How to calculate your Fargo performance in a particular tournament?

[JC]

You're missing the point. It means something to the player as he can get some feedback on how he performed in the tournament. Like you said, his overall FR isn't going to change much, but that doesn't mean he can't get verification of how he played that tourney. This is especially true, if you're like me and you would often times rather play well and lose instead of play poorly and win.

I understand the point. You have the common knowledge of your opponent's fargo and we all know when we're playing well or playing poorly compared to our expectations.

The question was how can I calculate my fargo for a given event and the answer is you can't really do so accurately. Fargo isn't designed to do that. You can use various methods to come up with a number I guess but what does it really mean based on wins and losses only as fargo is, if you know in your heart you played poorly yet won? Or shot strong and lost? I watched Alex lose a match 11-0 to another Filipino and he never made a ball in the entire match. What was his fargo for that event? Pool is a funny game when you take a small shapshot in time.

 

[Black-Balled]

Well the dumb ass says watch this video but refused to provide info or link. That is fooling stupid.

Real helpful, maybe they have some sort of asshat medal for service above and beyond.

You both deserve that award, asshat.

The only days you haven't astonished me with your ignorance are those which I haven't visited Azb.

 

[skip100]

I understand the point. You have the common knowledge of your opponent's fargo and we all know when we're playing well or playing poorly compared to our expectations.

The question was how can I calculate my fargo for a given event and the answer is you can't really do so accurately. Fargo isn't designed to do that. You can use various methods to come up with a number I guess but what does it really mean based on wins and losses only as fargo is, if you know in your heart you played poorly yet won? Or shot strong and lost? I watched Alex lose a match 11-0 to another Filipino and he never made a ball in the entire match. What was his fargo for that event? Pool is a funny game when you take a small shapshot in time.

That's just not true. At least in non-pathological situations like a single 11-0 loss, you can calculate your rating down to the decimal point for that particular event.

Now does it reflect your overall ability in a broader sense? Perhaps yes, perhaps no. That doesn't make it useless or wrong information.

 

[gerryf]

It looks like Fargo treats 8, 9 and 10 ball played on either 7 or 9 foot table as the same, that is, a game. I don't believe any extra weight is given to 9 ft vs 7 ft ... I hope I am wrong, but that seems to be the case.

It's just games played between Player X and Player Y.
... playing either 8 ball, 9 ball, or 10 ball.
... on either 7 ft, 8 ft, 9 ft, or 10 ft tables.
... with either winner breaks or loser breaks or alternate breaks
... with 1 on the spot, or 9 on the spot
... with or without jump cues
... with or without 3-point rule
... with or without 3-foul rule.

 

[JC]

It's just games played between Player X and Player Y.
... playing either 8 ball, 9 ball, or 10 ball.
... on either 7 ft, 8 ft, 9 ft, or 10 ft tables.
... with either winner breaks or loser breaks or alternate breaks
... with 1 on the spot, or 9 on the spot
... with or without jump cues
... with or without 3-point rule
... with or without 3-foul rule.

Imagine the nightmare if you started trying to handicap the equipment and the rules?

As long as ball spots aren't been given it makes no difference what the equipment is they both have to deal with it using pool skills to do so.

Equipment evens out over time just like the rolls.

If you can't see the logic of this in your mind's eye you probably aren't ready to be in this conversation.

 

[ipoppa33]

CocoboloCowboy said:
You are ignoring content by this member.
Well the dumb ass says watch this video but refused to provide info or link. That is fooling stupid.

Real helpful, maybe they have some sort of asshat medal for service above and beyond.

You both deserve that award, asshat.

The only days you haven't astonished me with your ignorance are those which I haven't visited Azb.

 

[Black-Balled]

CocoboloCowboy said:
You are ignoring content by this member.
Well the dumb ass says watch this video but refused to provide info or link. That is fooling stupid.

Real helpful, maybe they have some sort of asshat medal for service above and beyond.

You both deserve that award, asshat.

Lol. I feel bad sometimes...

Then I give my sister her blouse back.

 

[easy-e]

Well the dumb ass says watch this video but refused to provide info or link. That is fooling stupid.

Real helpful, maybe they have some sort of asshat medal for service above and beyond.

You both deserve that award, asshat.

HAHAHA, I told you about the video. If you cared, it would take you two minutes to find it. But we already know that you'll watch it and say something like "well fargo still dumb name for sytssem bout pool. How can dumb name of citty no how good player makes ball? Dont like tough tacos. pool suported by amachurs who by booz, not pool player"

 

[skip100]

Lol. I feel bad sometimes...

Then I give my sister her blouse back.

All he does is talk sh*t but then he can't handle the slightest criticism.

 

[Black-Balled]

HAHAHA, I told you about the video. If you cared, it would take you two minutes to find it. But we already know that you'll watch it and say something like "well fargo still dumb name for sytssem bout pool. How can dumb name of citty no how good player makes ball? Dont like tough tacos. pool suported by amachurs who by booz, not pool player"

Correction:
Tought tacos.

 

[Bob Jewett]

If you can't find the spreadsheet this can be coded up from scratch. We can imagine something like...

#Match 1 - 4-2 against a 600
#Match 2 - 1-4 against a 700
#Match 3 - 3-4 against a 680

The win probability for a single game is roughly 1 / (1 + exp(-(A-B) / 144)), where A is your rating and B is the opponents' rating.

Let's assume you start with a guess of 650. This rating suggests you would win 3.52 out of 6 games in match 1, 2.07 out of 5 games in match 2, and 3.14 out of 7 games in match 3. Adding all these up shows 8.72 games, which is higher than your actual win total of 8.

The optimization process involves iteratively testing out lower ratings until you find a match with the actual win total (or higher ratings, if your initial guess was too low). Here, the answer turns out to be 626.

Thanks for the explanation. It's pretty easy to put that into a spreadsheet if anyone wants to try on their own. If you note how much closer you got with the first guess, it's pretty easy to make a very good second guess.

 

[mikepage]

so anyway, to actually answer the question:

1 - Add up your number of wins at the event
2 - Make an initial guess at a rating
3 - For each opponent, multiply the single-game win probability based on your guess by the number of games you played against them. Add the results for each opponent to get an expected total tournament win count based on your current guess
4 - Adjust guess and repeat 3 until you get a close match

Nice!

A simple spreadsheet that looks like this below can be found under "files" at the "Oklahoma Poolplayers" facebook group. It's called pr.xls.

We do it in a non-iterative way. We first compute a single effective opponent rating--an average opponent rating weighted by the number of games against each opponent. So here you played 18 games, one each against these opponents

600 600 600 600 600 600 700 700 700 700 700 680 680 680 680 680 680 680

Those average to 659. So you won 8 out of 18 (44%) against a 659 pseudo opponent

44% is the expectation for a player 332 *log(0.44/(1-0.44)) away from his opponent. That computes to -32 points

 

[skip100]

Math is cool, yo

 

[Bob Jewett]

... A simple spreadsheet that looks like this below can be found under "files" at the "Oklahoma Poolplayers" facebook group. It's called pr.xls.

We do it in a non-iterative way. We first compute a single effective opponent rating--an average opponent rating weighted by the number of games against each opponent. So here you played 18 games, one each against these opponents ...

Nice and simple. It gives a very slightly different answer than the iterative way. Is either more correct?

I suppose for maximum accuracy (which isn't really needed here), you also need to consider the robustness of your opponents.

 

[hang-the-9]

Well my question is how does Fargo, or any handicapping system interpret data from Bar Box V/S Nine Foot table play?

Both venues have different skill set for success.

Plus you normally do not play Straight Pool, One Pocket, or Bank Pool on Bar Box.

It does not matter, after a large amount of games all those other table sizes will average out into it. It's like mixing two colors, at first you can see them separately but after a bit of time, it's just a single new color. Fargo does not track games much past 9 ball 8 ball and 10 ball so the other games don't matter much. But they tend to line up with the Fargo rating. I think in some one pocket tournament they did the Fargo rating to guess the scores of the matches and they were very very close.

At the end of the day, for statistics most of the specialty games and table sizes are within the noise level of the data, not much to move anyone up or down. If some 750 player can beat an 800 on a 13' table playing one handed banks using golf balls, won't matter much in the overall data.

 

[skip100]

Nice and simple. It gives a very slightly different answer than the iterative way. Is either more correct?

I suppose for maximum accuracy (which isn't really needed here), you also need to consider the robustness of your opponents.

Two things...

1) Averaging out can mess with the calculations a bit. The more heavy-tailed the inputs are, the worse it gets:

650 vs. six 650s: expected wins 3
650 vs. four 625s and two 700s: expected wins 3.001
650 vs. four 575s and two 800s: expected wins 3.03
650 vs. four 550s and two 850s: expected wins 3.07

All of those have the same prediction for a 6-game pseudo-player but the true expected win counts are different. Not a huge impact though and in most realistic situations we're talking about very small changes.

2) I also don't know where the 332*log(0.44/(1-0.44)) came from. A small difference there could easily cause a 1-point difference in the predicted rating.

 

[Bob Jewett]

... 2) I also don't know where the 332*log(0.44/(1-0.44)) came from. A small difference there could easily cause a 1-point difference in the predicted rating.

I'm guessing that because log() is used, 332 is 100/log(2), which is 332.1928, more or less.

 

[mikepage]

I'm guessing that because log() is used, 332 is 100/log(2), which is 332.1928, more or less.

Yes.

And the way we're thinking about it now is to take into account the uncertainties.

Suppose Bob is 80 points over Mike, and both have TONS of games in the system.
Bob and Mike play 50 games with a score of 25 to 25

Did Bob underperform by 80 points?
Did Mike overperform by 80 points?
Did Bob underperform by 40 and Mike overperform by 40?

With 50 games we know empirically the standard deviation is about 40 points. That is, if we approximate performance for either one of these players in many 50-game chunks, two thirds of those performance ratings will be within 40 points of their Fargo Rating. The third possibility above (each off by 1 sigma) is more likely than either of them being off by 2 sigma

Now what if Mike plays Skip instead of Bob. Skip has the same rating as Bob but is just barely established with a rating based on only 200 games before today. Now we have to take into account both Skip's day-to-day fluctuations like above and also the fact we are uncertain where Skip is fluctuating FROM. Our knowledge of Skip's home base, his rating, has an uncertainty with a standard deviation of 20 points.

So even though Mike is winning 25 games against both Bob and Skip and those opponents are rated the same, we judge Mike's performance as higher against Bob.

 

[JC]

Nice!

A simple spreadsheet that looks like this below can be found under "files" at the "Oklahoma Poolplayers" facebook group. It's called pr.xls.

We do it in a non-iterative way. We first compute a single effective opponent rating--an average opponent rating weighted by the number of games against each opponent. So here you played 18 games, one each against these opponents

600 600 600 600 600 600 700 700 700 700 700 680 680 680 680 680 680 680

Those average to 659. So you won 8 out of 18 (44%) against a 659 pseudo opponent

44% is the expectation for a player 332 *log(0.44/(1-0.44)) away from his opponent. That computes to -32 points

View attachment 615847

Playing with this spreadsheet and my personal results one thing is crystal clear comparing my fargo to my performance in events outside my area.

Of course I already knew I was sick and tired of and don't care one iota about league night, from where most of my rating comes from.

This confirms it using math.

 

[tomker]

Does anybody know if there is a way to calculate your speed you played at, Fargo wise, in a tournament that has finished? I've always been curious about this. When my Fargo goes up, it only moves a few points (1200 games in system, doesn't move much), but would be curious how I performed for certain individual tournaments.

I made a web page that does this, here:

Fargo - Tom Kerrigan's Home Page

Hope somebody finds it useful!