The statistics in today's world of sport in general, and in hockey in particular, are abundant. Every movement of the puck gets measured and categorized. From these events numerous metrics have been produced. What originally started with the mere Plus-Minus stat now grew up in sophistications like Corsi and Fenwick, and most recently the baseball stat of VORP (Value Over Replacement Player) has been attempted for ice hockey application.

There are two big problems with these metrics so far:

• They rely too heavily on one single factor - shots to describe the whole game.
• They do not produce any expectation metric, i.e. for a player or a team with the measured metric of value X so far, what should the expectation Y be.

The first problem is glaring. It ignores so many aspects of the game - the coaching style, the dynamic matchups, the value of the shots. They produce a dynamic where any shot attempt is encouraged, however harmless it might be. And then, remember if you want people to do A, but measure them over B, they will invest all their effort into doing B and whatever A comes out it's just a byproduct.

The VORP metric comes close to the solve the second problem, but applying an idea from a sport so distant in nature from ice hockey creates an impression of trying to fit a round bolt into a square hole. Unfortunately, without a good predictive capability a metric is similar to a character from the book "The Long Dark Tea-Time Of The Soul" by Douglas Adams, a girl that was relentlessly and subconsciously quoting yesterday's stock quotes without even having access to outside information. The director of the institution that was researching this capability was complaining how different the financing of the research would be if the girl were to quote tomorrow's stock values, rather than the yesterday's one.

MoreHockeyStats.com takes a wider approach. The creator of the site has a ...chess background, a sport where reliable statistical metrics have existed for many decades. The Buchholz and the Sonneborn-Berger metrics and their close variations have proven themselves to be reliable tools to determine the comparison between close rivals. The Elo rating has been producing reliable expectations for performances.

At first, applications of chess, a highly individualistic discipline, look absurd in hockey. But these applications are performed in environments that are similar to environments in which these metrics are evaluated. For example, we can compare outcomes of games between teams to outcomes of games between chess players. We can use the approach to rate hockey players that act mostly alone, or on metrics that are indicating their individual performances, specifically first and foremost, goaltenders and their saves percentage stat.

Another important aspect of Elo ratings is that they are good in underlining trends. They state that more recent performances value more than the distant ones, even throughout the same short competition, and definitely in spans of weeks and months. We try to hold onto that idea when we produce our metrics for fantasy predictions.

In addition, this site showcases a large number of measured stats for hockey games that have never been summarized before, both to showcase the capabilities of our platform, and to bring the attention of the entire hockey statistics community to them to encourage them to develop more encompassing metrics for evaluation of our beloved game.

We are most willing to share the data we scraped and the software we created, and you are welcome request analysis not provided on our website. We also encourage other hockey stats aficionados to share their work with us to produce even more comprehensive reports and predictions.