Can Fargo Ratings be more Accurate with Dr. Dave's TDF Statistic?

[dr_dave]

Cardigan Kid,

Thank you for suggesting this idea. I like the way you think. However, I agree with majority opinion in this thread.

The table difficulty factor (TDF) would be interesting to record with all match results, and it could be used to see how different players do on different equipment; but I agree with most people here that it probably wouldn't have much of an impact on the overall Fargo Ratings.

The top players play better on all equipment. Even in the bar table open events, where there are numerous strong amateur players, the top players in each game always rise above everybody else.

Regards,
Dave

After reading Bob Jewett's article "Fargo Ratings-How They Work" in the October 2015 issue of Billiards Digest, I have a much better grasp on the concept and thought process behind Fargo Ratings.

It's a great article and I recommend it as required reading for anyone wanting a thorough examination of the rating system.

My one question is this:

If Dr. Dave's TDF statistic (table difficulty factor) was added to the ratings equation, could a more accurate rating be made?

I bring this up because we are at a time when tables of all sorts (9 footers,7 footers, 4.5" corners, 4.75" corners, modified gold crowns, diamond bar boxes, etc) are played on all across the world in various leagues and tournaments.

About ten years ago, I recall a small time amateur who bragged about beating Parica in a bar box tournament. Had he played Parica on a full sized table, the odds of him winning would've been slim to none. After reading Mr. Jewett's article, it is understood that if there is an isolated pocket of players (Nome, Alaska was an example) and one player from that group goes to Vegas and plays a match, the results of that match now effect the entire isolated pocket of players because now there is a reference/connection to the rest of the world.

So in a hypothetical situation, if a small time player were to beat a heavy ranking player, but the TDF was recorded and it was well below 1.0 TDF, wouldn't it be more accurate that the numbers wouldn't adjust as much as if said small time player beat the heavy on a 4 1/4" corner modified gold crown?

Going forward, with US Open "bar box" tournaments (as Thorsten described it) becoming more common, should the ever important TDF statistic be valued when it comes to Fargo Ratings?

It would be great to hear thoughts on this.

For more reading on TDF, here is Dr. Dave's thread on it:

http://forums.azbilliards.com/showthread.php?t=324408

 

[iusedtoberich]

...snip...

The volatility of the scores is higher on the Valley table--they fluctuate more.

...snip...

Can you show this data (maybe in the future when there is more if you don't have it now)? That would show a weaker player is more likely to do better than his Fargorating suggests against a stronger player on a 7' table vs a 9' table? I think it makes sense, but thus far its been opinion. If you have data and analysts of that data that strongly supports this, I'd love to see it.

Thank you.

 

[dr_dave]

An AZ user created a TDF calculator that is linked to in Dr. Dave's thread. Upstate Al and I talked and he recommended that if an App was created that could run this calculator, then anyone could just take the three measurements, open the app, plug them in, and record the TDF.

Agreed. It doesn't take much effort to take three simple measurements and plug the numbers into an app; but many people would not want to do this, and not everybody would do it very carefully and accurately.

But for those who do want to record and report table difficulty factor (TDF) info, it is very easy to get with the free TDF app or TDF spreadsheet.

Regards,
Dave

 

[dr_dave]

There are actually two separate legitimate questions here that tend to get conflated.

The first is is whether in a race to 1000 SVB beats Brandon Shuff by a different amount on a Valley Bar Table and on a tight 9' Diamond. I think the answer to this is essentially no, that the final score will be about the same. This doesn't have to be the case, though. All players have relative strengths and weaknesses and one environment might play more to a particular player's strength.

Even if the above--that the table doesn't matter in the long haul--is true, though, Brandon is still way more likely to beat Shane in a tournament match on the Valley table than on the tight Diamond--The volatility of the scores is higher on the Valley table--they fluctuate more. We refer to this as the run-length issue. We understand it, and we have the ability to take it into account. It's the same thing that makes a win in a 10-ball match more meaningful than a win in a 9-ball match of the same length.

We do keep track of the game (9-ball, 10-ball, or 8-ball), and the table size but no other information about the difficulty of the table.

Good post, Mike.

Sounds reasonable to me.

It would be interesting to see some of this sort of equipment-comparison data some day.

Regards,
Dave

 

[BRussell]

There are actually two separate legitimate questions here that tend to get conflated.

The first is is whether in a race to 1000 SVB beats Brandon Shuff by a different amount on a Valley Bar Table and on a tight 9' Diamond. I think the answer to this is essentially no, that the final score will be about the same. This doesn't have to be the case, though. All players have relative strengths and weaknesses and one environment might play more to a particular player's strength.

Even if the above--that the table doesn't matter in the long haul--is true, though, Brandon is still way more likely to beat Shane in a tournament match on the Valley table than on the tight Diamond--The volatility of the scores is higher on the Valley table--they fluctuate more. We refer to this as the run-length issue. We understand it, and we have the ability to take it into account. It's the same thing that makes a win in a 10-ball match more meaningful than a win in a 9-ball match of the same length.


We do keep track of the game (9-ball, 10-ball, or 8-ball), and the table size but no other information about the difficulty of the table.

I don't think it's true that there's a difference between these two bolded parts, or at least I can't think of any reason why there should be. Volatility in the short run should lead to closer scores in the long run. If the weaker player is more likely to win a short race, it's because the probability of the weaker player winning increases, which should lead to a closer match in the long run as well.

 

[mikepage]

Can you show this data (maybe in the future when there is more if you don't have it now)? That would show a weaker player is more likely to do better than his Fargorating suggests against a stronger player on a 7' table vs a 9' table? I think it makes sense, but thus far its been opinion. If you have data and analysts of that data that strongly supports this, I'd love to see it.

Thank you.

Have to be careful here. I don't agree with with the highlighted statement. I would say both players are more likely to deviate from their expectations (in both directions) for a race of the same length on the bar table.

 

[DJ66]

I have been lurking for some time, and reading this post, several things come to mind, but first my disclaimer, I am an average player who has beaten or lost to no one of interest.
First, it seems to me, all for these ratings systems are only mathematical probabilities. One can study these forever and only get a general idea of what should happen, not necessarily what will happen.
Second, if you follow horse racing, how often does the favorite actually win? Pool is a human sport, with human faults, unfortunately. How many times have we seen great players miss seemingly simple shots? Not often, but it does happen, hence human factor affecting mathematical probability.
Third, how much has the table been played on, or has proper maintenance been performed since the cloth was last changed?
And last, I don't know what room owner or players are going to think if someone walks in to a strange room, pulls out a ruler, and starts measuring pocket width and depth. This might alter feelings as far as league or action.
Like I stated at the beginning, I am an average player. I do have a routine I go through when I get to a new table, but when is all of this too much? Please feel free to respond, I do enjoy learning, just realize I am not trying to be an a $$ about this.

 

[Cardigan Kid]

Agreed. It doesn't take much effort to take three simple measurements and plug the numbers into an app; but many people would not want to do this, and not everybody would do it very carefully and accurately.

But for those who do want to record and report table difficulty factor (TDF) info, it is very easy to get with the free TDF app or TDF spreadsheet.

Regards,
Dave

Agreed on the top tier players. They can and do play stellar on any sized equipment. At Large had some great statistics verifying the run outs, balls on the break, etc were similar in the bar box US Open 8 ball, and Accu Stat invitational 8 ball.

However, when we get to the lower tier players, or players just starting out. If player A plays on just valley tables, and player B works with gold crowns, after 100 matches with same level of competition and similar records, is some added weight given to player B? (There are so many other factors in Fargo Ratings, so I'm simplifying things for arguments sake, sorry Mr. Page).

Maybe not in Fargo Ratings, but the TDF can at least offer a little separation.

I always look at the great debate about Mosconi's high run on an 8 foot table with a wider pocket size as the first desperate need for the TDF stat. We could all at least, no matter what side of the divide, settle on the fact that the difficulty level is fluid depending on what the factor is.

So when Arthur "Babe" Cranfield, straight pool run, 420 balls, in Syracuse on a 10' table, before a sizeable audience, per George Fels....When TDF is added, the mosconi 526 on 8 foot table in 1954 now might have competition as greatest high run of all time.

This is just conjecture on my part, because it's all in the past, but going forward, to work around match results and records and avoid all the opinion based banter, the TDF + Fargo Rating would be the ticket to accurate pool predictions.

Maybe we can keep them separate but always in the conversation together. Like saying stripes and solids....Fargo Rating and TDF. :grin-square:

 

[mikepage]

I don't think it's true that there's a difference between these two bolded parts, or at least I can't think of any reason why there should be. Volatility in the short run should lead to closer scores in the long run. If the weaker player is more likely to win a short race, it's because the probability of the weaker player winning increases, which should lead to a closer match in the long run as well.

Let me compare two situations and then bring up a third.

(1) Two similar-speed top pros playing a race to 100 10-Ball

(2) Two similar-speed very weak novices playing a race to 100 straight pool

A "point" of straight pool plays the same role as a "game" here.

For the 10-Ball, you will see scores like
100-90 or 96-100 or 79-100 or 100-87, 74-100 and so forth. But you won't see 100-8 or 12-100.

For the two beginners playing straight pool, the scores will be similar to the scores for the pros playing 10-ball. You will see 100-89, but you won't see 100-13.

Here's the third situation

(3) the similar-speed top pros playing a race to 100 in straight pool

Now you will see very different scores, 100-88, 44-100, 65-100, 100-12, 100-57, 16-100.

The difference between scenarios (2) and (3) are the race to 100 in straight pool for the pros has far fewer innings (changes in control) and innings span many "games." This is the run-length issue.

A race to 13 on a table that has more table runs is effectively a shorter race. That is the source of the volatility I'm talking about.

 

[Cardigan Kid]

I have been lurking for some time, and reading this post, several things come to mind, but first my disclaimer, I am an average player who has beaten or lost to no one of interest.
First, it seems to me, all for these ratings systems are only mathematical probabilities. One can study these forever and only get a general idea of what should happen, not necessarily what will happen.
Second, if you follow horse racing, how often does the favorite actually win? Pool is a human sport, with human faults, unfortunately. How many times have we seen great players miss seemingly simple shots? Not often, but it does happen, hence human factor affecting mathematical probability.
Third, how much has the table been played on, or has proper maintenance been performed since the cloth was last changed?
And last, I don't know what room owner or players are going to think if someone walks in to a strange room, pulls out a ruler, and starts measuring pocket width and depth. This might alter feelings as far as league or action.
Like I stated at the beginning, I am an average player. I do have a routine I go through when I get to a new table, but when is all of this too much? Please feel free to respond, I do enjoy learning, just realize I am not trying to be an a $$ about this.

Thanks for posting to the thread, sir. Your opinion is absolutely valid in that it is a human sport and all sorts of intangibles go into the unpredictable results. That's the beauty of pool and why I love watching players match up because you never know.

The problem I have always found with the sport was lack of standardization. Not just in the equipment (different sized tables with different sized pockets) but also a standard way of ranking players or rating players.

Fargo Ratings purpose will be to bring the sport together. you've heard of the old argument an A player in one area of the country, would be a C player in another area of the country. That whole debate could be put to rest of Fargo Ratings take over and become the global standard.

Then comes the TDF where I believe is important to quantify the results of the play at the table. If you talk to a player who beat the ghost on a diamond nine foot, and another who beat the ghost on a valley bar box, it's not just opinion anymore if you nod your head and see the ghost defeat on the diamond as more impressive-now that the TDF is known.

Opinion can be backed by statistics.

 

[dr_dave]

Agreed on the top tier players. They can and do play stellar on any sized equipment. At Large had some great statistics verifying the run outs, balls on the break, etc were similar in the bar box US Open 8 ball, and Accu Stat invitational 8 ball.

However, when we get to the lower tier players, or players just starting out. If player A plays on just valley tables, and player B works with gold crowns, after 100 matches with same level of competition and similar records, is some added weight given to player B? (There are so many other factors in Fargo Ratings, so I'm simplifying things for arguments sake, sorry Mr. Page).

Maybe not in Fargo Ratings, but the TDF can at least offer a little separation.

I always look at the great debate about Mosconi's high run on an 8 foot table with a wider pocket size as the first desperate need for the TDF stat. We could all at least, no matter what side of the divide, settle on the fact that the difficulty level is fluid depending on what the factor is.

So when Arthur "Babe" Cranfield, straight pool run, 420 balls, in Syracuse on a 10' table, before a sizeable audience, per George Fels....When TDF is added, the mosconi 526 on 8 foot table in 1954 now might have competition as greatest high run of all time.

This is just conjecture on my part, because it's all in the past, but going forward, to work around match results and records and avoid all the opinion based banter, the TDF + Fargo Rating would be the ticket to accurate pool predictions.

Maybe we can keep them separate but always in the conversation together. Like saying stripes and solids....Fargo Rating and TDF. :grin-square:

You make some good points.

And thank you for being so supportive and embracing of the table difficulty factor (TDF) as a useful tool in pool.

Good thread,
Dave

 

[SUPERSTAR]

Maybe someone should add some luckbox/garbage/fate factors to the ratings.
As well as a rules factor and a table size factor.

Sometimes, people just crap the bed all day long.
Sometimes, rules are different. This tournament is luck 9ball, that tournament is call shot 9ball and safeties.
Most tournaments allow jump cues. Turning stone and the Joss tour does not.
Maybe there is a reason Shaw has won 3 Turning Stones in a row that has to do with lack of jump cues.

Maybe standardized stuff, or make different categories for different games.
Cause to say rotation games, 14.1, 8ball, and 1pocket are all the same game, is just wrong.
To say all those games are the same games on a bar table vs a 9footer is also stupid.
Just like when Earl clearly destroyed SVB on the 10 footer that one time.
But is the total underdog on all other equipment.
It's not the same game. So to mix statistics for all different tables is also wrong.

To mix all that varied info into one statistic is just not well thought out.

 

[JC]

TDF although well meaning still has way too many variables to be the nuts.

Cloth age, condition, type, ball cleanliness, type, humidity, rubber condition, etc, can swing things a great deal on two seemingly identical tables.

Table toughness is like supreme court justice Potter when he discussed pornography. He said he can't describe it but knows it when he sees it.

JC

 

[Poolplaya9]

First, it seems to me, all for these ratings systems are only mathematical probabilities. One can study these forever and only get a general idea of what should happen, not necessarily what will happen.

I'm going to paraphrase the above to perhaps help you see it from a different perspective. All you essentially said is that "probabilities are not certainties". And of course you are right. And Fargo (or any other probability statistic) doesn't try to tell you what is guaranteed, it just tries to tell you what the odds are on something. In the case of Fargo it tells you what player is favored in a match up and by what odds.

If player A is 80% favored to win over player B, and they play a match, it is of course still possible that player B will win that match as you point out. If player A was 80% favored in the match, it is the same thing as saying that if that same match up were played 100 times, player A would win 80 of them, and player B would win 20 of them. It is like if a gum ball machine has 80 red gum balls in it, and 20 blue gum balls. If you put in the your dime and crank the handle chances are (80% chance) that you will get a red one, but of course it is possible that a blue one comes out instead (will happen 20% of the time). A probability of 80% favorite doesn't guarantee what what happen any one time, but it does guarantee what will happen 80% of the time if it is done enough times.

Which brings up one other important point which is that the smaller the sample size, meaning the fewer the times something is done, the more the outcome can vary from what the probability says. What I mean by that is for example, we all know that if you flip a coin, the probability of it being heads is 50%, or 1 in 2. If you flip the coin only four times, all four flips might be tails, which would be 100% tails in the actual results, which is a big variation from the probability. It doesn't change the fact that it there was still a 50% chance that each flip would have been heads. It just means that you didn't flip it enough times for the results to accurately match up with the actual probability. If you flipped it 100 times, you will be much closer to the 50% heads, maybe you got heads 48 times, which is 48%. And if you flip it one million times, you will be even closer to 50% heads and 50% tails. Knowing the odds, even when a match is only being played once, is still very valuable information.

So just because anything can happen in the one match between player A and B doesn't change the fact that player A was still 80% favorite, or would win 80% of the time, or would win about 80 times out of a hundred in that match up. As you pointed out knowing that probability doesn't guarantee a result (nor is it designed to or claimed to), but what it does is tell you extremely accurately what percentage of the time you will get a certain result which is still extremely beneficial information to know.

Second, if you follow horse racing, how often does the favorite actually win? Pool is a human sport, with human faults, unfortunately. How many times have we seen great players miss seemingly simple shots? Not often, but it does happen, hence human factor affecting mathematical probability.

In horse racing a horse is rarely 80% favored to win a race, or even 50% to win a race. They are generally like 1 in 5 or maybe 1 in 3 to win the race, and that is about how often the favorite wins. The fact that the favorite often doesn't win is simply because he was the favorite only in the sense that he had more chance to win than any other single one of the other horses did, not that he had more chance to win that all of the other horses combined (which is probably fairly rarely the case). The reality is that the horses that are the favorite in their race win the race about as often as the mathematical probability predicts they will.

Third, how much has the table been played on, or has proper maintenance been performed since the cloth was last changed?

These things do make a small difference. Two points though. First, each player in the match on that table played on the same table and conditions, so it generally doesn't make much difference to their odds against each other. And second, as far as comparing to other players in other conditions there are literally hundreds maybe thousands of things that are a factor, like how hydrated each player was and how many hours before the match since they last ate a meal. As others have mentioned most of these things make very little difference in the scheme of things (since in a system like Fargo you have a pretty decent sample size) and it is too difficult to try to account for all the hundreds or perhaps thousands of things that can be a factor in a single match. Law of diminished returns comes into play here where you put in 2000% more time and work to get the system 1% more accurate.

And last, I don't know what room owner or players are going to think if someone walks in to a strange room, pulls out a ruler, and starts measuring pocket width and depth. This might alter feelings as far as league or action.

I can see an occasional room owner getting offended that his equipment was being measured for how tough or easy it was. But again, in the long run because you will have a big sample size of games for a player under lots of conditions it will kind of even out and things like this or how long since each player last ate are pretty inconsequential and will offer very little improvement in the system in exchange for the tons and tons more work they would create. And as others have mentioned, you are lucky and doing good if you can even get tournament directors to report all the match scores from an event and many of them aren't going to be willing to go measure all the table factors too, and even if they did they might not do it accurately anyway.

 

[robsnotes4u]

Very nice!

 

[DJ66]

Thank you for the replies

Thank you for the replies. As I stated, it is simply for my own edification, and I really appreciate not being chided for my ignorance. Obviously I have a lot more studying to do.