clock menu more-arrow no yes mobile

Filed under:

Revisiting a theme: Expected Win Percentages

On Feb 28th, I posted a look at the Leafs upcoming schedule and how we could expect them to do. The basic premise was to make use of the log5 method first devised by Bill James, and the home and away goal differentials via the pythagenpuck win probabilities model to assess the likelihood of the Leafs winning or losing individual games against their competition.

There's also been a recent posting by Garrett Baumann at MLHS that delves into the relationship between Goals For, Goals Against, and Expected Point Percentages in a bit of depth, so you can check it out here if you're interested. If you're intrigued, this is generally how most record prediction is done in sports analysis at this stage of things, though other factors obviously impact upon results.

Well, we've gone through 7 games since that posting, and the predicted results - while not exact game to game - are right on point as the Leafs predicted record was 1-5-1, and the Leafs have gone... 1-5-1. The games that swung things were the unexpected regulation loss to Florida, and slight boosts in Montreal from an OT loss to a win, and in Philadelphia from a regulation to an OT loss.

Either way, the predictive value seems to be fairly accurate at this late stage of the year. The benefit of using this method is, as the season moves on, we have more data to refine the predictions. So let's do so shall we?

GP DATE H/A OPP Prob on Feb 28th Prob Now W-L-OT
70 Tue, Mar 13, 2012 @ Florida Panthers 0.485 0.461 0-1-0
71 Thu, Mar 15, 2012 @ Tampa Bay Lightning 0.433 0.440 0-2-0
72 Sat, Mar 17, 2012 @ Ottawa Senators 0.523 0.505 0-2-1
73 Mon, Mar 19, 2012 @ Boston Bruins 0.314 0.293 0-3-1
74 Tue, Mar 20, 2012 New York Islanders 0.642 0.676 1-3-1
75 Fri, Mar 23, 2012 @ New Jersey Devils 0.501 0.432 1-4-1
76 Sat, Mar 24, 2012 New York Rangers 0.391 0.407 1-5-1
77 Tue, Mar 27, 2012 Carolina Hurricanes 0.607 0.599 2-5-1
78 Thu, Mar 29, 2012 Philadelphia Flyers 0.446 0.439 2-6-1
79 Sat, Mar 31, 2012 Buffalo Sabres 0.705 0.694 3-6-1
80 Tue, Apr 3, 2012 @ Buffalo Sabres 0.432 0.405 3-7-1
81 Thu, Apr 5, 2012 Tampa Bay Lightning 0.697 0.712 4-7-1
82 Sat, Apr 7, 2012 @ Montreal Canadiens 0.475 0.473 4-8-1

As in the previous iteration I applied the log5 formula to determine a win probability, and for probabilities between .475 and .525 I considered the game an OTL for the sake of balancing the point results.

Well obviously, the Leafs recent downturn has resulted in the likelihood that a number of games that were previously predicted to be overtime or shoot out losses will now likely be regulation defeats. As a result this method estimates that the team could end the season with 77 points on the year, bringing them into lottery status, possibly with the 28th best record in the NHL.

All of this being said, it should be pointed out that the Leafs downturn recently has lowered their numbers drastically, and perhaps below where they should actually be based on the quality of the team. Either way, it'll be interesting to see how this all pans out. The only way any of this is likely to shift drastically is if the Leafs suddenly improve defensively or offensively... or even less likely - both.

*Please note that these predictions are purely speculative in nature, and the actual results could vary drastically from reality. No liability is assumed should the reader choose to gamble on their basis.*