As I said, the sample size is small but let's do it anyways
Forwards:
NAME
RelQoC
OZS
CorsiRel
CorsiOn
P/60
Wellwood
1.444
52.9
57.6
42.62
0.00
Ladd
-1.715
65.6
26.9
18.47
0.00
Little
-1.608
61.5
17.7
12.53
0.00
Kane
2.825
37.0
12.3
08.04
1.34
Antropov
0.156
75.9
09.2
06.32
0.00
Burmistrov
2.803
70.6
05.2
03.55
0.00
Jokinen
3.159
40.0
04.5
03.70
0.00
Ponikarovsky
2.490
50.0
-01.8
-01.85
0.00
Wheeler
1.568
53.8
-07.5
-05.71
1.43
Scheifele
0.673
34.6
-22.0
-18.36
0.00
Slater
-1.867
15.4
-37.5
-30.75
4.39
Thorburn
-1.415
21.1
-46.5
-38.89
2.43
Wright
-0.505
0.00
-70.0
-57.37
0.00
Stats will never lie... but sample sizes sure will...
As the season goes by these numbers will become more representational on how they are actually doing. Problem is, IMO, it takes 30 games to get a real good sample and the season is only 48 games... Oh well, still fun stuff! I wouldn't worry about the degree or severity of the numbers but the sign is a lot more telling.
As we already saw, Noel kind of changed game plans a bit between game 1 and 2/3, whether that is due to home vs away, trying to push/protect players or because he was reacting, I have no clue. On top of that he did some line changing in the last game... So, this is more just fun observational stuff.
1) Noel is burying Wright (0 OZS poor guy!!), Slater and Thorburn into the defensive zone to push up his other lines into more offense prime minutes, much like he did with GST last year. What's new is these guys are no longer ALSO facing the toughest competition. I much rather it this way... Last season these guys were being outchanced 20:1/60mins... right now they are actually doing worse, but I'm hoping this will improve with the lower competition once the sample spread increases. Slater is currently our #1 5v5 P/60 player scoring at similar rates as Malkin and Crosby last season ... of course this will likely go down haha.
2) Looks like Kane's line and Ladd-Little's line have switched roles although it may be too soon to tell. My guess is Noel was sheltering LLW to try to get them going plus in interviews he noted that he thought both Kane's and Jokinen's defensive games have improved over the last season and a bit. Personally I prefer Kane-Jokinen scoring and Ladd-Little taking on toughs but it's early and players are rusty (as coaches are too!).
3) I like Burmistrov! If he continues to out-chance his opponents while facing tough match-ups as a 21 year old we may be seeing a good year for Burmi!
As I said, the sample size is small but let's do it anyways
Defense:
NAME
RelQoC
OZS
CorsiRel
CorsiOn
P/60
Enstrom
2.146
46.7
38.7
23.98
2.28
Byfuglien
1.997
44.7
23.1
12.83
0.99
Postma
1.413
70.0
10.0
07.24
1.81
Clitsome
2.123
65.0
-08.4
-06.69
0.00
Hainsey
-1.336
40.0
-32.0
-20.66
0.00
Stuart
-0.851
35.5
-37.9
-24.74
0.00
Stats will never lie... but sample sizes sure will...
As the season goes by these numbers will become more representational on how they are actually doing. Problem is, IMO, it takes 30 games to get a real good sample and the season is only 48 games... Oh well, still fun stuff! I wouldn't worry about the degree or severity of the numbers but the sign is a lot more telling.
1) Noel seemed to change his D deployment after the first game. In the first game he was pushing up Tobi and Buff, but in the 2nd game he started pushing up Clitty and Posty instead which will be interesting to see if Enstrom and Byfuglien will dominate as much as last season without the offensive zone push. My guess is once Bogosian returns a lot of this will change.
2) Hainsey and Stuart are getting the GST treatment with low O-zone starts, but their low QoC makes me wonder how off the QoC is with due to small sample sizes (especially since one team -WSH- doesn't exactly have their top6 firing on all cylinders).
3) When the eye test = stats test, it usually means we've found the truth. It does appear that Clitsome is indeed the weaker of the 3rd pair pieces. If this trend continues, we may see him as our #7 or #8 when Bogosian returns.
Too early for this to mean much but the Sdiff gives you a bit better look than GF/GA currently would... This is 5v5 only.
EDIT: added FenwickClose since Gabe has updated it on behindthenet.com... again this gives you even better look than the Sdiff... obviously some teams will pick up/fall down as games go, as it will take a lot of teams a few games to get to a proper groove.
If you are curious about the difference between our home and 2 away games.:
*Our home Fenwick was 45.83... which would have made us as terrible as WSH has been.
*Our away Fenwick was 53.84... which would put us ahead of LAK and BUF.
WPG vs OTT
*shows our terrible 3rd period collapse
WPG vs BOS
*definitely a positive sign us battling pretty much the whole game since BOS is such a strong Fenwick team
WPG vs WSH
*obvious and complete utter dominance by us
WPG vs PIT
*when ever you see something like this and the lower team wins like we did, this is when you can say Pavs won us a game (plus some great plays in some of the goals)
PDO or as us number nerds call it, the luck stat...
PDO is simple math, all you do is add a player's on ice save percentage (team's sv% with _____ player on the ice) and their on ice shooting percentage (team's shooting percentage with _____ player on the ice). In general, teams and players will end up with a value close to 1000. If the number is higher, that player can be considered the beneficiary of good luck, a low number is representative of bad luck.
The stat isn't perfect, but it can be a very good indicator of a player or team that is under / over achieving. Teams almost always end up close to 1000. Players generally have numbers a bit above or below.
Anyways...
Good news from the PDO perspective...
Kane, Olli and Wheeler are suffering from some of the lowest PDO on the team. All three are in the Jets bottom 5 with wright and Clitsome. Clitsome belongs there because he has been brutal, but the Jets top line has acutally s
All three players are corsi positive and PDO negative. The fact that they are scoring as much as they are without the benefit of puck luck is pretty impressive. Granted most of their PDO hit has come as a result of goaltending, they are a little behind as shooters too.
Probability owes Jokinen a goal or two.
Bottom line...
Our big guns aren't overachieving and the fact that they are all in the minus isn't likely to continue.
PDO or as us number nerds call it, the luck stat...
PDO is simple math, all you do is add a player's on ice save percentage (team's sv% with _____ player on the ice) and their on ice shooting percentage (team's shooting percentage with _____ player on the ice). In general, teams and players will end up with a value close to 1000. If the number is higher, that player can be considered the beneficiary of good luck, a low number is representative of bad luck.
The stat isn't perfect, but it can be a very good indicator of a player or team that is under / over achieving. Teams almost always end up close to 1000. Players generally have numbers a bit above or below.
Anyways...
Good news from the PDO perspective...
Kane, Olli and Wheeler are suffering from some of the lowest PDO on the team. All three are in the Jets bottom 5 with wright and Clitsome. Clitsome belongs there because he has been brutal, but the Jets top line has acutally s
All three players are corsi positive and PDO negative. The fact that they are scoring as much as they are without the benefit of puck luck is pretty impressive. Granted most of their PDO hit has come as a result of goaltending, they are a little behind as shooters too.
Probability owes Jokinen a goal or two.
Bottom line...
Our big guns aren't overachieving and the fact that they are all in the minus isn't likely to continue.
That's actually super interesting. Sounds like it gets more and more "accurate" as the season goes on. I'm taking it slow, but this advanced stat stuff is pretty useful at confirming or denying what your eyes see.
I'm assuming Bogo with all his hit posts last year was under 1.000 lol
WOOT dominance in puck possession... it's too bad that they capitalized on their chances more often than we did
Team Fenwick
TEAM
GP
PTS
GF
GA
FenwickClose
STL
6
10
24
13
69.69
CAR
4
04
11
13
58.40
LAK
4
03
08
12
55.75
OTT
5
07
16
10
55.70
PHI
6
05
13
18
53.80
COL
4
04
09
09
53.61
VAN
5
05
14
16
53.37
DET
5
05
11
16
53.33
MTL
4
06
13
07
52.78
PIT
5
06
15
14
52.42
BOS
4
07
12
08
52.15
PHX
5
02
17
20
51.46
NYR
5
04
14
16
50.29
DAL
5
05
11
12
49.79
CHI
6
12
22
13
49.48
ANA
4
06
15
14
48.97
NYI
5
05
18
18
48.92
WSH
5
03
11
19
48.84
CGY
4
03
11
15
48.80
WPG
5
07
15
14
48.32
EDM
4
04
14
16
48.15
BUF
5
04
13
15
48.15
NSH
5
05
10
14
47.67
SJS
5
10
23
08
47.66
NJD
4
07
11
07
46.92
TBL
5
08
24
13
46.15
MIN
5
05
13
15
45.86
TOR
5
04
14
17
45.70
CBJ
5
03
09
18
40.53
FLA
5
02
08
19
36.17
This Fenwick data shows you how much the team is out-chancing or being out-chanced compared to others. It is a good indication of team strength once the sample size gets large enough. Any discrepancy in this order from how teams end up are for one or both of two reasons:
1) Shooting percentages: a team may be capitalizing on their fewer opportunities more so than other teams are capitalizing on their many; this almost happened to the Jets' last night against NYI; this usually regresses over the season as players like Marleau can't consistently score throughout the year on 64% of his shots or Jokinen on 0% of his
2) Goaltending: unlike the one above, this may actually not regress as many teams will have different levels of goaltending
Teams can sometimes ride these crazy percentages for awhile as seen last season with MIN and TOR (and kept LAK down for almost a full season).
That's actually super interesting. Sounds like it gets more and more "accurate" as the season goes on. I'm taking it slow, but this advanced stat stuff is pretty useful at confirming or denying what your eyes see.
I'm assuming Bogo with all his hit posts last year was under 1.000 lol
It basically boils down to luck and streaks evening out over time, especially with the team numbers. There is a little more individual ability involved in the personal PDO numbers.
Bogo's PDO was 987 last year. Definitely a little on the low side.
Maybe I dont' fully understand these figures, but when I look at these images I veiw them from a slightly different perspective. vertical distance between the lines is only meaningful if it is indicative of two different slopes in the lines. Parallel lines only indicate that at some point one team outdid the other, not that it was consistent. The second period of the Washington game is pure dominance by Winnipeg. The slope is much steeper indicating a sustained advantage. Brief spurts, such as those by Pittsburgh at about 27 minutes and just before Ladd's goal create an illusion of dominance. Since the lines are essentially parallel after that, Pittsburgh did not outplay the Jets from that point forward. Nor did they really outplay them in the first period, since they end the period with no distance between them at all. The use of time on the horizontal axis will create these illusions. So yes, Pavelec was great, but did he steal a game for us. I dont' think so.
The islanders image is quite telling. from 5 minutes to 30 the distance increases, the slopes are different, WPG is dominating. The last ten minutes of the second period the Isles close the gap, but it is about parallel after that. Pretty much what I thought watching it. The goals in the third weren't about the Islanders dominating the Jets. They were jsut a lot better than they were earlier, and they got some good bounces.
lond story short: with time on your horizontal axis, it's the slopes that matter more than vertical distance.
Quote:
Originally Posted by garret9
WPG vs OTT
*shows our terrible 3rd period collapse
WPG vs BOS
*definitely a positive sign us battling pretty much the whole game since BOS is such a strong Fenwick team
WPG vs WSH
*obvious and complete utter dominance by us
WPG vs PIT
*when ever you see something like this and the lower team wins like we did, this is when you can say Pavs won us a game (plus some great plays in some of the goals)
Quote:
Originally Posted by garret9
WOOT dominance in puck possession... it's too bad that they capitalized on their chances more often than we did
Team Fenwick
TEAM
GP
PTS
GF
GA
FenwickClose
STL
6
10
24
13
69.69
CAR
4
04
11
13
58.40
LAK
4
03
08
12
55.75
OTT
5
07
16
10
55.70
PHI
6
05
13
18
53.80
COL
4
04
09
09
53.61
VAN
5
05
14
16
53.37
DET
5
05
11
16
53.33
MTL
4
06
13
07
52.78
PIT
5
06
15
14
52.42
BOS
4
07
12
08
52.15
PHX
5
02
17
20
51.46
NYR
5
04
14
16
50.29
DAL
5
05
11
12
49.79
CHI
6
12
22
13
49.48
ANA
4
06
15
14
48.97
NYI
5
05
18
18
48.92
WSH
5
03
11
19
48.84
CGY
4
03
11
15
48.80
WPG
5
07
15
14
48.32
EDM
4
04
14
16
48.15
BUF
5
04
13
15
48.15
NSH
5
05
10
14
47.67
SJS
5
10
23
08
47.66
NJD
4
07
11
07
46.92
TBL
5
08
24
13
46.15
MIN
5
05
13
15
45.86
TOR
5
04
14
17
45.70
CBJ
5
03
09
18
40.53
FLA
5
02
08
19
36.17
This Fenwick data shows you how much the team is out-chancing or being out-chanced compared to others. It is a good indication of team strength once the sample size gets large enough. Any discrepancy in this order from how teams end up are for one or both of two reasons:
1) Shooting percentages: a team may be capitalizing on their fewer opportunities more so than other teams are capitalizing on their many; this almost happened to the Jets' last night against NYI; this usually regresses over the season as players like Marleau can't consistently score throughout the year on 64% of his shots or Jokinen on 0% of his
2) Goaltending: unlike the one above, this may actually not regress as many teams will have different levels of goaltending
Teams can sometimes ride these crazy percentages for awhile as seen last season with MIN and TOR (and kept LAK down for almost a full season).
Maybe I dont' fully understand these figures, but when I look at these images I veiw them from a slightly different perspective. vertical distance between the lines is only meaningful if it is indicative of two different slopes in the lines. Parallel lines only indicate that at some point one team outdid the other, not that it was consistent. The second period of the Washington game is pure dominance by Winnipeg. The slope is much steeper indicating a sustained advantage. Brief spurts, such as those by Pittsburgh at about 27 minutes and just before Ladd's goal create an illusion of dominance. Since the lines are essentially parallel after that, Pittsburgh did not outplay the Jets from that point forward. Nor did they really outplay them in the first period, since they end the period with no distance between them at all. The use of time on the horizontal axis will create these illusions. So yes, Pavelec was great, but did he steal a game for us. I dont' think so.
The islanders image is quite telling. from 5 minutes to 30 the distance increases, the slopes are different, WPG is dominating. The last ten minutes of the second period the Isles close the gap, but it is about parallel after that. Pretty much what I thought watching it. The goals in the third weren't about the Islanders dominating the Jets. They were jsut a lot better than they were earlier, and they got some good bounces.
lond story short: with time on your horizontal axis, it's the slopes that matter more than vertical distance.
The slopes show short bursts of dominance and yes, you want to have more vertical slopes, but the bottom line is you just want to out chance your opponent and that does result in vertical distance.
The slopes show short bursts of dominance and yes, you want to have more vertical slopes, but the bottom line is you just want to out chance your opponent and that does result in vertical distance.
yeah, my response more had to do with the idea that the result of the Pittsburgh game was some kind of anomaly because Winnipeg was the lower team. That Pittsburgh had dominated them. The truth is... no, not really.
Pittsburgh outplayed Winnipeg for the first 15 minutes. After that, not really so much. At 57 minutes or so, the two teams are barely separated. The last minutes barrage due to 6 on 5 play makes the end result look worse.
yeah, my response more had to do with the idea that the result of the Pittsburgh game was some kind of anomaly because Winnipeg was the lower team. That Pittsburgh had dominated them. The truth is... no, not really.
Pittsburgh outplayed Winnipeg for the first 15 minutes. After that, not really so much. At 57 minutes or so, the two teams are barely separated. The last minutes barrage due to 6 on 5 play makes the end result look worse.
Yeah. The Jets played them pretty even and even the late game outburst was largely due to "score effect", a combination of the Jets sitting back and the Pens pressing. That said, you ideally want to come out ahead in every game.
Keep in mind that the top 4 minutes have been waywaywaywayway tougher than the bottom pair minutes....
Here we are in order of minutes...
Name
RelQoC
OZS
SA/60
GA/60
Corsi
Byfuglien
1.616
45.6
29.6
3.34
-01.90
Enstrom
1.545
47.2
25.3
2.63
00.53
Hainsey
0.327
43.1
33.4
0.56
-20.05
Stuart
0.662
39.4
34.6
1.19
-21.46
Clitsome
-2.298
63.6
31.1
5.33
01.78
Postma
-2.555
64.7
25.6
2.74
13.71
Well... this makes me sad because it agrees with the eye test... Postma is doing SO much better than Clitsome... why is Noel giving (slightly) more time and (slightly) less sheltering to Clitsome... Clitsome is the weak player...
Clitsome should not be having shot/goal suppression numbers similar to Stuart and Hainsey who are buried in the Dzone...
Keep in mind that the top 4 minutes have been waywaywaywayway tougher than the bottom pair minutes....
Here we are in order of minutes...
Name
RelQoC
OZS
SA/60
GA/60
Corsi
Byfuglien
1.616
45.6
29.6
3.34
-01.90
Enstrom
1.545
47.2
25.3
2.63
00.53
Hainsey
0.327
43.1
33.4
0.56
-20.05
Stuart
0.662
39.4
34.6
1.19
-21.46
Clitsome
-2.298
63.6
31.1
5.33
01.78
Postma
-2.555
64.7
25.6
2.74
13.71
Well... this makes me sad because it agrees with the eye test... Postma is doing SO much better than Clitsome... why is Noel giving (slightly) more time and (slightly) less sheltering to Clitsome... Clitsome is the weak player...
Clitsome should not be having shot/goal suppression numbers similar to Stuart and Hainsey who are buried in the Dzone...
Interesting, Noel did cut Clitsome's time very sharply in the Habs game. It'll be interesting too see exactly how much Clitsome is dragging Postma down. Even looking beyond the stats for a second it can't be too easy for Postma to play his natural game while worrying if Clitsome can handle his part. I wonder if there is a way to see Clitsome-less stats for Postma i.e Postma's stats when Postma is on the ice but Clitsome is not. I realize it would be a skewed number as the amount of time Postma has played without Clitsome is miniscule but it will be interesting to see.
Interesting, Noel did cut Clitsome's time very sharply in the Habs game. It'll be interesting too see exactly how much Clitsome is dragging Postma down. Even looking beyond the stats for a second it can't be too easy for Postma to play his natural game while worrying if Clitsome can handle his part. I wonder if there is a way to see Clitsome-less stats for Postma i.e Postma's stats when Postma is on the ice but Clitsome is not. I realize it would be a skewed number as the amount of time Postma has played without Clitsome is miniscule but it will be interesting to see.
I'm at school, but I can show you this when I'm home (it's a VERY small sample size)
I have placed all the goal events as goals for, against and as a percentage.
I've done the same thing for Corsi events (which is goals, shots, blocked shots and missed attempts).
So the sample sizes are extraordinarily small... normally I say Corsi events are better because they give a larger sample size and are influenced less by luck/streaks/goalies, but in this case the sample sizes are so small I'd say go for Corsi events not cos it's better but because it's less bad.
Looking at goals:
The Jets get out scored when they are together (no surprise there). There away time is ridiculously small so I wouldn't take too much value in it, but it is interesting that Postma without Clitsome the Jets have only scored, and Clitsome with out Postma the Jets have only been scored on.
Looking at Corsi events (or "chances" or an estimation of zone possession time):
Postma has had better possession in total than Clitsome. They have been in the black together by about 3.5% which isn't that special when you consider how ridiculously sheltered they've been. You'd want/expect competent D to have that number in the low 60s (as a rather random guess). Postma has strongly out-chanced his opponents when away with Clitsome and Clitsome has been out-chanced by opponents. While the eye-test makes me think this is due to talent difference, the sample size isn't large enough to statistically verify the eye test as an absolute.
The battle for the bottom pair once Bogosian and Byfuglien return.
OK... so we all know who will be #8, Clitsome, but who will be #6 and who will be #7.
Even Strength
Name
RelQoC
OZS
SA/60
GA/60
CorsiOn
P/60
Redmond
-0.829
75.0
10.8
0.00
27.09
0.00
Clitsome
-1.027
57.1
31.3
4.56
03.91
0.65
Postma
-1.339
59.6
24.8
3.72
16.72
1.86
*Redmond has looked very promising as being a solid 3rd pair D-man and maybe even a good stop-gap on the second pair to be placed with Hainsey until Bogosian returns. Although he was sheltered slightly, to only have 10.8 SA/60 in two very one-sided losing games is quite impressive. On average he was being less sheltered that Clitsome and Postma match-ups-wise but he has been pushed very heavily into the offensive zone which is probably helping him out a lot possession-wise. Obviously it's a small sample size so the numbers are inflated but at least their inflated in a good direction.
*Clitsome is quite obviously the weakest in the bunch. Being used almost identically to Postma (and usually with Postma) he's had far worse numbers in every sense. Before he could use goaltending as a slight excuse, but now there are 2 players who have worse OnIceSv% who still have better GA/60 than him (it's not you; it's me).
*Postma has consistently been improving his SA/60. Although his GA/60 worsened these last few games, I'm still under the belief it wouldn't be as bad if a) Pavelec didn't have three terrible games in a row and b) he had a different shift partner (especially when he does better than his shift partner).
are there any ways to manipulate OZONE starts and OZONE finish's to get an advanced stat for how players "push" the play?
i've tried a number of things but having no real statistical background i feel like my equation is doing something wrong
mostly i took zone finishes subtract zone starts to determine a "push" factor.
But i noticed this was highly influenced by your zone starts deviation from 50. Everyone with a negative deviation (slater and co) got really high push factor, and anyone with a positive deviation (sheltering a la Redmond) got a very negative deviation.
So i decided to do this
PUSH / OZS Differential
(OZF-OZS) / (50-OZS and remove the integer)
This gave me numbers/ordering that mad ea little more sense with what you'd expect.
My question is, i feel like that division by the OZS differentiall would unfairly punish someone..but i'm nto sure who.
how should i be doing that to add "diminishing returns" on the farther you get away from 50 for you oz's
are there any ways to manipulate OZONE starts and OZONE finish's to get an advanced stat for how players "push" the play?
i've tried a number of things but having no real statistical background i feel like my equation is doing something wrong
mostly i took zone finishes subtract zone starts to determine a "push" factor.
But i noticed this was highly influenced by your zone starts deviation from 50. Everyone with a negative deviation (slater and co) got really high push factor, and anyone with a positive deviation (sheltering a la Redmond) got a very negative deviation.
So i decided to do this
PUSH / OZS Differential
(OZF-OZS) / (50-OZS and remove the integer)
This gave me numbers/ordering that mad ea little more sense with what you'd expect.
My question is, i feel like that division by the OZS differentiall would unfairly punish someone..but i'm nto sure who.
how should i be doing that to add "diminishing returns" on the farther you get away from 50 for you oz's
From what I recall, the differential between OZS and OZF didn't strongly correlate to much...
I think someone at nhlnumbers did something like this but I can't remember who (probably EricT??)...
Makes sense though... shifts can be long and play is very back-and-forth, rarely do you just start in one zone, move to the other and end your shift. The variance and back and forth nature of the game is going to mean that play likely ends (somewhat) 50/50ish in zones irregardless of starting point, so guys who get super pumped into d-zone (Slater + Malholtra) or super pumped into o-zone (Sedins + Karlsson) are going to automatically be penalized/gifted by any differential.
Not saying you are or aren't on to something. But, I'm not sure how strongly it can correlate to possession or scoring chances.... hmmm... interesting...
From what I recall, the differential between OZS and OZF didn't strongly correlate to much...
I think someone at nhlnumbers did something like this but I can't remember who (probably EricT??)...
Makes sense though... shifts can be long and play is very back-and-forth, rarely do you just start in one zone, move to the other and end your shift. The variance and back and forth nature of the game is going to mean that play likely ends (somewhat) 50/50ish in zones irregardless of starting point, so guys who get super pumped into d-zone (Slater + Malholtra) or super pumped into o-zone (Sedins + Karlsson) are going to automatically be penalized/gifted by any differential.
Not saying you are or aren't on to something. But, I'm not sure how strongly it can correlate to possession or scoring chances.... hmmm... interesting...
Truck?
thats' sort of what it looked like, the whole 50 50 thing as a baseline.
either way, when i did the division it did give m numbers that made sense to me- the highest rated forwards on th jets thus far this year were: wellwood ladd little and kane, and the lowest were burmi, clitsome, redmond, and schiefele.
of those 8 names the only one that looked out of place to me was burmi.
EDIT post Truck: Would it be possible to factor in "diminshing returns" on deviation from the 50 baseline? My math isn't good enough to go about that, but it seems like the right approach. all of this of course hinges on zone finish actually meaning something. Though i suppose is if you can establish that the actual baseline truly is 50% then you can figure something out.
EDIT: using my current method sedins wouldbe right around schiefele level right now- so i guess thats true. It does seem helpful when you restrict it to players with a zone start deviation of 40%-60%
are there any ways to manipulate OZONE starts and OZONE finish's to get an advanced stat for how players "push" the play?
i've tried a number of things but having no real statistical background i feel like my equation is doing something wrong
mostly i took zone finishes subtract zone starts to determine a "push" factor.
But i noticed this was highly influenced by your zone starts deviation from 50. Everyone with a negative deviation (slater and co) got really high push factor, and anyone with a positive deviation (sheltering a la Redmond) got a very negative deviation.
So i decided to do this
PUSH / OZS Differential
(OZF-OZS) / (50-OZS and remove the integer)
This gave me numbers/ordering that mad ea little more sense with what you'd expect.
My question is, i feel like that division by the OZS differentiall would unfairly punish someone..but i'm nto sure who.
how should i be doing that to add "diminishing returns" on the farther you get away from 50 for you oz's
is this for 5v5 only? if so, then it would seem to be appropriate if starts and finishes are approximately 50/50 overall (probably pretty close). if a player is, on average, ending a higher percentage of shifts in the OZ than they've started, then they're doing something right.
The main issue is the division by 50 - OZS. The closer someone is to 50-50, the screwier the stat would get due to possiblity of a small denominator. or conversely, the farther someone is form 50-50. imagine someone at 51 OZS, and someone else at 52 OZS. Let's say they both finish 10 points better then they started (61 and 62). so you get 10/1 and 10/2. half the stat for something pretty identical. somebody else could start at 51 and finish at 56 and get the same as the guy at 52 and 62. or someone coudl start at 60 and go to 100 and still not appear as good as the guy at 51 - 61.
so this stat would still have its issues.
I'm not sure what you mean by 'remove the integer'.
... of course this will likely go down haha.
If he continues to out-chance his opponents while facing tough match-ups as a 21 year old we may be seeing a good year for Burmi!
