In fantasy hockey, we often talk about streaky players. Guys who have a history of going on a run for periods of time, and being relatively worthless in others. Jordan Staal comes to mind. It feels like one time each season he goes on a productive tear and everyone is kicking themselves for not adding him. For the rest of the season though he is languishing on the waiver wire. It turns out this is kind of a thing, out Frozen Tools has a report for it.

This week on Frozen Tool Forensics: Consistency and the Streaky Player.

Before we dive in too far, let’s talk about streaky-ness. Generally, we understand it to be a group of games where a player does the same thing multiple times (i.e. score) followed by groups of games where they do something different (i.e. don’t). In the case of weekly fantasy matchups, we would recognize this by seeing a player put up a week, or a couple of weeks of good point production and then a week or multiple weeks of poor point production. One thing that can definitely exacerbate this is multi-point games. A player with a lot of multi-point games is likely to have more condensed scoring (more points over a small number of games) which makes it more likely they will end up with good production one week and poorer production the next, and therefore feel streakier.

Now on to the reports. When we navigate to the reports page there are a number of reports that we can pull, and honestly, I never noticed the Most Consistent report until now. As of Thursday (1/30) here is the top page for the 2019-20 season.
 


In a revelation that surprises no one, our top scorers lead the pack, both in points per game and consistency markers. Leon Draisaitl, Nathan MacKinnon, Artemi Panarin, and Connor McDavid are four of the five current top scorers in the NHL and are four of the five highest in consistency percentage. We should define quickly that Consistency % is the number of games a player has a point divided by total games played. Given that, it seems pretty logical that the more points a player has the more likely they are to have gotten them across more games. If a guy has 80 points in 80 games, he is likely to have scored in more games than a guy who has 40 points and therefore have a higher consistency percentage.
 


The data points bear that out, and if we plot it, we get a pretty clear trend. 

The trend line indicates that points per game and consistency percentage are related and that generally, players fall into a rough correlation. What about the exceptions though? Those dots that are far from the trend line?

The dots above the line are players who have points in more games than we would have expected given their points per game number. These are players that potentially have fewer multi-point games, but perhaps more single point games. For fantasy purposes, these guys have been more likely to give you a similar number of points, week in and week out.

The dots below the line are players who have a higher point-per-game rate than we would have expected from their consistency percent. These players are more likely to have higher multi-point outings. For fantasy purposes, these guys are more likely to have wild variations in their point production over any given week – regardless of their general point pace. In other words, these are players who feel “streakier” as different small samples of their games are more likely to have wildly different point totals.

One thing to point out here. A lot of streaky-ness might just be the random grouping of points (i.e. a 40-point guy scores in four straight, and then doesn’t for four games), but we can use these data to look at if there are players who have a bit of an extra edge to their streaky-ness.

So who are the most streaky players for 2019-20 by this metric? Well in order to find out I had to figure out how consistent I would have expected them to be based on their points-per-game numbers and find the difference. It involved some equations and details that aren’t precisely necessary to go into here, but just know that the Difference column is like it says: The difference between the percent of games we would have expected that player to get a point in and the percent of games they actually got a point in. A negative difference means we expected that player to point in more games than they did.

Here are the top (bottom?) 10 for 2019-20.
 

 

Name

PTS/G

Consistency %

Expected Consistency %

Difference

1

JOONAS DONSKOI

0.66

36.4

49.3

-12.9

2

MIKA ZIBANEJAD

1.11

60

69.1

-9.1

3

JONATHAN TOEWS

0.86

51

59.2

-8.2

4

OSCAR KLEFBOM

0.57

36.7

44.3

-7.6

5

ALEXANDER RADULOV

0.61

39.1

46.6

-7.5

6

OLIVER BJORKSTRAND

0.71

44.7

51.9

-7.2

7

ANTHONY CIRELLI

0.69

43.8

50.9

-7.1

8

JAMES VAN RIEMSDYK

0.58

38

44.9

-6.9

9

DAVID KREJCI

0.81

50

56.9

-6.9

10

ANDRE BURAKOVSKY

0.72

45.7

52.5

-6.8


Joonas Donskoi tops the list and it isn’t that much of a surprise. He went on a huge binge when he got up on the top line with MacKinnon in the middle of November. He put up 19 points in 15 games with six multi-point outings. In the 14 games since that streak ended, he has two points in 14 games. This is a pretty clear case of deployment changing a player’s production in the short term – in short he went on a scoring streak. On a completely related note, it is not a surprise to see Andre Burakovsky on this list.

Mika Zibanejad is a slightly different case. He was injured for a bit, and the lines have been juggled over the course of the season in New York, but there are no immediately evident rationales for his ranking here. Scanning his game log, we see that he has 14 multi-point games and only eight games with one point (that leaves 14 games with zero points). Each seem to be scattered fairly randomly across the season. At least for 2019-20 so far, Zibanejad is more likely to get multi-point or zero-point games than a one-point game. Is this repeatable though? Is this something that is specific to Zibanejad, or is it just a small sample thing?

I pulled three season’s worth of data (again, a useful feature of Frozen Tools Reports) and filtered it only using players with at least a full season’s worth of games played.
 

 

Name

GP

PTS/G

Consistency %

Expected Consistency

Difference

1

MICHAEL CAMMALLERI

127

0.47

32.3

37.5

-5.2

2

COREY PERRY

221

0.57

38.5

43.6

-5.1

3

TJ BRODIE

279

0.42

29.7

34.2

-4.5

4

ANDREAS JOHNSSON

117

0.53

36.8

41.3

-4.5

5

JEFF CARTER

235

0.61

41.7

45.9

-4.2

6

BRYAN RUST

234

0.62

42.3

46.5

-4.2

7

DYLAN STROME

146

0.66

44.5

48.7

-4.2

8

TRAVIS ZAJAC

272

0.5

35.3

39.4

-4.1

9

ALEXANDER KERFOOT

202

0.52

36.6

40.6

-4.0

10

MIKA ZIBANEJAD

244

0.8

51.6

55.6

-4.0


The scale of the difference is clearly muted over the larger sample, but interestingly Zibanejad still makes the list. The implication is that even over large samples of games, some players still perform differently than expected, and in the case of this list might be streakier than their point paces would imply.

The difference is not huge though. For a lot of the players on this list, we are talking about a max of 13 or so games where we would have expected them to point and they did not. That is not a huge number spread out over three seasons. So while it does seem true that some players group their scoring more than others, even over long time periods, the far more likely explanation for the feeling of player streaky-ness is that strict scoring patterns just don’t happen. It isn’t like a 40-point player literally scores every other game. There are going to have at least 40 games where they don’t score a point and the games where they do are going to be fairly randomly distributed in spurts throughout the course of the season.

So, Jordan Staal. He did not make either of the top ten lists. Over the last three seasons, he ranks 26th.
 

 

Name

GP

PTS/G

Consistency %

Expected Consistency

Difference

26

JORDAN STAAL

254

0.54

39

41.86

-2.86


That means we would have expected Staal to point in about seven more games than he has over the past three seasons. That would imply that his point production is a little bit more condensed than we would expect, but definitely not dramatically so. His streaky-ness appears to just be a product of changes in deployment and the random spurts of a 45-50-point player.

One interesting side note about Staal though is he appears to have been more productive in the second half of his last few seasons. So do with that what you will.

Until next week, thanks for reading.