First and second assists
I compiled from-to matrices for all the Team USA goals (who passed to whom) and similar matrices for passes just before goal passes - so called second assists. (I wanted to compare these passes and their correlation to the total number of passes, as there was some time ago short discussion in the ultimate statistics Yahoo group about value of 2nd assists compared to goal passes).
Similar results as yesterday. Goal passes correlated closer to total passes, although the correlation was not high (Pearson correlation only 0.476 compared to 0.381 for second-assist passes). However there was virtually no correlation between 1st and second assists - sort of interesting.
Another interesting tidbit was that the highest number of goal passes in one game from one person to another was 2. In the summary matrix the highest number was 5 (Watson to Ziperstein and Ziperstein to Eastham). And there was no player who passed at least one goal to every other player. It is no surprise that Namkung was closest to reach this milestone (no goal passes to Fontenette though) - as one probably remembers Namkung was the bridge between male and female players.
September 7th, 2005 at 9:57 pm
2 questions unrelated to the entry:
1) What is the yahoo statistics groups’ address?
2) Given only points played, goals, assists, Ds, and turns as statistics for a team over an entire tournament, what are the most relevant statistics you can come up with?
I like (goals + assists - (.8 * turns)) / points as an offensive rating (assuming a turn is 4/5 of a goal for the opponent), but I’ll be damned if I can think of a good way to evaluate the D - it may just be too little information.
Peace
Bill Mill
bill.mill at gmail.com
September 7th, 2005 at 10:36 pm
1) FrisbeeStats, but we warned that there has not been much activity lately.
2) So you are looking for a single, magic number, right? I do not have a good answer yet, as I have not been thinking of that question lately.
One question though. When you say points played, do you have information for making a plus-minus statistics (on field while score is plus, on field while allowing score is minus, and some adjustments for how the point was started - either on offense or defense)? Or just that person A was on the field during 8 points?
About your formula. The multiplier might need to be different for different divisions. Based on the data I have seen and without any exact calulcations I would say there are more turns in women’s games than in men’s games. On average.
Or are you saying that the multiplier should be different for each of the game a team plays (based on the opponent efficiency)
I am not sure how you came up with 0.8. Now you award two points for each goal but only 0.8 points for each turn. However both end one possession. I would say the multiplier has to be higher.
Hartti
September 8th, 2005 at 12:11 am
Unfortunately, the stats I was given only have points played - the major difference I want to suggest to the team is that they keep plus/minus and whether it was an O or D point.
As for the .8, it’s just a number I’m playing with. I started with 1.5, figuring on penalizing turnovers (in my head, they’re BAD), but the stat seemed to jive better with my subjective assessment when I turned it down to the .66-.8 range. I figure that creating a goal is 1 point (either a goal or an assist) whereas turning it over is worth a likely point for the opponent (somewhere below 1 point, but not far).
I feel like the multiplier should differ not only by division, but by team. Perhaps the opponent’s actual turnover conversion rate for a tournament could be tracked and used as the multiplier? It should be clear that some teams are easier to score on after a turn than others.
Hmmm, I do know how many points the opponent scored, and how many turns there were in each game. If only I knew the number of *opposing* turns, I could figure out what the actual turn conversion rate was. Maybe I should try to come up with some estimate of turn conversion rate - similar to the approximation of possession (http://www.basketball-reference.com/about/glossary.html#Poss) in basketball?
(In case you were wondering, the stats are for a mixed division team).
Peace
Bill Mill
September 14th, 2005 at 4:09 am
I’ve been trying to figure out such an offensive measure as well. So far, I’ve just been sticking to the fantasy type rating of (goals thrown + goals caught + defensive blocks - turnovers) and it seems to be correspond reasonably well with my notion of the players’ impacts. Of note, is that this is on a team where the range of playing time is pretty small (i.e. the player with the most PT might only play 8-10 more points during a tournament than the player with the least amount of PT).
I also have stats on number of touches and completion percentages, so I would love to be able to figure out some way to work that into some sort of “handler rating” but don’t really have any ideas yet. Any thoughts?
September 15th, 2005 at 1:25 am
Handler rating would probably be affected by the number of touches (as well as all the fantasy ratings you mentioned), but shouldn’t it also be affected how much one advances the disc with passes (which requires some pass categorization or storing field position info for all throwers). One problem is opposing defense. Against zone defense the passes are different and the number of touches tend to concentrate more on few people than against man defense.
If we go one step further, should there be difference in handler rating based on the distribution of receivers for one’s passes. Although I am not sure if it would be better for a handler to have many receivers or few receivers… Or does that matter at all. (Maybe I should spend more time thinking than writing comments on my blog…)
Maybe we could do an experiment. We could ask people on RSD (or somewhere else) to rank players based on their handling contributions (subjective ranking) in some game which is available on DVD. Then find out from stats for that game which kind of rating correlates best with the subjective ranking. Although the subjective rankings are probably affected by single highlight catch or one bad throwaway/drop. I could work out the stats for Condors/Furious George UPA Finals 2003 which I am almost halfway through. Or then we could do the same thing during UPA nationals this year.