The three-games-from-the-mid-season mark is probably about acceptable to make some non-binding statistical conclusions about the Caps performance so far. The never-lying numbers tell us that there seems to be a clear correlation between the starting lineups and the final results. Or better to say, it does make a difference whether or not the three Young Guns play on the same line.

Right now the Caps are 21-12-5. Here's the split for the most frequent starting combinations of the first two lines (sorry, bottom six--you're out of this comparison):

Line	W	L	O
8-19-22	7	11	2
8-19-28	9	1	2
14-21-28	4	2	2
21-28-85	3	2	2
14-21-22	7	0	0

 

There is certainly some correlation here as well: in all seven games with 14-21-22 in the starting lineup, 8-19-28 were playing together. However, it seems rather obvious that the Ovi-Backstrom-Semin combination is the most successful line so far. In fact, 19 and 28 starting together (there were two 19-21-28 cases) have even a better output of 11-1-2. On a different note, it looks kind of worrisome that putting Knuble on the first line does not work well. At least it hasn't so far.

We all have heard the endless talks about how fancy 8-19-28 become after a while and so on. I just don't buy it. What looks fancy for someone may be quite natural for somebody else. Was Semin's behind-the-back pass to MP too fancy? Should he have not tried it? And even if sometimes a fancy-shmancy thing does not work, I think the advantage of building even better understanding between the top three players heavily outweighs occasional mishaps.

..and the question is:

Poll

Would you keep 8-19-28 together?

46%

Yes

23 votes

53%

No

26 votes

49 votes | Poll has closed

This plot shows the distribution of votes for several players. Solid lines are the votes from a couple of days after the poll was open for a particular player, and dashed lines are the final votes after the poll was closed (except for Morrisonn, his poll is still open). If the votes were statistically (i.e., independently) distributed, the change from an earlier to an older post would be more or less as it is for Knuble's votes, that is, the peak values would increase the most (around rating of 8 in his case) and moving farther away from the peak, one would see smaller and smaller  increases. This is also the case for Fleischmann's results, and more or less for Backstrom's too. Regretfully, both Green's and Morrisonn's results were twisted by the later voting. For Green, the later votes with low rates ≤ 8 look okay, and so do Morrisonn's votes below 8. As for the higher rates (9 and 10)...well, you can see it yourself...

Neuvirth is the only other victim of misvoting so far, I didn't put his data on the plot to avoid overcrowding. All other closed polls generally look good.

There should be some mechanism to prevent such troubles in the future, say, only registered Rink Rats/Rabbits can vote in specific polls or there's a minimal membership time to allow voting. In any case, this is a problem that must be solved if we want to see unbiased results.

During the last game in Toronto, either Locker or Beninati mentioned that AO does not score much in the 3rd period. Indeed, only 2 of his 21 goals, that is, less than 10%, were scored in the last 20 min. And this is not because he shoots less -- of his 130 shots, 42 (32%) were made in the 3rd, compared to 40 in the 1st and 48 in the 2nd. As a result of such disbalance between shooting and scoring, the percentage drops dramatically from 0.275 to 0.167 to 0.048(!!!).

Let's take a look at the top 10 goal scorers in the league (as of 12/14) and their scoring percentage and TOI:

                         GP  G     1st   2nd   3rd   Av.SHFT    TOI/G   H/G

Marian Gaborik      NYR  29  22   .125  .237  .310   0:50.4     22:12   0.5
Alex Ovechkin       WAS  25  21   .275  .167  .048   1:04.5     21:19   2.7
Dany Heatley        SAN  34  21   .146  .300  .176   0:47.8     20:20   0.9
Patrick Marleau     SAN  34  20   .192  .209  .150   0:46.2     21:23   1.0
Sidney Crosby       PIT  32  20   .179  .174  .158   0:57.6     22:04   0.9
Dustin Penner       EDM  32  18   .129  .219  .194   0:47.7     19:56   0.8
Michael Cammalleri  MON  33  17   .190  .174  .111   0:48.0     19:29   0.3
Steven Stamkos      TAM  32  17   .125  .212  .182   0:46.9     19:32   0.9
Jarome Iginla       CGY  32  17   .194  .156  .161   0:50.8     20:48   1.0
Ilya Kovalchuk      ATL  24  17   .318  .107  .188   1:08.9     21:41   0.4
----------------------------------------------------------------------------
Average                                              0:52.9     20:52

Obviously, no other top-10 scorer has such sharp drop in percentage as Ovi, and only Kovi has longer average shifts. Ovi's shift is almost 12 seconds longer than the average in this table, and this does not feel good. The TOI/G seems not too high for #8 but this stats includes two games when Ovi was ejected. Excluding those games, his TOI/G is 22:43 which becomes the highest in this list. Another component inevitably adding to the fatigue is Ovi's high percentage of slap shots that require more physical effort than any other shot (not that I am complaining about it too much). His ratio of the scored slap shots to the wrist shots is 1.00, while it is 0.67 for Kovi, 0.33 for Marleau and only 0.18 for Gaborik.

Crossing my fingers in hope that the Russian machine never breaks gets tired in the 3rd...

UPDATE: Hits added...

In addition to the average rating, which is obviously calculated taking average over all votes, another useful parameter to look at is the standard deviation of the voting results which is denoted as σ. Basically, it shows how (non)uniform the fans are in their votes -- the larger standard deviation, the wider opinions and votes on a specific player. So, fitting all results as of this morning (06/25/2009) with a normal (Gaussian) distribution that gives both the average value and the standard deviation (and by the way, the fit quality is mostly excellent), we come to the following list ordered by σ:

Player	standard deviation	average rating
Backstrom	.865	8.52
Green	.892	8.58
Semin	.900	7.63
Laich	.914	7.97
Steckel	.982	7.87
Jurcina	1.00	6.30
Ovechkin	1.09	9.69
Alzner	1.13	6.73
Johnson	1.19	6.93
Varlamov	1.19	8.86
Gordon	1.21	6.17
Erskine	1.24	6.58
Bradley	1.26	6.41
Morrisohn	1.27	4.65
Sloan	1.37	7.09
Clark	1.39	2.65
Fehr	1.39	5.48
Kozlov	1.40	4.85
Poti	1.41	5.94
Fleischmann	1.42	5.68
Nylander	1.45	1.84
Brashear	1.46	4.45
Theodore	1.52	4.76
Fedorov	1.53	6.19
Schultz	1.63	5.25
Pothier	1.95	7.68

 

These numbers quite well reflect the spread of opinions and viewpoints on different players. Not surprisingly, the least controversy arises for the three out of four Young Guns + Laich + Steckel. Ovi's 1.09 is likely a reflection of maniacal alien fans trying to distort his results (there are 14 ratings of 1...how could that be?!?..) On the other side, Nylander, Theodore and Schultz have all been a subject of endless discussions which is reflected in their values. The highest σ is for Pothier which clearly is a manifestation of more emotional ratings due to his return from injury rather than a purely technical evaluation of the player's performance. Moreover, Pothier's season was one of the shortest for Caps and this also affected the spread. Finally, a large standard deviation for Fedorov is probably also due to a non-negligible emotional component in evaluation of an aging superstar...

And the very last note, due to quite significant statistics (450 to 950 votes per player) these results are not going to change if more votes are added. In fact, most of the polls are already closed.