Another thread about shooting percentages

[ChrisE]

Using MLS's statistical categories, there's really three possible results of a shot:

goal, shot on goal,
not goal, shot on goal,
not goal, shot not on goal.

I thought I would take a look at the second and third categories, the misses, and see how they relate to each other. I don't think the statistic is particularly good at predicting anything (nothing that I've come up with, at least), but it's an interesting descriptor of what kind of a player someone is.

So, to get those two categories, I subtracted out penalty goals and shots (since, again, they're substantially different from a regular game situtation), and then subtracted goals from shots on goal, and shots on goal from shots. Then, I divided (non-scoring) shots on goal by shots not on goal.

Using career stats (and including only players with 50 or more career shots, a total of 212 players), we get an average of .65, so approximately 2 missed shots on goal for every 3 missed shots off goal. There's a rather weak correlation between shooting % and this SoG/SnoG ratio, .30.

The numbers were distributed like such:


Bin Frequency
0.243 1
0.331 2
0.419 19
0.507 27
0.595 34
0.683 33
0.771 33
0.860 32
0.948 13
1.036 10
1.124 1
1.212 1
1.300 3
1.388 2
More 1


The top 25 in SoG/SnoG were:



Elliott, Gerell 1.48
Torres, Johnny 1.37
Marino, Pete 1.36
Santel, Mark 1.26
Comas, Alex 1.25
Takawira, Vitalis 1.21
Wright, Paul 1.17
Conteh, Abdul Thompson 1.10
Rammel, Steve 1.03
Hermosillo, Carlos 1.03
Cerritos, Ronald 1.00
Naveda, Alberto 1.00
Convey, Bobby 1.00
Guevara, Amado 1.00
Olsen, Ben 0.99
Lozzano, Lawrence 0.96
Doyle, John 0.96
Joseph, Miles 0.95
Beasley, Jamar 0.94
Moreno, Jaime 0.94
Lassiter, Roy 0.93
Savarese, Giovanni 0.91
Ruiz, Carlos 0.91
Wolyniec, John 0.90
Myers, Roy 0.89


The biggest surprise here, in my opinion, was Bobby Convey, who I've always taken to be a guy who can't put the ball on goal. Apparently, he's just a guy who can't get it in. Some of the league's proudest poachers, Pete Marino, Steve Rammel, Digital Takawira, lie at the top of the list, and there's obviously a pretty heavy trend towards goal-scorers.

The bottom 25 were:


Balboa, Marcelo 0.42
Hernandez, Daniel 0.42
Johnson, Edward 0.42
Bowers, Sean 0.42
Pena, Danny 0.42
Petke, Mike 0.41
Gutierrez, Diego 0.41
Salcedo, Jorge 0.41
Donadoni, Roberto 0.40
Wilmar Perez, John 0.40
Pope, Eddie 0.40
Franchino, Joe 0.39
Gorter, Edwin 0.39
Ramos, Mauricio 0.39
Kubik, Lubos 0.38
Sorber, Mike 0.38
John, Stern 0.38
Vermes, Peter 0.37
Jara, Guillermo 0.36
Vermillion, Scott 0.35
Villegas, Peter 0.35
Kotschau, Ritchie 0.34
Aunger, Geoff 0.33
Okoh, Matt 0.31
Zavagnin, Kerry 0.24


Two names immediately jump out from this list: Stern John and Eddie Johnson. I never saw Stern John play, it's pretty amazing to me that he's so very low, considering he's ninth all time in shooting percentage (can anyone provide an explanation?) Johnson, on the other hand, is not: he's good at getting himself in position to score, he's crummy at actually scoring. However, to be fair, some pretty proficient scorers are quite close to his rate: Damani Ralph and Edson Buddle, both at .45.

Finally, here's the top 10 all time in shooting percentage:


Savarese, Giovanni 0.91 0.228
Molnar, Miklos 0.64 0.226
Shannon, Musa 0.46 0.215
Pineda Chacon, Alex 0.71 0.214
Prampin, Alan 0.78 0.205
Lassiter, Roy 0.93 0.204
Rammel, Steve 1.03 0.203
Marino, Pete 1.36 0.202
John, Stern 0.38 0.197
Twellman, Taylor 0.88 0.197

[beineke]

As usual, you've come up with some interesting stuff. I wonder if you might post some version of this info on the Crew board, to get an opinion about Stern John.

1. What jumps out at me from these numbers is the contrast of power vs. control (also long-range vs. short-range) -- if you rely on placement, you'd better be able to keep it on net. (Incidentally, I don't think EJ hits his shots with much pace yet, so his lack of placement is worrisome.)

2. As young players fill out and get stronger --Convey and Donovan in particular -- I wonder if their numbers will reflect a greater willingness to "have a crack," even if they risk mis-hitting a few.

3. A related comparison you might look at is a shooter's "save percentage" -- how often his shots on goal get saved.

4. I'd be interested to know whether John Doyle's accuracy (mostly headers on set-pieces) resulted in a few extra goals. One of my pet peeves is how a lot of guys try to aim these headers for the top corner or side netting. Almost invariably, they end up putting them off-frame. If they had tried put them downward and on-net, they would have had a much better chance of scoring or creating a goal for a teammate.

[NoSix]

ChrisE, please check your PM's.

[ChrisE]

Originally posted by NoSix
ChrisE, please check your PM's.

Might as well use this thread, NoSix. Gets about the same amount of attention. :)

Also, please, just Chris.

[ChrisE]

Originally posted by beineke
As usual, you've come up with some interesting stuff. I wonder if you might post some version of this info on the Crew board, to get an opinion about Stern John.

1. What jumps out at me from these numbers is the contrast of power vs. control (also long-range vs. short-range) -- if you rely on placement, you'd better be able to keep it on net. (Incidentally, I don't think EJ hits his shots with much pace yet, so his lack of placement is worrisome.)

2. As young players fill out and get stronger --Convey and Donovan in particular -- I wonder if their numbers will reflect a greater willingness to "have a crack," even if they risk mis-hitting a few.

3. A related comparison you might look at is a shooter's "save percentage" -- how often his shots on goal get saved.

4. I'd be interested to know whether John Doyle's accuracy (mostly headers on set-pieces) resulted in a few extra goals. One of my pet peeves is how a lot of guys try to aim these headers for the top corner or side netting. Almost invariably, they end up putting them off-frame. If they had tried put them downward and on-net, they would have had a much better chance of scoring or creating a goal for a teammate.

As usual, beineke, your comments are much appreciated, and far better and clearer than anything I can muster. Let me address them, obsessive-compulsively, point-by-point:

1. I originally thought of it more in terms of athletes vs. scorers. It makes sense at the top, with Rammel and Marino and then Twellman and Ruiz and Hermosillo, but then there's people like Roy Lassiter (17th), Jamar Beasley (23rd), and Cobi Jones (28th) who sort of ruin that perspective. I don't really like the power vs. control idea because you see very technical players, like Clint Mathis, Brad Davis, and Edson Buddle well below average. I like the long/shot shooting idea better, but I don't think the stat can be predicted very well from any single feature (unfortunately; hopefully I'm wrong).

2. Who are you talking about exactly? Young players in general, or specifically the P-40, U-20 crowd? I'd certainly be interested in looking at this, but I don't know if our sample right now is nearly big enough.

3. I took a look at this, and to me, it looked pretty much identical to sog/snog. Of the top 10 in save percentage, 6 were in the top 10 S on/not, 3 were in the top 20 (Sutter wasn't). What exactly do you think save % is going to tell us that th Shots on/not isn't?

4. Doyle's shooting percentage was slightly above average (.104, vs an average of .083 for defenders); the second best defender, Tony Sanneh, was even better (.84, .135); so was #4 Jose Vasquez (.83, .131); however, #3 Zak Ibsen (.83, .067) and #5 Ryan Nelsen (.82, .089) weren't. I, of course, don't know how many of those shots were headers, or how many of a defender's shots are, on average, headers, so it's hard to say precisely what effect this had.

[beineke]

Originally posted by ChrisE
3. I took a look at this, and to me, it looked pretty much identical to sog/snog. Of the top 10 in save percentage, 6 were in the top 10 S on/not, 3 were in the top 20 (Sutter wasn't). What exactly do you think save % is going to tell us that th Shots on/not isn't?


Remind me ... "sog" was MISSED shots on goal or ALL shots on goal. I was taking it to be missed shots; am I wrong? If it is missed shots, then what you've just said makes me interested to see the correlations between the three numbers -- goals, missed sog, and shots off-goal. It could have some interesting implications about predicting
future shooting %.

BTW, I think the power/placement thing isn't so much about whether a player is technically skilled. After all, shooting with power is very technical. Rather, it's about whether his shots have enough zip to beat a keeper ... and Mathis and Davis have cannons. [Though I do agree ... as with most sports stats, there probably won't be a silver bullet explanation.]

[ChrisE]

Originally posted by beineke
Remind me ... "sog" was MISSED shots on goal or ALL shots on goal. I was taking it to be missed shots; am I wrong? If it is missed shots, then what you've just said makes me interested to see the correlations between the three numbers -- goals, missed sog, and shots off-goal. It could have some interesting implications about predicting
future shooting %.



The first stat was, basically, (missed shots on goal)/(missed shots not on goal).

I took your suggestion, save percentage, to mean (missed shots on goal)/(shots-pka); so, obviously, they ended up with the same numerator. I may be doing the math wrong, I already screwed up once doing this, but if not, you see a correlation of .928 between the two stats.







BTW, I think the power/placement thing isn't so much about whether a player is technically skilled. After all, shooting with power is very technical. Rather, it's about whether his shots have enough zip to beat a keeper ... and Mathis and Davis have cannons. [Though I do agree ... as with most sports stats, there probably won't be a silver bullet explanation.]

Good point.

[beineke]

Originally posted by ChrisE

I took your suggestion, save percentage, to mean (missed shots on goal)/(shots-pka); so, obviously, they ended up with the same numerator. I may be doing the math wrong, I already screwed up once doing this, but if not, you see a correlation of .928 between the two stats.
[/B]

Ahh ... what I intended was

missed shots on goal/(goals + missed shots on goal)

That would be (roughly?) equal to the save percentage of keepers who face that shooter.

[ChrisE]

Originally posted by beineke
Ahh ... what I intended was

missed shots on goal/(goals + missed shots on goal)

That would be (roughly?) equal to the save percentage of keepers who face that shooter.

Oh, well, that's a lot more interesting. Instead of using save percentage, I thought I'd use scoring percentage (so 1-save%); just seems like a more intuitive statistic when you're talking about goalscorers.

Correlation between on-goal goal % and on goal ratio is -.08; average % is .230. Unsurprisingly, this list is going to make the guys who were at the bottom of the last list look a lot better.

Stern John is number 1, scoring on 47.3% of his shots, Edson Buddle is 5th, Eddie Johnson is 12th.

Here's the top 20:


John, Stern 0.473
Shannon, Musa 0.465
Molnar, Miklos 0.429
Pineda Chacon, Alex 0.396
Buddle, Edson 0.386
Savarese, Giovanni 0.383
Gilmar 0.370
Prampin, Alan 0.370
Valencia, Adolfo 0.362
Welton 0.358
Serna, Diego 0.350
Johnson, Edward 0.350
Lassiter, Roy 0.348
Martinez, Antonio 0.348
Klein, Chris 0.346
Wolff, Josh 0.343
Twellman, Taylor 0.343
Wilmar Perez, John 0.343
Nowak, Peter 0.338
Rammel, Steve 0.333


I think the most impressive here are Taylor Twellman, Giovanni Savarese, Roy Lassiter, and Steve Rammel, who manage to show up on both lists.

Bottom 20 (note, a lot of a-mids and d-mids):


Jara, Guillermo 0.133
Ramos, Tab 0.131
Martino, Kyle 0.130
Gori, Mario 0.130
Hernandez, Daniel 0.121
Convey, Bobby 0.119
Hejduk, Frankie 0.119
Lewis, Eddie 0.118
Vaca, Joselito 0.116
Clark, Mike 0.115
Cloutier, Braeden 0.114
Alvarez, Leonel 0.111
Torres, Johnny 0.106
Corrales, Ramiro 0.105
Albright, Chris 0.088
Sutter, Alain 0.077
Mulrooney, Richard 0.067
Burns, Mike 0.063
Guevara, Amado 0.056
Rodas, Jorge 0.040


That's all I've got time for right now.

[ChrisE]

So, it occurs to me that we can get a lot of these boutique shot percentage stats from just two basic ones, shooting percentage (goals/shots) and on goal percentage (sog/shots).

If we take shooting percentage = a, and on goal percentage as b, we can get:

the statistic I adjusted from beineke's suggestion, goals/sog: a/b

beineke's suggestion, saved %: (1-a)/b

non-goal shots on goal %: a-b

shots not on goal %: 1-b

the weird ratio I started the thread with: (a-b)/(1-b)

etc.

and, because these are the only three stats that we really have in regards to shooting the ball, if you can't make the stat with those two formulas, you can't make it.

Not really relevant, just made me feel sort of dumb.