|
Post by meanmug on Apr 19, 2013 12:48:19 GMT -5
Building the database is the expense and you're looking at at about 5-10k for the initial set up.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Apr 20, 2013 19:32:44 GMT -5
Building the database is the expense and you're looking at at about 5-10k for the initial set up. I don't consider that expensive. I suspect if the project was useful to people--finding people willing to donate money and/or time to the project is possible. I have no computer skills to do it--just to be clear on that--but contributing something financially to get a volleyball-reference database set up...I'd be willing to help with that.
|
|
|
Post by uvavolleystats on Jul 8, 2013 12:40:28 GMT -5
|
|
|
Post by meanmug on Jul 8, 2013 14:46:21 GMT -5
I wouldn't start getting into rating passers by actual FBSO instead of expected FBSO just based off the quality of pass. Actual FBSO can be really erratic and there's a lot of noise involved.
As you mention in the article, you have to be sure that you're actually measuring what you think you're measuring, instead of lots of other factors.
|
|
|
Post by n00b on Jul 8, 2013 14:58:34 GMT -5
I wouldn't start getting into rating passers by actual FBSO instead of expected FBSO just based off the quality of pass. Actual FBSO can be really erratic and there's a lot of noise involved. As you mention in the article, you have to be sure that you're actually measuring what you think you're measuring, instead of lots of other factors. I think that's the goal. But you have to use past data to be able to calculate expected FBSO. Also, am I interpreting that correctly that UVA's opponents had over 11% error on overpasses? That seems high...
|
|
|
Post by meanmug on Jul 8, 2013 15:45:08 GMT -5
To calculate expected FBSO based off each type of pass, a league average would be better than a team average (although a team average is better than nothing), otherwise you are measuring passing + hitting instead of just passing.
The overpass figure is probably exaggerated due to the small sample size. With only 52 overpasses for the season, if your opponent gaffs 6 of them, they are at 11.5%, but if they gaff 2 less (which could easily be due to variances in quality of opposing blocker, whether your setter was frontrow or backrow, referee calls, etc), then you are 7.7%, which is a pretty substantial difference.
|
|
|
Post by meanmug on Jul 8, 2013 15:46:59 GMT -5
One more reason to use league average:
"For reference, Chris determined an FBSP benchmark as 42.4%; the team average for ACC winners Florida State." "Current: FBSP = .426(3P%)..."
Those two statements are not compatible.
|
|
|
Post by uvavolleystats on Jul 9, 2013 9:00:15 GMT -5
meanmug is correct in their explanation of why the overpass error rate is higher than expected. The sample size is small which makes it less reliable.
When calculating the FSU averages, I did not use the same scoring percentages for each pass type as I did with the UVA averages. I figured out FSU's percentages and plugged them into the formula instead. FSU was in 3P 48% of the time compared to 34% for UVA which makes a big difference. They also got a first swing kill 57.8% of the time they were in 3P compared to 42.6% for UVA. This makes it very evident that the FBSP is heavily weighted by the success of your attack. Obviously, the FBSP for the UVA players would be higher if their passing percentages were calculated with the FSU scoring percentages. I am not saying that this is a proper way of weighing things but it gives me a different look from the typical 3/4 point passing scale. I am still in the exploratory phase of this entire process.
|
|
|
Post by meanmug on Jul 9, 2013 9:17:34 GMT -5
uva,
It's a great method. What I am pointing out is that you can't use 42.4% as a benchmark and have your perfect pass be worth 42.6%. Otherwise, to hit your benchmark, you will essentially need to pass every ball perfectly. I've been using this method for a couple years now, the process you want to go through is:
(1) Get league averages for FBSO rates off different quality passes. (2) Get leagues averages for overall pass distribution (ie, what an average expected FBSO number is in your league) (3) Compare to your team's performance.
|
|
|
Post by uvavolleystats on Jul 9, 2013 10:05:08 GMT -5
Now I understand what you are saying. I know that the 42.4% "benchmark" in unobtainable given our scoring percentages during the 2012 season. I wouldn't necessarily call it a benchmark as much as I would call it a long term goal. If we can get to a point where our FBSP is at or above 40% we should have much more success.
I have developed a similar formula for digs and have been working with the guy that writes these articles on that as well. He should be posting that article within the next week or two. Once we wrap up camps I want to start a new project that can quantify the effectiveness of a setter. Have you ever done anything with this? I'd appreciate any information you can share based on your experiences.
|
|
|
Post by meanmug on Jul 12, 2013 13:19:10 GMT -5
uva, I sent you a sitemail.
|
|
|
Post by meanmug on Jul 17, 2013 9:20:13 GMT -5
Bumping this for uva...
|
|