themselves still aren't handicapped. This means there is no reason for players/teams to sandbag in those games. The APA is fundamentally different than the sports you are mentioning. I've already told you how easy it would be to sandbag using an ELO system, and this is a huge factor that you seem to be ignoring.

I'm not arguing for something that I don't believe in. I think the APA's system is flawed, but it also works well for its intended purpose. I also like the ELO system, and hope that someday, at least at the highest level of pool, a form of the ELO system is used.

If you run down and mess up your out, it's not hard for a 5/6/7 to win the game. Maybe they'll have to play a couple Ds to win, maybe not. But it's not unbelievable for somebody who can win games in 1-2 innings to usually lose against 6s and 7s and be fairly rated as a 5. I know a couple people like this personally, so this isn't just a hypothetical.

We can just agree to disagree on this one. There are edge cases where the handicap system may not work perfectly, but the basic premise holds true 99% of the time. If you run down to your last ball and mess up the out, and a 5 then wins the game, well that means that he just won the game in 1-2 innings, and it will go towards moving him up because 5's don't run out in 1-2 innings on average. If the 5 plays smart and plays defense, and that's what wins him the game, well every defense shot cancels out an inning, so the 5 is still winning in 1-2 innings. Two-way shots may get the 5 more chances at the table while keeping innings up, and that may be an edge case that works in favor of the 5 a small percentage of the time, since two-way shots aren't going to win the game. He eventually has to run the balls out without giving you another chance at shooting your last ball. Furthermore, a 6 or 7 would be able to run out a wide open table when you mess up your run, so this 5 should definitely not be a 6 or 7 if it takes him 20 two-way shots to get out.

I admit that this "lazy 2-way" strategy doesn't work against very good players who will manage to do something even in difficult situations but it can be very effective against anybody who isn't a stronger 6 or a 7. In fact, I do it myself fairly often so I know exactly how effective it can be. I don't do it to sandbag or run up my innings, it's just that I might not see a way to get out, so I'll be really lazy about making my shots while I try to figure out what to do. I will also try for low-percentage shots fairly often in such situations rather than hit all my ducks in and possibly be left with some problem balls and a losing situation.

Again, this will only be an issue for 5's most likely. If you are good enough to see the two-way shots and execute them as to never leave your opponent a shot, but you aren't good enough to negotiate the rack and run out without letting your opponent to the table, well that's an "edge case" that probably only applies to some 5's, and because of their inability to run out without "lazily" figuring things out first, they should be a 5.

I didn't know that innings-per-win was the main variable in the rating system but I think maybe a lot of people do. Luckily I don't think anybody in my local division is focused on sandbagging but when I go to other locations and to tournaments, there are tons of people who talk about padding their innings and get upset if you mark down Ds for them.

Since you said you're a software developer, surely you're familiar with the idea of "security through obscurity" and the reasons why it's a bad idea. The APA rating system is based on this faulty premise and people are hacking it just as they would a closed security system.

First of all, the ELO system you keep promoting would be 10 times worse as far as sandbagging goes, and much easier to exploit. I did not say innings per win worked well because it was hidden. I merely said it doesn't hurt that it's hidden. I originally fully explained why innings per win makes sandbagging more difficult, and it has nothing to do with being hidden. It also isn't perfect, but no system is, and there will always be some way to exploit it.

 

[tomker]

Yes the ELO system could be used for pool, and it could work very well, but that does not mean the APA's system is completely flawed. What you aren't realizing is that using the ELO system is not actually going to help with handicapping the races. The ELO system will not directly translate into a 50/50 race no more than the APA's sytem will.

Asked and answered. Differences in ELO ratings translate mathematically to different probabilities of winning individual games which translate directly to handicapping matches. I already posted a link to the Fargo page that shows races based on rating differences.

I will agree, as I have 10 times already, that you are not going to hit 50-50 without playing 100 games, just because of the quantization of ratings and games. But the Fargo races will get pretty close to 50-50. (And you can easily determine how close each one gets with a few equations.) It seems that everybody agrees that the APA races don't even try to even the playing field, so using math to get closer seems like a better system, right? Or are you completely happy with uneven APA races in their "handicapped" system?

I've already told you how easy it would be to sandbag using an ELO system, and this is a huge factor that you seem to be ignoring.

Nope, not ignoring it. I've already addressed it a couple of times but I'll do it again. First, to manipulate an Elo-based system, you have to throw matches which have a direct impact on your team's score. So that's a direct deterrent to sandbagging, instead of the APA's system where we just have to assume that there's some kind of indirect deterrent.

Second, other games/sports address sandbagging by having rating floors. If this is good enough for chess, where there's much more money involved and thus a bigger incentive to sandbag, then why in the world wouldn't this also be good enough for pool? Do you somehow think that pool is the only game/sport that has sandbaggers and nobody who plays any other game/sport has given any thought to the problem?

As for how good innings-per-win is at approximating skill, we could easily determine this mathematically if we had access to a bunch of match data. But unfortunately I don't. If anybody else does I would be happy to do some analysis.

 

[justadub]

There's a difference between purposely missing shots to keep your rating down and just not trying very hard to make certain shots. Or strategically choosing to take low-percentage shots.

"Not trying very hard to make shots" is most certainly a method of sandbagging. Its a matter of not trying to make the shot, pure and simple. If you accept that it is OK, people will (and are, I suppose) padding their innings by "just missing".

I've seen that, plenty. Difficult to mark too many of them as defensive shots, because they are making a good performance of it. Still, it is a function of not trying to make the shot for the purposes of driving up the inning count.

Now two way shots, or taking the more difficult shot (while still trying to make it) are less overt ways of making the match longer. While the spirit or intent is the same, if the shooter is actively trying to make the ball on the more difficult shot, that satisfies the requirement. Because someones judgment on which shot to play isn't in and of itself something that can be faulted....(of course save for those who make it far too obvious and obnoxious)

In my humble opinion only. Those who wish to sandbag will definitely sandbag, and justify it anyway they can. While our area isn't rife with the problem, I've seen it enough to annoy me. I can't imagine dealing with it in the parts of the country where it does go to a ridiculous level.