Probabilities and Pool

scottjen26

AzB Silver Member
Silver Member
Long post, but something I recently did that I found interesting...

I've been in a slump for several months now, with a few flashes of mediocrity in between but nothing too exciting. Decline in practice, lack of competition, etc. all played a part. I took a break over the holidays, and did some thinking about how to attack my first practice session when I got back, what I did differently before vs now, etc. Happy to report that 3 days running now it looks like "I'm back"! Running out much more often, making far fewer mistakes, and the ones I am making are immediately recognizable and correctable, unlike before when stuff was just happening seemingly randomly.

Anyway, because of my slump, the difficulty of this game was really thrown in my face. Especially when you reach a certain level and have certain expectations, you see how minor drops in focus or execution can impact things severely. For instance, at one point I think I went several practice sessions without a single runout. I would make good shots, even series of shots, but there was always that one mistake that would stop the run, and it could come at any point. Frustrating for sure.

So I started thinking about probabilities of running a rack and how consistency impacts it. Hopefully my math and assumptions are correct...


The probability of running a rack of 9 ball should be equal to the each of the invidiual shot probabilities multiplied together. So assuming ghost-type format, if no balls are made on the break you have 9 shots, so 9 individual probabilities multiplied together to determine the overall likely runout percentage. In my estimation, a top pro would be 90 - 99% on most shots to both make the ball and achieve position on the next ball, assuming they stay in line. If we average that out to 95% for each shot, then they would have a 63% chance of running the rack. Again, this is a ghost-style format with BIH to start, not an expected break and run percentage in competition. This seems reasonable, top players would typically beat the 9 ball ghost in a race to 9 unless the balls were breaking poorly, their chances would be even better because they are making 1 or 2 balls on a lot of the breaks. So don't bet against a top player playing the 9-ball ghost!

If we drop the average down to 90%, which might be a good average for an A player, the runout percentage drops to 39%. Again, pretty reasonable, and again these players would typically do well against the ghost (your percentage would only need to be just over 50% to beat the ghost in a race), especially making balls on the break etc. For a B player, whose average might be in the 75% - 80% range, the overall runout percentage drops dramatically to 7% - 12% making beating the ghost very difficult to achieve.

If the same top player above with an average 95% shot percentage gets out of line once, forcing them to shoot a 75% shot (3 out of 4 chance to make and get position), their runout percentage for that rack drops from 63% down to 50%. If that one shot is even more of a flyer, say 50% (which they almost never shoot these days), it drops to 33%. Having three slightly harder shots in the runout, say in the 75% - 80% category, is equivalent to having just one very hard shot, so it shows the merit in playing safe instead of taking that flyer to try and get out.

I calculated a pretty typical rack for myself and came up with a 27% runout percentage. That seems about right, I beat the ghost in a race to 9 a third to half the time, and when I lose it's usually somewhere between 9-5 and 9-7. Again, factoring making balls on the breaks, differences in how the balls spread (some racks are just easier than others), etc.


It's pretty simple to throw together a spreadsheet and start playing with percentages to see what happens. I wasn't really trying to overanalyze things, more than I already do, but I had the thought and I was able to draw some good conclusions from actually seeing the results on paper. First, the game if fricking hard! But we all know that… Mathematically it really shows how improving your consistency on your weaker shots will improve your run out percentage more than further improving your already strong shots. It shows the value in shotmaking or position drills with the intent to make more and more shots routine and repeatable. Reading the rack and pattern recognition in order to keep each invidual shot percentage as high as possible. It reinforces the value in playing safe when faced with a difficult shot, and shows the foolishness in trying to get out by making one spectacular shot after another. More than anything, it shows the extreme importance of the mental game and the need to stay focused on every shot and not take anything for granted.

Scott
 
Pretty reasonable analysis, although you need to be very careful with the math.

Exponentiating the arithmetic mean of shot probablities by the number of shots is not quite correct in determination of the likelihood of a stated number of consecutive successes. In fact, you need to exponentiate the geometric mean by the number of shots.

Example:

Shot 1 Probability: .5
Shot 2 Probability: .6
Shot 3: Probability:.7
Shot 4 Probability: .8
Shot 5 Probability: .9

With your method, the probability of succeeding on all five shots would be 0.7 (the arithmetic mean) to the fifth power, which comes to .168, or 16.8%.

In fact, however, this is the wrong answer as the answer is best computed as .5*.6*.7*.8*.9, which comes to .151 or 15.1%, corresponding to a geometric mean of about .685, or 68.5%.

I think the most important takeaway from your analysis is that unless you are playing really good position, your chances for running out are not very good.

Nice post.
 
The numbers don't lie!

Actually knowing where you stand statistically is very important. You hear a lot players throw out the phrase - "I'm really in a slump." I think for a lot of guys they just don't have a realistic image of where their game is at. They remember that they ran out a few racks last week but they forget all those racks where they didn't run out. Or they make a lot of difficult shots and don't get out because they missed position but all they remember is the great shots they made. It's sort of human nature to have this sort of selective memory. So, when these guys go a while without running out or playing their "best" game they just start saying they are in a slump. It's much easier on the psyche than admitting they really aren't as good as they thought they were.

Personally, I've noticed that I don't really get into slumps much anymore. I guess my game has reached a certain level of mediocrity, and paired with knowing exactly what I'm capable of my game sort of just does what it's supposed to. I have days where I can't focus like I need to and start pounding the balls into the rails. These days I'm usually just too tired to play.

Don't get me wrong though I do believe you can still get into a slump or a funk and I'm not questioning you on that. The numbers you have broken down are really interesting and give me some food for thought.

Good luck with waking your game back up.
 
9541142.jpg
 
Pretty reasonable analysis, although you need to be very careful with the math.

Exponentiating the arithmetic mean of shot probablities by the number of shots is not quite correct in determination of the likelihood of a stated number of consecutive successes. In fact, you need to exponentiate the geometric mean by the number of shots.

Example:

Shot 1 Probability: .5
Shot 2 Probability: .6
Shot 3: Probability:.7
Shot 4 Probability: .8
Shot 5 Probability: .9

With your method, the probability of succeeding on all five shots would be 0.7 (the arithmetic mean) to the fifth power, which comes to .168, or 16.8%.

In fact, however, this is the wrong answer as the answer is best computed as .5*.6*.7*.8*.9, which comes to .151 or 15.1%, corresponding to a geometric mean of about .685, or 68.5%.

I think the most important takeaway from your analysis is that unless you are playing really good position, your chances for running out are not very good.

Nice post.

I did the math as you stated in your second example (the correct one), multiplying the individual percentages for each shot together. I didn't take an average and then multiply, that would only be close if the shot percentages were well distributed. I only used 95%, or 90%, as a good starting average on a per shot basis, but then assigned other percentages to each individual shot as needed.

So in a typical decent rack, it might look like .90 * .90 * .80 * .70 * .75 * .80 * .90 * .90 * .95, or about 21% total runout percentage for that shooter for that layout.

While I studied math extensively in school, I haven't done formal mathematics of any sort for quite some time. This was really just a simple analysis based on my memory that the probability of multiple events is equal to the product of the invidivual events. Just like flipping coins, probability of getting heads twice is 1/2 * 1/2 or 1/4. The premise seemed to make sense, and if so it certainly does show just how impressive the top players in the game really are with regards to their analysis, precision, and concentration. One of the toughest games/sports out there for sure.

Thanks for the feedback!

Scott
 
Last edited:
Tuff action

The numbers dont lie...., this is the tuffest game in the world...., I golf and bowl at a good speed...., but havent gotten to same level at pool yet..., tuff action!
 
Great post

... I love this post... can we use this to answer some questions?

who has the advantage - shot maker or position player?

Obviously the most deadly are a combination ( Lassiter / or Hall, etc)

Great post...
 
There is no way to properly put it down in numbers for the total chance of a runout from starting position. If anywhere along the way of shooting, you take the wrong shot, or go for the wrong leave, you will hurt your chance of running out the table, even if you are a good shot. Can have a table that could be wide open, all shots at 95% of being made and getting perfect position, but if you attempt to go for something else, and put slightly the wrong english then suddenly shots could be 30% chance of make and position. Choosing the correct run is more crucial than what the difficulty of the shots could be.
 
There is no way to properly put it down in numbers for the total chance of a runout from starting position. If anywhere along the way of shooting, you take the wrong shot, or go for the wrong leave, you will hurt your chance of running out the table, even if you are a good shot. Can have a table that could be wide open, all shots at 95% of being made and getting perfect position, but if you attempt to go for something else, and put slightly the wrong english then suddenly shots could be 30% chance of make and position. Choosing the correct run is more crucial than what the difficulty of the shots could be.

This is really well said. Differing conceptualization skills will cause players of equal pocketing ability and identical storkes to have very different levels of success in running the table.
 
Slump

Building a great pre shot routine can definitely help with slumps
you would be surprised what you can forget to do when everything is going good and then how quick you correct in the wrong areas when its not
 
He DID do it the way you're saying is correct. He multiplied the individual probabilities of each success to compute the probability of succeeding everytime.

Pretty reasonable analysis, although you need to be very careful with the math.

Exponentiating the arithmetic mean of shot probablities by the number of shots is not quite correct in determination of the likelihood of a stated number of consecutive successes. In fact, you need to exponentiate the geometric mean by the number of shots.

Example:

Shot 1 Probability: .5
Shot 2 Probability: .6
Shot 3: Probability:.7
Shot 4 Probability: .8
Shot 5 Probability: .9

With your method, the probability of succeeding on all five shots would be 0.7 (the arithmetic mean) to the fifth power, which comes to .168, or 16.8%.

In fact, however, this is the wrong answer as the answer is best computed as .5*.6*.7*.8*.9, which comes to .151 or 15.1%, corresponding to a geometric mean of about .685, or 68.5%.

I think the most important takeaway from your analysis is that unless you are playing really good position, your chances for running out are not very good.

Nice post.
 
Last edited:
Spending time playing the ghost is as we all know one of the most entertaining forms of solitaire and the measurement is valid for part of our game however it does little to predict your chances of winning a given game/set. I have dedicated a couple hours a week for the last year to practicing only safeties that recur frequently in rotation games. I can't shoot any straighter than I could 35 years ago but this one discipline has helped me win much more often than before this light bulb came on in my stubborn head. First you have to recognize when to hold em and when to fold em based on an honest evaluation of your current level of play and then you have to have the skill set necessary to execute it. And if you don't practice it often you won't do it under pressure. It's amazing the games you can win by smart "grinding" without being able to aim any better or run out any more often. Once you start having success like this it becomes addicting. People you play frequently know you're playing better but they can't quite figure out what changed. They think you're shooting straighter but all you've really changed is your decisions.
 
He DID do it the way you're saying is correct. He multiplied the individual probabilities of each success to compute the probability of succeeding everytime.

Not so. His post includes the following,

... If we average that out to 95% for each shot, then they would have a 63% chance of running the rack...

This statement shows that the artimetic mean, generally referred to as the average is what is being exponentiated, and this is in error. Here's an example of why.


Shot Player 1 Player 2
Number Prob Prob
Shot 1 0.91 0.95
Shot 2 0.92 0.95
Shot 3 0.93 0.95
Shot 4 0.94 0.95
Shot 5 0.95 0.95
Shot 6 0.96 0.95
Shot 7 0.97 0.95
Shot 8 0.98 0.95
Shot 9 0.99 0.95

Runout 62.8% 63.0%

Each player has an average success rate of .95 on their shots, but their runout percentages are different. That’s because you can't merely exponentiate the average success rate to get to the runout probability.
 
Last edited:
First you have to recognize when to hold em and when to fold em based on an honest evaluation of your current level of play and then you have to have the skill set necessary to execute it.

And if you are asking other players to evaluate what your current skill level is, you may be told things you don't want to hear, but don't dismiss them, you will at least know what is holding you back.
 
I once ask an old timer when I should play safe an he told me when there was less than 90 per cent chance of making a shot i. He instinctively knew what he was talking about. If you run into a couple of less than 90 percent shots your chances of running out are greatly diminished. Thus a good look at the table as it lies before starting is a must.

Good thread. Gives me things to consider.
 
... That’s because you can't merely exponentiate the average success rate to get to the runout probability.

Correct. And if the numbers vary a lot more from the mean, the results will show a greater difference than in your example.

Ex.: The mean of .99, .50, and .01 is .50. But the cube of .50 is over 25 times as great as the product of .99, .50, and .01.

[Edit: not that the numbers in my example would really be illustrative of shot probabilities for two different players.]
 
Last edited:
I once ask an old timer when I should play safe an he told me when there was less than 90 per cent chance of making a shot ...

And did you discuss what to do if the safety is at least as difficult as the shot?
 
Back
Top