Most corrupted home takeaway stats

[RapidKnight]

In this exercise I compared each team’s home and away takeaway rates from this season at 5v5 only.

	HT/60	AT/60	Dif
FLA	13.70	5.89	+133%
SJS	11.31	6.84	+65%
NSH	8.46	5.21	+62%
CHI	9.22	5.73	+61%
TOR	10.42	6.65	+57%
EDM	9.55	6.08	+57%
NJD	9.66	6.29	+54%
VGK	11.13	7.28	+53%
COL	8.65	6.12	+41%
PIT	9.37	6.88	+36%
WPG	8.10	6.05	+34%
CAR	9.16	6.97	+31%
STL	9.49	7.61	+25%
WSH	7.38	5.96	+24%
CGY	7.90	6.47	+22%
NYR	8.31	6.93	+20%
ARI	7.58	6.38	+19%
BOS	7.25	6.11	+19%
VAN	7.09	6.34	+12%
MIN	6.49	5.98	+9%
SEA	7.70	7.14	+8%
CBJ	7.77	7.32	+6%
ANA	6.30	6.42	-2%
TBL	6.29	6.78	-7%
DAL	5.58	5.98	-7%
MTL	5.93	6.54	-9%
BUF	5.94	7.01	-15%
PHI	5.51	6.45	-15%
NYI	5.74	6.91	-17%
OTT	5.66	7.44	-24%
DET	4.37	6.54	-33%
LAK	4.04	7.31	-45%
PAC	8.15	n/a	n/a
CEN	7.95	n/a	n/a
MET	7.86	n/a	n/a
ATL	7.44	n/a	n/a

There’s no consistency around the NHL and there’s also a sizable amount of home team bias.

Adjusted takeaways TBA…

Click to read more...

 

[GermanSpitfire]

Very good post - I’m quite surprised Ottawa isn’t higher Given how dominant Stone was for so long I thought that the stats team put a premium on this.

 

[morehockeystats]

There is a "home" bias and an "arena" bias in this stat (as well as in giveaways).

 

[Golden_Jet]

In this exercise I compared each team’s home and away takeaway rates from this season at 5v5 only.

	HT/60	AT/60	Dif
FLA	13.70	5.89	+133%
SJS	11.31	6.84	+65%
NSH	8.46	5.21	+62%
CHI	9.22	5.73	+61%
TOR	10.42	6.65	+57%
EDM	9.55	6.08	+57%
NJD	9.66	6.29	+54%
VGK	11.13	7.28	+53%
COL	8.65	6.12	+41%
PIT	9.37	6.88	+36%
WPG	8.10	6.05	+34%
CAR	9.16	6.97	+31%
STL	9.49	7.61	+25%
WSH	7.38	5.96	+24%
CGY	7.90	6.47	+22%
NYR	8.31	6.93	+20%
ARI	7.58	6.38	+19%
BOS	7.25	6.11	+19%
VAN	7.09	6.34	+12%
MIN	6.49	5.98	+9%
SEA	7.70	7.14	+8%
CBJ	7.77	7.32	+6%
ANA	6.30	6.42	-2%
TBL	6.29	6.78	-7%
DAL	5.58	5.98	-7%
MTL	5.93	6.54	-9%
BUF	5.94	7.01	-15%
PHI	5.51	6.45	-15%
NYI	5.74	6.91	-17%
OTT	5.66	7.44	-24%
DET	4.37	6.54	-33%
LAK	4.04	7.31	-45%
PAC	8.15	n/a	n/a
CEN	7.95	n/a	n/a
MET	7.86	n/a	n/a
ATL	7.44	n/a	n/a

There’s no consistency around the NHL and there’s also a sizable amount of home team bias.

Adjusted takeaways TBA…

What does green, red and yellow mean.

 

[Dr Beinfest]

What does green, red and yellow mean.

Appears to have marked top 8 (top 25%), bottom 8, and median differential (yellow).

 

[RapidKnight]

Top 32 in adjusted takeaways (2021-22)

1. Pietrangelo 74
2. Matthews 73
3. Fox 71
4. Paul 66
5. Slavin 65
6. Cirelli 62
7. Aho 60
8. Kopitar 60
9. Draisaitl 58
10. C. Brown 58
11. McDavid 57
12. Marner 57
13. Zibanejad 57
14. Iafallo 57
15. Connor 56
16. Keller 55
17. Bjorkstrand 55
18. Skinner 55
19. K. Miller 54
20. Thomas 54
21. Kaprizov 54
22. Hedman 54
23. P. Suter 54
24. Thompson 53
25. Nelson 53
26. Pelech 53
27. Burns 52
28. Pettersson 52
29. JT Miller 52
30. Gourde 52
31. Suzuki 52
32. DeBrincat 51

To find the adjusted number I multiplied by a factor of the player’s team’s standardized away takeaways (sAT/60) and total takeaways (home & away). The standardized rate factors removes divisional advantages.

		sAT/60	T/60
1	STL	7.64	8.55
2	OTT	7.41	6.55
3	VGK	7.37	9.22
4	CBJ	7.35	7.55
5	LAK	7.10	5.69
6	SEA	7.08	7.42
7	CAR	7.06	8.06
8	BUF	6.99	6.47
9	PIT	6.98	8.12
10	NYR	6.98	7.62
11	SJS	6.93	9.07
12	NYI	6.86	6.32
13	TOR	6.81	8.55
14	TBL	6.78	6.53
15	MTL	6.52	6.23
16	DET	6.46	5.46
17	CGY	6.42	7.19
18	NJD	6.39	7.98
19	PHI	6.39	5.97
20	ARI	6.33	6.98
21	ANA	6.32	6.36
22	VAN	6.27	6.71
23	FLA	6.15	9.93
24	BOS	6.14	6.68
25	COL	6.11	7.39
26	EDM	6.09	7.82
27	WPG	6.02	7.08
28	MIN	5.90	6.23
29	WSH	5.97	6.66
30	DAL	5.87	5.78
31	CHI	5.74	7.47
32	NSH	5.20	6.82

 

[Golden_Jet]

Top 32 in adjusted takeaways (2021-22)

1. Pietrangelo 74
2. Matthews 73
3. Fox 71
4. Paul 66
5. Slavin 65
6. Cirelli 62
7. Aho 60
8. Kopitar 60
9. Draisaitl 58
10. C. Brown 58
11. McDavid 57
12. Marner 57
13. Zibanejad 57
14. Iafallo 57
15. Connor 56
16. Keller 55
17. Bjorkstrand 55
18. Skinner 55
19. K. Miller 54
20. Thomas 54
21. Kaprizov 54
22. Hedman 54
23. P. Suter 54
24. Thompson 53
25. Nelson 53
26. Pelech 53
27. Burns 52
28. Pettersson 52
29. JT Miller 52
30. Gourde 52
31. Suzuki 52
32. DeBrincat 51

To find the adjusted number I multiplied by a factor of the player’s team’s standardized away takeaways (sAT/60) and total takeaways (home & away). The standardized rate removes divisional advantages.

		sAT/60	T/60
1	STL	7.64	8.55
2	OTT	7.41	6.55
3	VGK	7.37	9.22
4	CBJ	7.35	7.55
5	LAK	7.10	5.69
6	SEA	7.08	7.42
7	CAR	7.06	8.06
8	BUF	6.99	6.47
9	PIT	6.98	8.12
10	NYR	6.98	7.62
11	SJS	6.93	9.07
12	NYI	6.86	6.32
13	TOR	6.81	8.55
14	TBL	6.78	6.53
15	MTL	6.52	6.23
16	DET	6.46	5.46
17	CGY	6.42	7.19
18	NJD	6.39	7.98
19	PHI	6.39	5.97
20	ARI	6.33	6.98
21	ANA	6.32	6.36
22	VAN	6.27	6.71
23	FLA	6.15	9.93
24	BOS	6.14	6.68
25	COL	6.11	7.39
26	EDM	6.09	7.82
27	WPG	6.02	7.08
28	MIN	5.90	6.23
29	WSH	5.97	6.66
30	DAL	5.87	5.78
31	CHI	5.74	7.47
32	NSH	5.20	6.82

Think you need to show a couple of examples, like a teams calculation and a players.

 

[RapidKnight]

Think you need to show a couple of examples, like a teams calculation and a players.

The standardized rate is complicated but the result is very similar to the raw away totals. Divisional differences in outputs cause minor changes. To find a players’ total it’s their initial takeaways times their teams standardized rate divided by their team home and away rate.

T • sAT/60 / T/60 = adjusted T

I hope this clarifies my results for you.

I would also like to point out:
1) I made a minor mistake by not removing a teams impact on the league average before finding the teams standardized rate.

2) The divisional adjustments aren’t perfect since I’m assuming each arena is equally biased in tracking their takeaways.

 

[morehockeystats]

2) The divisional adjustments aren’t perfect since I’m assuming each arena is equally biased in tracking their takeaways.

Garrett Hohl made a nice presentation at VANHAC2018 about that issue.

 

[ThankGord]

Pretty interesting. So Florida is reporting over twice as many takeaways as the opponent reports which means Barkov's takeaway stats are inflated? He's obv still an incredible two-way player but that Florida number stands out.

 

[fm]

The standardized rate is complicated but the result is very similar to the raw away totals. Divisional differences in outputs cause minor changes. To find a players’ total it’s their initial takeaways times their teams standardized rate divided by their team home and away rate.

T • sAT/60 / T/60 = adjusted T

I hope this clarifies my results for you.

I would also like to point out:
1) I made a minor mistake by not removing a teams impact on the league average before finding the teams standardized rate.

2) The divisional adjustments aren’t perfect since I’m assuming each arena is equally biased in tracking their takeaways.

 

[Dr Beinfest]

Pretty interesting. So Florida is reporting over twice as many takeaways as the opponent reports which means Barkov's takeaway stats are inflated? He's obv still an incredible two-way player but that Florida number stands out.

Hard to come to that conclusion immediately. Florida was disgustingly good at home. 2.5x the losses on the road. Plus, I mean, they did finish in first place. It’s somewhat believable even if statistically anomalous.

 

[Alwalys]

.

 

[pabst blue ribbon]

There is a "home" bias and an "arena" bias in this stat (as well as in giveaways).

Hit totals suffer from the same thing as well

 

[Hynh]

There is a "home" bias and an "arena" bias in this stat (as well as in giveaways).

Giveaways are funny since they usually punish the home team. For example, Edmonton has recorded 10,470 giveaways at home all time and only 4,802 giveaways on the road.

 

[Naster Okan]

I feel like the definition needs to be a very concrete black and white description. It has a very broad bias. It should only count when it’s directly taken off another’s stick then because any pass or chip that is giving possession away should be deemed a give away.