Scoracle: The Update (3/14/2012)

(This article was originally published by me as a guest author at the site called “Surly and Scribe”)

As promised, or perhaps as threatened, I am now fully geeked into this tedious and stupid idea of trying to invent a serviceable predictor for the final season standings, so we can pretend to know who will make it into the playoffs. What a chore, I curse whoever it was that came up with the idea. But wait…It was my idea. Okay then, now I like it again. I am either fully committed, or else I am fully crazy and I should be committed. Anyway, the results are in and we are not gonna like them.

Basic premises: Teams have an average points-gained to points-possible ratio over the season. The points ratio differs significantly between home and road.

Here, first, is a reprint of the 5 teams in contention for the final two playoff spots, 7th and 8th. It shows their own points percentage against their opponent’s points percentage, game by game, in rows separating home and road games. From this, you can see which team is favored in any particular game.


It seems, anecdotally, (meaning without any effort from me to confirm or prove it) that predicting home and road success is largely independent from goals for and against, in that two “defensive” style teams match correctly, but when a “defensive” team (Phoenix, Kings) plays against an “offensive,” high-scoring team (Detroit, Vancouver) the goals scored would favor the offensive minded team, yet would not be accurate as to which team could play its’ own style best and win the game. You could then devise a formula to use goal differential, again separated by road and home, but I haven’t really thought it through and besides, if it’s gonna get complicated I like my way better since I already did so much work. I give a nod here to reader/commenter VagabondJim, in that he mentioned the idea of using goals differential to me and I am hoping that if he has more info he will share it with us in detail so that we can all learn from it.

As I went thru and tracked each team’s remaining games and opponents, I had to compute Vancouver’s home and road percentages. It is of peculiar note, and it must be mentioned, that Vancouver has the same odd percentage of exactly .666 for both home and road games. Not only is that a weird number to arrive at, but doubly so when Vancouver is the only team even close to getting the same number twice. Plus, it’s terrifically annoying for them to be successful in any way, and to have the same proficiency on the road as at home is even more bothersome.

And now, back to the idea of using points-gained percentages to judge team performance and as the basis for a predictor of winning probability. Instead of picking a winner of particular games, or all the games, and adding up points that way, I wanted to use each teams’ average performance of gaining points over the season as the main factor. Here is how I did it.

To predict the final standings, I need to award points from games according to a formula using the teams’ point-getting percentages so far. I took the regulation points available in future games and broke them down by opponent, for example Kings v. Ducks. The Kings home percentage is .559, and the Ducks road percentage is .471. To define how points are to be awarded, I added the two percentages to make a whole, representing each teams proportionate chance of winning the two points. I call this a “weighted total.”

The “weighted total” in our example of the Kings v. the Ducks at Staples is .559 plus .471 = 1.030. The Kings therefore have a Weighted Expectation Percentage in Regulation (WEP-R) of winning the 2 points equal to .559 of 1.030, which is .542. I therefore credit the Kings .542 of 2 points, equaling 1.084 points. Now, of course there are no partial points, but we are counting expectations here; it will become valid by using the same comparison to all teams. So, you have 1.084 points from regulation, added later to the expected total of OT points.

Assigning OT points gets really fun. Approximately 1/4 of all games go to OT, so, if a team had 7 road games left, in those 7 games the expected overtime points will 1/4 of 7, or 1.75 points. Store this fact.

We can’t know which games will go to OT, but we can expect one quarter of them to do it. So, the OT points are spread among all the remaining games. That means: OT = .25 x #games = points available. But how to assign OT points? Same as with regulation points, almost.  I took each individual games’ Weighted Expectation Percentage in Regulation, and computed their average. Following our Kings example, the Kings WEP-R for each remaining home game is .542+.507+.521+.468+.474+.623+.521= 3.656, divided by 7 gives an average Weighted Expectation Percentage in OverTime (WEP-OT) of .522. Next, I took the WEP-OT of .522 and computed it against the likely 1.75 OT points (one fourth of 7 games likely to go to OT) and the result is .914 points. In 7 games, the Kings can be expected to net .914 points from among the likely overtime points to be allotted. It’s all very simple, right? (By now, I expect your answer to be, “Fuck off, X.”) And again, the partial points become valid when used comparatively among all teams.

The WEP-R is created for each game, one at a time. It is different than the WEP-OT because the WEP-OT is the average among many games, from all opponents remaining.

So, here is the indisputable, irrefutable, iron-clad, guaran-damn-teed final standings (reads “maybe”):

7th:      Coyotes

8th:      Sharks

9th:      Flames

10th:    Kings

11th:    Avalanche

Sad, but true. Of course, it seems that every year some team goes out of its’ mind and makes a run, defying the season’s stats and simply refusing to lose. Let’s hope that it’s the Kings, cuz if it isn’t, then the Kings are fighting for one spot from among 4 teams, and the chances get much worse.

For those of you who like to see sausage made, here is the ugly chart in a simple, non-excel layout. What you see in the first line for each team is that team’s home Pts. Percentage against the opponent’s road Pts. Percentage, shown game by game. The second line is the weighted total next to the resulting Weighted Expectation Percentage in Regulation. For example, for the Kings, line 1 shows Kings/Ducks=.559/471. After a comma, the next game is shown, and so on. The second line shows 1030/.542: this is the weighted total of 1.030 and the computed WEP-R of .542. The beginning of the second line shows the OT points formula, containing .25 multiplied by games remaining equaling likely OT points, then the WEP-OT number for that team. The third line shows the WEP-OT calculation added to the total points from home and away games, which is at last added to the exisiting point total as of before play began on March 13.

This entry was posted in Math Predicts Future? Ya Right.. Bookmark the permalink.

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>