Fargorate tourney performance and group avg questions

[iusedtoberich]

@mikepage

1) When you have a "tournament specific" or "time period specific" rating for a player, do you assume all of their opponent's ratings are static throughout that calculation, and only the player in question has the rating change? Or, something else? Does Digital Pool do it the same way when looking at a Tournament Specific Performance?

2) Sometimes you will take the fargorate average of a group of players, and compare it to a single player. Such as your quote here:

Here is her record against world-top-100 types. These opponents average around 810, so winning a third of the games is about what is expected of a 710. That's about where she's at here, winning 54 games and losing 105 [specific performance rating of 717 for these 159 games].

Can an average actually work in this case, because the scale is logarithmic? I've been confused about that. Does it only work because it's a large population of players? For contrast, if I'm trying to match up a 2 vs 2 partners game, my thinking is we can't simply take the avg of each pair and then put the difference in the odds calculator. Or, if I'm playing me single vs a partner's pair, their average can't be compared to me as a single. Would you expand on these thoughts and let us know the correct math in these scenerios?

Thanks.

 

[mikepage]

@mikepage

1) When you have a "tournament specific" or "time period specific" rating for a player, do you assume all of their opponent's ratings are static throughout that calculation, and only the player in question has the rating change? Or, something else? Does Digital Pool do it the same way when looking at a Tournament Specific Performance?

We--FargoRate-- do something else when we are estimating performance for a small number of games, like one or a few tournaments. Turns out it is pretty complicated. Need to consider likelihood opponent is playing better or worse than his rating and likelihood opponent's rating is running high or low. When there are several opponents, need to consider how coupled are the opponents' ratings. For example is you have multiple opponents from the same league or same small country, errors in the ratings are more correlated than usual. This gets pretty involved.

The easy thing to do is--as you suggest-- is assume the resistance provided by each opponent is exactly at their rating (their rating is right and nobody is getting the rolls). The result is what we call the Specific Performance Rating. It's simple and easy to do and becomes meaningful when you have lots of matches against lots of opponents (cancellation of errors). That's what you see in digitalpool.

Of course when Pia beats Josh 6-5 she's not really playing at 875 speed or something. It's probably a combination of her performing a little better than usual, him performing a little worse than usual, and her getting lucky.

2) Sometimes you will take the fargorate average of a group of players, and compare it to a single player. Such as your quote here:

Here is her record against world-top-100 types. These opponents average around 810, so winning a third of the games is about what is expected of a 710. That's about where she's at here, winning 54 games and losing 105 [specific performance rating of 717 for these 159 games].

Can an average actually work in this case, because the scale is logarithmic? I've been confused about that. Does it only work because it's a large population of players? For contrast, if I'm trying to match up a 2 vs 2 partners game, my thinking is we can't simply take the avg of each pair and then put the difference in the odds calculator. Or, if I'm playing me single vs a partner's pair, their average can't be compared to me as a single. Would you expand on these thoughts and let us know the correct math in these scenerios?

An average works fine. If you are a 600 and you play 20 games against another 600 and 20 games against an 800, you are expected to win 10 of the first set and 4 of the second set. This (14 wins) is pretty much the number of games you are expected to win if you play 40 games against a 700. The averages I use are weighted by the number of games against each person.

For partners, I don't really know. Their strength is going to depend on whether it is regular doubles or scotch and whether they may communicate. When comparing one scotch team to another, it seems to work pretty well. A 420 and 580 seem to perform pretty competitively with 2 500's.

 

[Justaneng]

For partners, I don't really know. Their strength is going to depend on whether it is regular doubles or scotch and whether they may communicate. When comparing one scotch team to another, it seems to work pretty well. A 420 and 580 seem to perform pretty competitively with 2 500's.

As a thought experiment how do we think this works at the extremes? Say SVB and a Fargo 170 vs. 2 Fargo 500's.

 

[iusedtoberich]

As a thought experiment how do we think this works at the extremes? Say SVB and a Fargo 170 vs. 2 Fargo 500's.

Alternate shot I’d take the 500’s. Alternate turn I’d take Shane.

 

[Justaneng]

Alternate shot I’d take the 500’s. Alternate turn I’d take Shane.

Same, I was trying to think through the specifics durnig this "enlightening" e-lecture I just sat through.

Scotch doubles each member takes 50% of the shots so the impact of the lesser player can't be avoided. In this case since we're talking about a sub-200 player, even if Shane sets them up for a straight in 2-ft stop shot it's still not a gimme.

In regular doubles, the shot distribution will gear more heavily to the better player - in this case Shane is going to take 95% of the shots that team SVB/170 takes, minimizing the impact of the lesser player.

 

[iusedtoberich]

Nah I’ll take Shane and the 170 against 2 500’s scotch only though with walk by coaching. My woman(3) and me(7) usually roast the 5x5 teams when we play scotch doubles tournaments. Given they aren’t 500’s. Probably 450ish. Thats in 8 ball. The whole point is Shane will make that table simple for his partner where all the partner will have to do is shoot easy shots and automatically leave Shane his next shot. I tell Kim when we play just hit the ball in because I set up two balls ahead when we play together. The walk by coaching is needed though for when a mistake happens.

The 5x5 teams like to hook each other a lot.

A 170 player can't make a bridge, and is likely to miscue on a straight in shot.

 

[Willowbrook Wolfy]

I don’t know I think your reference for Shane and a low would be pushing it a bit. But I’ll take Shane and a 250 over 2-550’s in a scotch doubles match. Shane and a 350 vs 2-600’s I’ll take the 600’s to win.