Article Index
The Power of Numbers
Page 2
All Pages


Frolik and Marchand are in similar situations, as both saw a decent 2nd quarter production given their allotted ice-time. Given how deep each of their respective teams are, look for these two to trend downwards as the season progresses.

*

King saw a boost in production during the 2nd quarter thanks to a Jeff Carter injury. He really used that to his advantage as he’s seen his average ice-time climb from 14:46 in the first quarter to 15:51 in the second quarter. He’s essentially cemented himself in the Kings’ top-nine. A 0.5 point-per-game rate the rest of the way is definitely achievable.     

 

kj

 

A few hits on this list. Hagelin, Kadri, Kreider, Penner, Vrbata and Backes were the most notable ones as they all took sizeable hits in terms of production in the second quarter. Backes was a career 0.61 point-per-game producer prior to this season, so his hot start of point-per-game production was always going to regress. 0.64 is probably the expected production from here on out. Same could be said for “Pancakes” Penner.

*

Grabovski continues to defy trends. He’s producing at a 0.79 point-per-game clip, despite averaging less than 16 minutes per contest (also only 1:37 on the PP). I’m not sold on him keeping this pace at all, but with that said Ribeiro did post point-per-game numbers last season while garnering just 17:50 per contest, so something must be could be up with that 2nd line center slot in Washington. The key difference Ribeiro appeared in 63.4 percent of the Caps PP chances last season compared to Grabovski’s 29.7 percent rating this campaign.

*

The Bruins duo of Soderberg and Smith are also tipping the scales. Both are seeing a decent portion of PP chances, 40.9 and 33.8 percent respectively, but the 13:37 and 14:14 TOI average will catch up with them sooner or later. Expect the point production to be closer to Soderberg’s 0.56 numbers than Smith’s 0.94 moving forward.

*

A couple of veterans round out our list with Roy and Alfredsson. Roy’s 27 points so far this campaign has been largely been propped up by his 13 PPP. If that Blues’ PP efficiency slumps, so will Roy’s numbers. Alfy’s production is slightly elevated, but given his responsibility, and his experience, the 0.94 point-per-game pace probably isn’t that far-fetched.

 

54

 

Gonchar’s slow start was always going to turn around. So his big boost in production is not surprising at all.

*

There’s nothing special about Kulikov, Goligoski, Timonen or Letang’s 2nd quarters, but at least it appears that the situation is on the way up. They might be building towards a stronger 2nd half.

*

Streit and Markov have seen a decent point production rate, but there’s still plenty more to give in my opinion. Now that the Flyers’ offense has woken up, Streit’s production is following suit. Markov is getting his points chewed up by Subban, if he can manage to reclaim that “top dog” status, his points could really skyrocket.

 

54

 

Garrison, Chara, Josi, Irwin and Gardner have seen some really nice increases in the 2nd quarter. Garrison has really benefited from Edler’s injury and Chara should see more responsibility with the absence of Seidenberg for the remainder of the year. Josi and Irwin both have fairly favourable schedules in the 2nd half, so they should keep pace the rest of the way.

*

Del Zotto and Hamonic are two players who struggled immensely in the 2nd quarter while posting just three and one point respectively. More on them in next week’s column.

 

54

 

Three notable hits in this grouping: Bieksa, Bouwmeester and Leddy all experienced their expected drop offs in production. Bouwmeester’s production is still propped up by the Blues’ out-of-the-ordinary offense, so that could take a tumble if St. Louis runs into scoring troubles.

*

Seabrook continues to rack my brain, but much like what I mentioned about Hossa above, the Hawks could be in trouble if the offense dries up. I would strongly consider moving him in “points only” leagues. His current 55-point pace would clear his career-high of 48 by a fair margin. His current 0.67 point per game rating is well above his career average of 0.41.

*

Hedman has recorded 20 of the Lightning’s 70 points from the blue-line, but is appearing in only 28.1 percent of their PP chances. The puck luck is bouncing his way at the moment, but it probably won’t continue.    

 

54

 

Pretty much the entire list fell into place as a lot of these “hot starters” regressed in the 2nd quarter.

Now there’s been a lot of banter on the forums regarding the “predictable power” of data. I will be the first to accept the fact that when looking at numbers, you are looking at it from a “summative” perspective (it’s already happened) and that there’s absolutely no certainty that the situation will necessarily change for the future. What I would argue is that, numbers tend to “normalize” to some degree and the proof is in the pudding. If you look at my predicted lists above, 57 percent of the data fell in the “correct” direction with definitive margins, and 75 percent of the data falling on or in the “correct” direction. Just like everything in life, there’s always a margin for error, and for every three that I get “right” I’ll miss one. But if you ask me honestly, I’d take a 75 percent efficiency rating any day of the week. It’d certainly hold more water than a lot of the various reasons being tossed around out there.

Hopefully with this article I’ve turned you into a stronger believer into looking more deeply at the numbers and perhaps you can use my findings, to better your position in your fantasy leagues.

 

54

 

54 

 

Want to know who the outliers are? Stay tuned for next week’s article as I go through my greens and red players for the 2nd half.


This is just some of the gold you will find to set yourself up for the second half - pick up the seventh annual Midseason Fantasy Hockey Guide here and seize victory!





Write comment
Comments (4)add comment

number54 said:

number54
Covariance Just inspecting your correlation coefficients tells me you have an ENORMOUS chunk of covariance tied up between your top 3 predictor variables. Remember, R^2 tells you the proportion of total variance explained by a SINGLE predictor; your lowest coefficient (overall TOI) explains 70% of the total variability in scoring. This clearly means that your other (even stronger) predictors MUST covary. Otherwise, you're explaining more than 210% of the variability in scoring, and I'm sure you don't need to be told that this is impossible.

Moreover, the amount of covariance you're seeing is intuitive and expected: it's simply probable that players with more ice time are those who get PPTOI in addition to normal TOI; also, players with more TOI get more SOG; further, players with more TOI will accrue more points even if they are not actually "better" players (P / TOI * TOI = P); finally, players with more SOG are more likely to accrue goals (and thus points). So, while I don't doubt that your predictors are useful, it's misleading to suggest that any of them truly accounts for a full 70% of their performance (as measured by points).

If you're truly interested in discovering how much each of these variables contributes to the overall model, you need to calculate the partial or semi-partial correlation coefficients. Essentially, these are just correlations among residuals (left-overs) once other predictors are accounted for. I'd suggest trying a model where the coutcome variable is P/TOI or P/PPTOI, if you wanted to identify talented players who're buried on the 4th line, or not getting enough PPTOI. That'd be a great way to find a diamond in the rough.
January 07, 2014
Votes: +1

Wrist_Shot said:

Wrist_Shot
Solid Solid Solid Love the article and use of data to extract what "should" happen. Can't wait for the next article Ma!
January 07, 2014
Votes: +0

Dikoi said:

Dikoi
Pure gold Keep them coming, only good managers know how useful this stuff is!
January 06, 2014
Votes: +0

SavageGardener said:

SavageGardener
Incredible analysis What an excellent article that gives you the best chance of picking up on the trend of a player's production - nothing is absolute when it comes to predicting future performance, but this is as good of a model as any! I'll take these "3 out of 4" odds any day... I can back up my wins with solid rationale and not dumb luck, and if it doesn't happen at least I have the consolation that I played my best cards and take solace that it just wasn't to be. Keep writing these, Ryan, and I'd love to hear how your data modelling has helped you finish ahead in your leagues!
January 06, 2014
Votes: +0
You must be logged in to post a comment. Please register if you do not have an account yet.

busy