I think he doesn't understand statistics at all.
But you don't have to run Monte Carlo simulations if you believe the theory. The theory says that as long as neither player is allowed more than N breaks in a game to N, the format doesn't matter.
which favors the higher-rated player, winner break or alt. break?
- Thread starter evergruven
- Start date
But you don't have to run Monte Carlo simulations if you believe the theory. The theory says that as long as neither player is allowed more than N breaks in a game to N, the format doesn't matter.
I mention this and everyone hates the idea. I was interviewing Allen Hopkins and we got on this topic. His suggestion was the first time I heard this idea.
LOSER BREAKS
Allen said, "Can you imagine a football team scores a touchdown and then they immediately get the ball back?" His example turned the light on me. People say stuff like, "I like a close match where it gets down to hill-hill" Really?
In one pocket it is always alternating breaks because the break averages 1.5 ball advantage to the breaker. That game should stay as it is. But all other games (9-ball, 10-ball, 8-ball, etc) should be done in loser breaks format.
Best of rolls,
Ken
LOSER BREAKS
Allen said, "Can you imagine a football team scores a touchdown and then they immediately get the ball back?" His example turned the light on me. People say stuff like, "I like a close match where it gets down to hill-hill" Really?
In one pocket it is always alternating breaks because the break averages 1.5 ball advantage to the breaker. That game should stay as it is. But all other games (9-ball, 10-ball, 8-ball, etc) should be done in loser breaks format.
Best of rolls,
Ken
Perhaps I'm not understanding the issue here. Can one of you elaborate?
Statics = extrapolation of reality
I prefer realtime reality.
lol... what flaw....? sample size of player performance on a micro case by case basis isn't flaw. It's as accurate as it gets. Not stretched out to the point of ignoring details.
Statistics is a funny tool. Its good at keeping history and tracking change. Speculators are out of control using statistics concepts with wild abandonment.
At the local level there are not enough qualified pool players to fill a 256 field. Unless you are in a high density and high pool population area. Maybe 8 cities have enough pool players where it would take awhile to rate and rank them.
That is what speculators do, they attempt to order high variability data. Predicting randomness is mathematically incoherent and illogical.
Back to the story locally once you hit threshold practice time and proficiency then you become the constant not the variable.
A room has its regulars and the order holds until players make breakthroughs.
To predict the future other factors have to be studied. If your interested in analysis, post and make requests.
At the local level there are not enough qualified pool players to fill a 256 field. Unless you are in a high density and high pool population area. Maybe 8 cities have enough pool players where it would take awhile to rate and rank them.
That is what speculators do, they attempt to order high variability data. Predicting randomness is mathematically incoherent and illogical.
Back to the story locally once you hit threshold practice time and proficiency then you become the constant not the variable.
A room has its regulars and the order holds until players make breakthroughs.
To predict the future other factors have to be studied. If your interested in analysis, post and make requests.
This was my thought too. And then also Turning Stone has a fairly low break and run percentage compared to other major events. According to Atlarges stats for the recent Turning Stone, there was a 22% break and run percentage which is lower than I usually see for winner breaks events. This may not be enough to see a huge difference in scores since players are likely getting roughly similar opportunities as they might in an alternating break event.
There are two things I’d be interested in seeing.
- How does this outcome change in a tournament with looser break rules and a table that is more productive on the break. It’s possible that the pros at turning stone are not running off the break enough to substantially take advantage of the rule set.
- What differences do we see if we focus on rating differences of 25, 50 and 100 points. The 200 point gap is so large that I don’t think it matters what the format is. They make too many mistakes to truly benefit from the alternate breaks format. But what about an 830 vs. 800 player?
Looking at individual game results is good and will be a more accurate than match wins alone assuming your model is good.
Having the break is an advantage to winning the rack. Alternate break formats will result in the weaker player having more chances to break than winner break. Thus, the best model factors in knowledge of who broke the balls. In tournaments with an alternate break format, the ratings produced would be more tightly grouped than those with winner break formats.
True. If the psychological factors affect both players equally (in proportion to their skill level), then the effects simply cancel out. But many times that is not the case. Psychology may be precisely what differentiates an overall 'stronger' player from an overall 'weaker' player.
For example, let's say Player A and Player B have the same skill level/rating in an absolute vacuum absent any psychological factors. Player A is effectively a robot and maintains his speed regardless of the situation and score. However, Player B is more emotional and tends to fold during high pressure situations, effectively playing at a lower speed late in the match when the score is tied or very close.
Given the player models above, the probability that Player B beats Player A is actually higher in a winner break format compared to alternating break. This is simply because in winner break Player B is more likely (compared to alternating break) to jump out to a decent lead early on in the match, such that B is more likely to maintain play at his higher rating throughout the entire match. To be clear, Player B still won't be the favorite in either format, but his chances of beating Player A does increase with winner break (again, given the player models I stated previously).
I don't have any hard data to point to. Though, the final four of the latest Spanish Open (Bjisterbosch, Labutis, Woodward, Dang) shows that the cream doesn't always rise to the very top with the Matchroom format (though, I would include Sky as part of the 'cream'). And that's not a bad thing!
You could use my simulations to get a feel of this. It doesn't use Fargo rates, but you can vary the odds that player A and player B win a rack, and lower their chances of break & win rack to reflect tougher equipment.
If I do this, keeping the gap between players constant, here are the things I notice:
- winner break, loser break, alternate break -- still don't matter as to how likely the better player is to win a match
- the easier the equipment is, the more likely the better player is to win, although the difference isn't very big*
- bigger rating differential between players doesn't change the above, except that the better player is more likely to win with a better rating differential (obviously)
That’s interesting. Though a few things come to mind,
- For equipment difficulty, As you say, different players aren’t equally affected by equipment. Looking the stats between the UK and World Championships this year, the pros maintained a similar break and run percentage despite the UK Open having tighter pockets. So we may need to account for the fact that at a certain threshold, tighter tables don’t affect performance. But at what skill and below do we see it start playing more of role?
- How does a players likelihood of winning the next rack change after sitting out an opponents package? For example, in winner breaks a 2 pack encompasses 3 consecutive rack wins. So to what extent would this variable influence the outcome?
- If I’m right, then the likelihood of the weaker player winning the rack would theoretically change between match formats depending on the likelihood of either player stringing a two pack or better.
- And finally, in a winner break format where a player needs to break serve after losing a game, should it change the likelihood of winning any given rack afterwards if you only have a 34% chance of winning any given game thereafter until you do manage break serve? For example, assuming you lose the lag,
- in alternating breaks, you have a 34% chance of winning game 1. 66% chance of winning game 2, 34% chance of winning game 3 etc.
- In winner breaks, you have a 34% chance of winning game 1, 2 and 3 assuming you lose each game.
Last edited:
So far as I know, there is no sound theory that includes the human variables. It can be fun to speculate, though.
There was a player around here who seemed to have a specific psychological quirk. When he felt he did not have a fair chance, he would play worse. Or at least that's the way it seemed to me, having seen him turn from Godzilla to Goofy multiple times. The "no fair chance" might be when he felt he was being cheated or hustled or the gods were just against him on the rolls. In order to work that quirk into a statistical analysis, you would have to figure out how likely it would be to get triggered. I suppose I should have tried to figure out how to trigger the change, but I'd rather just play.
tomatoshooter
Well-known member
This was comparing players with a 200 point difference, I do have to wonder if that difference is so great that no break format would make a difference. Then again, normal performance variation is said to be about +/- 50 Fargos so a smaller difference in player strength may be hard to discern from the noise. There may also be slightly different results if the matches were 600s vs 400s. The mid 700 players are running racks reliably, and the mid 500 players are not running many racks.
For your scenario of some of the players getting to the hill, and some getting blown out, if you are going to have a sample large enough to draw useful conclusions, you need to lump them together. Since none of the lower ranked players won more than 9 games at the International Open, at most 17 of 63 players could have gotten to the hill, if every other weaker player got shut out.
It would be interesting to be able to sort the statistics by match score and look at the Fargo difference instead of sorting by the Fargo differences.
But it they win from the (alternate) break, in an opportunity that wouldn't exist in winner break, they also lose the opportunity to break and win the next rack. While alternate break may give more opportunities, there is a limit to how much a player can capitalize on those opportunities.
Yup and especially with stuff like golden breaks where it doesn't take much skill to win a game it should be pretty obvious it could affect the final score.
Baysian networks....
Except for golden breaks and wired 9-balls after the break.
I can see in a game like 10-ball where you can't win off the break shot it not mattering as much, but it should be pretty obvious if you can win games in one shot and are guaranteed N/2 shots then in a game against a strong player a weaker player will have better odds of a higher score.
Just some perspective here. Let's think about the support for the claim having the break is an advantage to winning the rack. Correct me if I'm wrong, but I believe it comes from the ATLARGE here, who wonderfully documents streamed matches in major professional tournaments. I believe he finds breaker won the game maybe 55-60% of the time. Is that right? Importantly, the data comes not only from the most elite tournaments but from the most elite matches within the most elite tournaments.
The plot below shows the most recent 10,000 matches that went into FargoRate (hundreds of tournaments played in the last week). The matches are sorted by the rating of the higher-rated player in the match. Only the matches in the far right bin have players who even might be included in ATLARGE's analyses.
In my pool room in Fargo, we had in ten years about 300,000 alternate break 8-Ball league games. From the same population of players we had about 100,000 winner-break 8-Ball games in our weekly 8-Ball tournament. Both league and tournament games might have been about 10% single inning games. We have failed to detect a difference. That means any difference is smaller than the noise in the data. We also can't detect a difference in other places we've looked (Turning Stone vs International Open as one example brought up earlier)
There is another issue that doesn't affect Fargo Ratings that is often confused with the one you raise, and that is that AB leads to less volatility in the match scores than WB. This leads people to conclude, incorrectly, that AB matches are more closely contested. That's an illusion though. Consider Gorst and Filler playing 30 9-Ball races to 10 with template rack and 4.5" pockets.
AB: match count would be about 15 to 15, and because there are many 10-9 and 10-8 scores, game count in the end might be 280 to 280
WB: match count would be about 15 to 15, and because there are many 10-0 and 10-1 scores, game count in the end might be 220 to 220
So though the scores are tightened up with AB, that doesn't mean the matches are tightened up competitively. If Gorst has an 8-6 lead here with AB, that's pretty much insurmountable. Even an uncharacteristic mistake or a cueball getting kicked in on the break probably won't turn things around. With WB, though, Gorst is one weak safety away from losing this match.
Thanks for providing the data showing that with AB scores are 'tightened'. But this is where psychological factors might come into play. If the weaker player plays more poorly during a 'tight' match, then on average the weaker player has a greater chance to win with WB compared to AB, simply because he'll have a smaller chance of a tight match in WB.
On the flip side, one could also argue that a certain weaker player actually plays stronger during a tight match, in which case this weaker player has a greater chance to win with AB.
Are you aware of the Accu-stats analysis of breaker-wins percentages from the pro tour in the 1980s?
Trial and error. Start by aiming a laser pointer at his ferrul and go from there.
I don’t know the theory, but if I had the data at my disposal I could compare winning percentages across different skill levels, formats, tournaments, venues etc. The human element would be baked into those outcomes. For example, by this point matchroom has given us a nice sampling of Pros vs. 600 level players on a tv table with mid-level production. That should give us an idea of whether performance meaningfully dips due to those circumstances.
That’s all more granular than we really need to see. I guess all I’m getting at is my original point that I would like to compare outcomes across specific skill differences (25 points, 50 points, 100 points etc.) and different categories (830 vs. different skill gaps, 700 vs. different gaps etc.). I may not be correct, but my hunch is that format has a bigger or lesser influence depending on the gap and skill level of the player.