How Does an APA 7 Go Down to a 6?

hang-the-9

AzB Silver Member
Silver Member
HELL YES!!!! A SL 7 is expected to lose to another 7? If that SL 7 is WEAKER, then he should be a 6. This is the jist of my beef with APA ratings system. Weaker players SHOULD NOT have to play more skilled players even up. Would you gamble with a more skilled player without getting some weight? Not unless your mad at your money. All you’ve stated proves my point; the APA will raise someone’s hcp far quicker than they will lower it.

This will just lead to bouncing around in skill levels all the time, one week you are a 7, then you lose to a 7 go to a 6 then you beat a 6 and go up to a 7. This is really tough on APA since there is a large skill level between the higher ranks. A 7 can be a B player or an A++ player. Plus it's silly to assume that if a 7 loses to another 7 they are worse or need to move down, SOMEONE has to lose that match even if they two players are dead even in skill. And since the 7s in APA are so widely spread out, even if a 7 loses to three other 7s because they happen to be much better, does not mean they should be a 6 when compared to the other 6s in their league area. They can be higher than other 6s but weaker than other 7s, then what happens?

What the skill ratings should do is track your overall long term ratings and not move around so much, and also be more in tune with the real world like Fargo tries to do. Instead of going by hard numbers for the races, they take into account the actual difference between the players. For example a weak 5 vs a strong 5 has a tougher match than a strong 5 vs a weak 6 but one gets a handicap and one does not. Fargo takes care of that by going with differences in skills in 1/100th and ratings between players not just a large range. So if a 599 plays a 601 they don't automatically get a spot because one is in 500 range and other is in 600 range, but in leagues with a lesser value range you can have a guy that is a 6.99 play a 7.01 with a spot, but a 6.01 vs a 6.99 is even, even though the latter is a much greater difference in skill.
 
Last edited:

APA Operator

AzB Silver Member
Silver Member
HELL YES!!!! A SL 7 is expected to lose to another 7? If that SL 7 is WEAKER, then he should be a 6. This is the jist of my beef with APA ratings system. Weaker players SHOULD NOT have to play more skilled players even up. Would you gamble with a more skilled player without getting some weight? Not unless your mad at your money. All you’ve stated proves my point; the APA will raise someone’s hcp far quicker than they will lower it.
This is hard to argue with. Not because it's right, but because it's hard to make the argument without sounding like I'm insulting your intelligence, I just can't find words kind enough. Your last sentence is a point with which I agree, except I'd phrase it as "it's easier to go up than down". It just is, and it should be.
 

pogmothoin

AzB Gold Member
Gold Member
Silver Member
There are a couple of problems with the Equalizer system (ES) in my opinion. #1 is that the ratings are really local to the league you play in. So if you play in a league where players are generally weaker and less skilled your rating will increase quicker than if you play in a stronger league with more skilled players. #2 - It is not really based on your level of proficiency, I've seen 5's that can't execute a stop shot or move a cue ball to play basic position, while other 5's have all the shots but have other weaknesses in their game. It should generally be harder to move up from one level to the next. Winning is NOT always an indication of skill, you might still suck but the other person sucked more.

It is my understanding that going down is really difficult due to the use of applied scoring. Meaning that no matter how you shoot the ES will give you an applied score based on your current rating and winning percentage. So if you're a 5 with a winning percentage of 60% and you shoot like a 3, win or lose, on a given night the ES will still give you a score in the 5 range.

All that said league pool is still supposed to be about having a good time. So take it for what it is and play or get frustrated with it and quit.
 
Last edited:

APA Operator

AzB Silver Member
Silver Member
#1 is that the ratings are really local to the league you play in. So if you play in a league where players are generally weaker and less skilled your rating will increase quicker than if you play in a stronger league with more skilled players.
Can you name a system where this is not true? I'd say it's less true than in any other system I know of, but I do know it could happen. Years ago APA found an area that was tiering their league play this way, strong players in one division and weak in another, and it did result in teams from that area that seemed to be outliers on each end of the "team strength" spectrum pretty regularly (they didn't mix the two tiers of teams in their WQ's, so their representation at nationals consisted of both 'strong' teams and 'weak' teams). APA discovered this occurrence in their national tournaments through their observer program, and sent a representative to that area to work with that LO - the result was elimination of tiered play.

The APA's use of observers at national events has been a huge benefit to them. LO volunteers from all over the APA, big ponds, small ponds, U.S., Canada watch individual matches and grade individual players totally subjectively. All observers are highly skilled players who grade different aspects of a player's skills (mechanics, speed, spin, strategy, successful defense, unsuccessful defense, luck shots, long shots, cuts, banks, etc.), coming up with their assessment of skill level in the end. The observer does not know what size each player's pond is and doesn't know the player's skill level number on the sheet, and they don't observe anyone from an area with which they have familiarity. This gives the APA data to compare area versus area and team against team. The majority of observer ratings are spot on or even a little low (players play bad matches way more than they play above their heads) compared to the number, but from the data APA can see trends like a player or team that's consistently high or low, an area that's consistently high or low, even an observer who's consistently high or low with his/her rating. Observers can't raise/lower anyone's rating but do give the APA a point of view from which they can spot trends and make decisions. The APA identified the area above through observers. They noticed half the teams consistently high and the other half consistently low. This allowed them to investigate further and led to identification and elimination of tiered division play. They work with areas that are consistently high or low to find out why, or to find out why specific teams are high or low. That's one of the things APA does to "equalize" across ponds - it's been in place for about 20 years and isn't a secret anywhere. Observer-type programs have been in place locally and across regions for longer than that. Think about it - every time you decide whether a player is or is not properly rated, you're doing the same thing. The APA just formalized it and came up with an objective way to include some subjectivity.

#2 - It is not really based on your level of proficiency, I've seen 5's that can't execute a stop shot or move a cue ball to play basic position, while other 5's have all the shots but have other weaknesses in their game.
Hence grading aspects of your play. The combination of the different aspects is what makes someone a '5', and it's not always the same combination. It shows in wins and losses, and in other things we measure on the scoresheet. Why do you think that's a problem?

It should generally be harder to move up from one level to the next. Winning is NOT always an indication of skill, you might still suck but the other person sucked more.
It is. Correct me if I'm wrong, but doesn't this say you shouldn't automatically go up a little if you win or down a little if you lose? I mean, if you just sucked less it doesn't necessarily mean you're better than the other player or even better than your rating, right?

It is my understanding that going down is really difficult due to the use of applied scoring. Meaning that no matter how you shoot the ES will give you an applied score based on your current rating and winning percentage. So if you're a 5 with a winning percentage of 60% and you shoot like a 3, win or lose, on a given night the ES will still give you a score in the 5 range.
You need to change your understanding here. All APA does is cap how badly you can win or how badly you can lose, which is standard anti-sandbagging procedure. Look at golf, for example. With my handicap, I can't record a triple-bogey even though that's what I might actually shoot. For the purposes of posting it becomes a double-bogey. That keeps me from making lots of birdies and pars, and one or two 10's to manage my handicap. In APA's case we keep both numbers but use the capped one in the calculation. And yes, it does make it harder to go down. It's supposed to - hence the term anti-sandbagging.
 

mikepage

AzB Silver Member
Silver Member
pogmothoin said about APA:
#1 is that the ratings are really local to the league you play in. So if you play in a league where players are generally weaker and less skilled your rating will increase quicker than if you play in a stronger league with more skilled players.


Can you name a system where this is not true?

[...]

This is not true of 99+% of leagues/divisions that use Fargo Ratings.

The reason I don't say 100% is if you have a group of players that is TOTALLY isolated, then we have no idea how they play compared to the outside world. Reality, though, is even groups that people imagine are isolated are not. Your 50-player group in Mayberry RFD seems isolated. But 3 of you have played tournaments in Raleigh or Elizabeth City. Another 3 have played in a league tournament in Vegas. Then there is the guy who moved there from Ohio and one other player who moved away to Florida and another who retired to Arizona continues to play. There are more of these connections than most people have any idea.
 

Tom1234

AzB Silver Member
Silver Member
I can't record a triple-bogey even though that's what I might actually shoot. For the purposes of posting it becomes a double-bogey. That keeps me from making lots of birdies and pars, and one or two 10's to manage my handicap. In APA's case we keep both numbers but use the capped one in the calculation. And yes, it does make it harder to go down. It's supposed to - hence the term anti-sandbagging.
Your golf analog; I’m 75 years old and I play the back tees where handicaps are determined. 35 years ago I was a scratch golfer. Now, playing at 7400 yards, the BEST I can do is triple freakin bogey! Your analogy means that I give strokes to a 7 handicapper!! How crazy is that. First, no golfer in his right mind would approve of that situation. Second, golfers have handicap adjustments WHEN THEY AGE AND SKILLS DIMINISH! Hell, we even have a SENIOR tour for PRO golfers. Please, have some common sense when looking at keeping someone’s skill level higher than their true ability. Skills do diminish when we age.
 

garczar

AzB Silver Member
Silver Member
I still find it amazing(to say the least) that any pool league that requires this level of explanation as to handicapping gets ANYONE to play. I quit leagues yrs ago mostly because of all the nitty drama and they were all heads-up/no handicap. I've read most of this thread and it just re-enforces why i quit. There HAS to be a system that is BOTH simple and equitable. Nothing i've read here looks to be either one.
 

APA Operator

AzB Silver Member
Silver Member
pogmothoin said about APA:
#1 is that the ratings are really local to the league you play in. So if you play in a league where players are generally weaker and less skilled your rating will increase quicker than if you play in a stronger league with more skilled players.




[...]

This is not true of 99+% of leagues/divisions that use Fargo Ratings.

The reason I don't say 100% is if you have a group of players that is TOTALLY isolated, then we have no idea how they play compared to the outside world. Reality, though, is even groups that people imagine are isolated are not. Your 50-player group in Mayberry RFD seems isolated. But 3 of you have played tournaments in Raleigh or Elizabeth City. Another 3 have played in a league tournament in Vegas. Then there is the guy who moved there from Ohio and one other player who moved away to Florida and another who retired to Arizona continues to play. There are more of these connections than most people have any idea.

It is absolutely true in leagues that use Fargo ratings. The fact that Fargo has to indirectly adjust players up or down to level X because they play in a stronger or weaker area suggests that they would already be X in their area if it wasn't stronger or weaker. This inter-area automatic adjustment is simply an attempt to adjust for that, but it doesn't happen until it is discovered. Maybe that adjustment is faster in Fargo or maybe the difference in the rate of increase is lower in Fargo (debatable) or maybe an adjustment on a 1000-point scale is less noticeable or meaningful than an adjustment on a 6-point scale, but that wasn't what was stated. I maintain that the difference does exist, and I maintain that it exists more often in Fargo where your rating is more dependent on who you play than in APA where it's more dependent on how you play.
 

APA Operator

AzB Silver Member
Silver Member
Your golf analog; I’m 75 years old and I play the back tees where handicaps are determined. 35 years ago I was a scratch golfer. Now, playing at 7400 yards, the BEST I can do is triple freakin bogey! Your analogy means that I give strokes to a 7 handicapper!! How crazy is that. First, no golfer in his right mind would approve of that situation. Second, golfers have handicap adjustments WHEN THEY AGE AND SKILLS DIMINISH! Hell, we even have a SENIOR tour for PRO golfers. Please, have some common sense when looking at keeping someone’s skill level higher than their true ability. Skills do diminish when we age.
I've already said skills do diminish when we age. I've already said the APA has a way for your rating to reflect that, so what's your argument? I'm simply pointing out that it's not the same for everyone (I know some 75-year-olds that are still scratch golfers), so the fact that it's not automatic might not be such a bad thing. Two years ago I had some medical problems and my golf skills went back to zero overnight, but the next several times I played I still couldn't post a triple-bogey, even though that's the best I could do. I understood that and didn't accuse the USGA of lacking common sense, I accused them of protecting their system from abuse.

I'm assuming that your golf handicap has decreased as you age, so of course triple bogeys count for you. They don't for me again. Please don't accuse me of being the one without common sense. By the way, a golf handicap can be determined from any tee. That's why they all have different slope and rating numbers. Perhaps you should move up to a tee box that's more handicap-appropriate. You might have more fun, and I'm sure the group behind you would appreciate it.
 

mikepage

AzB Silver Member
Silver Member
It is absolutely true in leagues that use Fargo ratings. The fact that Fargo has to indirectly adjust players up or down to level X because they play in a stronger or weaker area suggests that they would already be X in their area if it wasn't stronger or weaker. This inter-area automatic adjustment is simply an attempt to adjust for that, but it doesn't happen until it is discovered.

This is wrong.

We think of League1 and League2 and then an adjustment between the Leagues just because that's a convenient mental model. But the FargoRate optimization starts completely from scratch every day with all the data. And there is no concept of "these people are in the same league" or "these people are in the same city" or "these people are in the same state." A set of ratings--a complete set of ratings--is found that maximizes the likelihood of all the data. And that's it. There is no inter-area adjustment. There is no concept of area.

Maybe that adjustment is faster in Fargo or maybe the difference in the rate of increase is lower in Fargo (debatable) or maybe an adjustment on a 1000-point scale is less noticeable or meaningful than an adjustment on a 6-point scale, but that wasn't what was stated.

Again, FargoRate doesn't go through some sort of machine learning process for which it starts out with a presumption about two areas and then is informed by new data. There is no starting presumption.

I maintain that the difference does exist, and I maintain that it exists more often in Fargo where your rating is more dependent on who you play than in APA where it's more dependent on how you play.

Here is an example. You can think of WOMEN as like a big league and MEN as like another big league. I just looked at a tournament that went in a couple days ago--the Black Widow Legacy Tournament. The field had 3 females, and they played 107 games against men. Based on ratings, the women were expected to win about 53.6 of the 107 games. They actually won 51. This is of course just one of hundreds of tournaments that have gone in in the last week. And there is league play as well. If women tended to be high or low compared to men, it would tend to show up in these little comparisons that are happening every day that nobody is actually looking at.

But here is the key. These are new games. And it is natural for us to think of the new ratings as "adjustments" of the old ratings updated with this new information. But these games are fully in the mix with the others. It's like a new reality with all the games including these.

1618636501483.png
 

APA Operator

AzB Silver Member
Silver Member
We think of League1 and League2 and then an adjustment between the Leagues just because that's a convenient mental model. But the FargoRate optimization starts completely from scratch every day with all the data. And there is no concept of "these people are in the same league" or "these people are in the same city" or "these people are in the same state." A set of ratings--a complete set of ratings--is found that maximizes the likelihood of all the data. And that's it. There is no inter-area adjustment. There is no concept of area.
You can think of it however you want, one pond, lots of ponds, it doesn't matter. If on a specific day a single match is played somewhere and I have some indirect connection to one of the players in that match, the next day my rating could change without me ever playing a match. That means you have today's guess at my rating and if the rating is lower than it was yesterday, it's been adjusted downward (because I'm connected to the "weaker" side of yesterday's match, even though the system didn't give any thought specifically to that point, it's just in today's data and not in yesterday's data), it can and often is interpreted as you think I was too high, or went up too fast. Anyone connected to the winner is in the winner's "affinity group" (let's remove geography) and anyone connected to the loser is in the loser's affinity group. A player can be in one or the other, or both, or neither.

Again, FargoRate doesn't go through some sort of machine learning process for which it starts out with a presumption about two areas and then is informed by new data. There is no starting presumption.
It doesn't need to. It has yesterday's data and today's data. If ratings change from day to day differences in the data, they have been adjusted. Those are the adjustments about which I speak. No, the system doesn't start with yesterday's ratings and arrive at today's through some magic transformation, but neither does APA. It just appears that way. That's what everyone sees.

Here is an example. You can think of WOMEN as like a big league and MEN as like another big league. I just looked at a tournament that went in a couple days ago--the Black Widow Legacy Tournament. The field had 3 females, and they played 107 games against men. Based on ratings, the women were expected to win about 53.6 of the 107 games. They actually won 51. This is of course just one of hundreds of tournaments that have gone in in the last week. And there is league play as well. If women tended to be high or low compared to men, it would tend to show up in these little comparisons that are happening every day that nobody is actually looking at.

But here is the key. These are new games. And it is natural for us to think of the new ratings as "adjustments" of the old ratings updated with this new information. But these games are fully in the mix with the others. It's like a new reality with all the games including these.
This is exactly what I'm talking about. If you define the affinity group as men and women and look at it that way, the little comparisons (though not done explicitly by the system) will show the difference, and anyone looking at those affinity groups will see the ratings being "adjusted" because of it. Whether the system defines the affinity group or the observer does is immaterial - indirectly, without any thought by the system of those groups, it has been done.

Another way of looking at it is that Fargo has redefined the problem so geographic affinity groups do not effectively exist. The original statement made the assumption that 'ponds' exist, and I said whenever they exist, the behavior described is true of any system. You said it doesn't happen in FargoRate, at least 99% of the time. All you really said is that the ponds don't effectively exist, meaning the behavior isn't geographically observable. You also said that when they do exist, the behavior exists. If FargoRate had existed worldwide before it was practical to have a centralized database, that might not have been true. It might have been very much harder to take geography out of the equation. That's the problem APA has now, but is making progress toward. In 2011, APA went from a database that was totally distributed geographically to one that is now centralized. However, some of its data is still virtually distributed in that centralized database as the paradigm shift was too big to appease everyone. As a result, geographic differences sometimes exist, although they're not as frequent or as observable as they once were. Eventually, they will get geographic differences eliminated, but with a system that big that's been around as long as they have, it takes time.

In the end, the two systems do reflect different things and both do a good job at what they're intended to do. An APA rating is intended to reflect one's skill set on a small scale used to determine races for a specific game. A Fargo rating is intended to put everyone in a relative order. The guy at the top might change in Fargo daily regardless of skillset, whereas in APA the focus is on the groups and who should or should not be in those groups based on their skillset. Both are based primarily on measurable data.
 

mikepage

AzB Silver Member
Silver Member
You can think of it however you want, one pond, lots of ponds, it doesn't matter. If on a specific day a single match is played somewhere and I have some indirect connection to one of the players in that match, the next day my rating could change without me ever playing a match. That means you have today's guess at my rating and if the rating is lower than it was yesterday, it's been adjusted downward (because I'm connected to the "weaker" side of yesterday's match, even though the system didn't give any thought specifically to that point, it's just in today's data and not in yesterday's data), it can and often is interpreted as you think I was too high, or went up too fast. Anyone connected to the winner is in the winner's "affinity group" (let's remove geography) and anyone connected to the loser is in the loser's affinity group. A player can be in one or the other, or both, or neither.


It doesn't need to. It has yesterday's data and today's data. If ratings change from day to day differences in the data, they have been adjusted. Those are the adjustments about which I speak. No, the system doesn't start with yesterday's ratings and arrive at today's through some magic transformation, but neither does APA. It just appears that way. That's what everyone sees.


This is exactly what I'm talking about. If you define the affinity group as men and women and look at it that way, the little comparisons (though not done explicitly by the system) will show the difference, and anyone looking at those affinity groups will see the ratings being "adjusted" because of it. Whether the system defines the affinity group or the observer does is immaterial - indirectly, without any thought by the system of those groups, it has been done.

Another way of looking at it is that Fargo has redefined the problem so geographic affinity groups do not effectively exist. The original statement made the assumption that 'ponds' exist, and I said whenever they exist, the behavior described is true of any system. You said it doesn't happen in FargoRate, at least 99% of the time. All you really said is that the ponds don't effectively exist, meaning the behavior isn't geographically observable. You also said that when they do exist, the behavior exists. If FargoRate had existed worldwide before it was practical to have a centralized database, that might not have been true. It might have been very much harder to take geography out of the equation. That's the problem APA has now, but is making progress toward. In 2011, APA went from a database that was totally distributed geographically to one that is now centralized. However, some of its data is still virtually distributed in that centralized database as the paradigm shift was too big to appease everyone. As a result, geographic differences sometimes exist, although they're not as frequent or as observable as they once were. Eventually, they will get geographic differences eliminated, but with a system that big that's been around as long as they have, it takes time.

In the end, the two systems do reflect different things and both do a good job at what they're intended to do. An APA rating is intended to reflect one's skill set on a small scale used to determine races for a specific game. A Fargo rating is intended to put everyone in a relative order. The guy at the top might change in Fargo daily regardless of skillset, whereas in APA the focus is on the groups and who should or should not be in those groups based on their skillset. Both are based primarily on measurable data.

I don't think we have major areas of disagreement.

Note that with an ELO-type system like FIDE uses for chess, geographic parity is a slow and clunky process even with a centralized database. With FargoRate, the equilibration is complete and done on a daily basis.

FargoRate orders players by skill and also tells how separated they are by skill, all with no reference to absolute skill. Once the database is as large as it is now, though, there is a certain inertia and each rating can be associated with an absolute skill level or proficiency at any particular task. That is, there exists a Fargo Rating that is 50% to run 4 balls in rotation with ball in hand on a particular table or beat the 8-ball ghost on a Valley, etc.

All ratings are best estimates given everything currently known. So your rating today may depend on the games you won and lost against, say, 50 different opponents. As the system gets to know those opponents better (some it may not have known at all), it knows you better because it can now better interpret your games. So if you see your rating change without you playing, that's because the system has more information about you now and can make a better estimate.
 

APA Operator

AzB Silver Member
Silver Member
So if you see your rating change without you playing, that's because the system has more information about you now and can make a better estimate.
Agreed. More data in the rating window usually means a more accurate estimate. There exists the possibility, though, that yesterday's rankings are more "accurate" than today's (I quote accurate because it depends on what those rankings are intended to reflect), based on outside influences not related to skill. Say, for example, today my allergies are acting up and my ability to aim is affected. As Bill Parcell says though, you are what your record says you are. Fargo might have a different result for you tomorrow based on my match being included, but any difference caused by just that match probably won't change any of the races you play tomorrow, at least not when the system has plenty of data on both of us.
 

orion21

AzB Silver Member
Silver Member
Any player can be manually locked in at a 7 never to move again. We found out that one of our 7's was "locked in" after many losses, simply because he had been in APA for so long. The makeup of the local players changed and there was no way he should have been a 7 relative to other 7's.
 

lorider

AzB Silver Member
Silver Member
Gotta love an apa thread ...they go on and on and on lol.

I am going to use a pro player as an anology to an apa 7...in 8 ball that is. An apa 7 is the top tier in apa leagues much like a pro is top tier in the game of pool.

Just because you can not beat other 7's does not mean you should not be a 7...if you are capable of playing at that level.

Lets use robb saez as an example. He is a known pro player. Can he hang with the likes of svb....dennis orcullo...efren etc etc. ? Hell no ! So ask yourself this ..should he not be a pro since there is a long list of pros he cannot beat.

Basically what this thread is about if you think on it.
 
Top