I am a firm believer in the work that the fancy stats community is doing in hockey. There seems however, some that want to use them at the exclusion of other analysis. Our eyes can sometimes play tricks on us, just as our fandom does. We are often harshest on our teams players and hopeful they could be like the guy on another team. Often not based in reality. This is where stats provide some clarity. The example that comes to mind is a conversation about a Boston game in which Victor Hedman was minus -5. Yet the same individual saw no issue with Drew Doughty being minus -5 versus the Lightning in Tampa. Doughty plays a preferred style for this particular fan.
Depending on the stats you review (www.extraskater.com or www.stats.hockeyanalysis.com ) both Hedman and Doughty, reside in the top 5 or top 30 in most possession and production measures among defensemen. The fancy data provides some perspective as to what we see on the ice. Neither paints the complete picture however. Possession is still a surrogate of shots which can be deceptive in its own right. Teddy Purcell, for example, is 2nd on the Lightning in possession stats yet he is known league wide as a pass first Winger. Does this allow the defenders and goalie to cheat on him making the inevitable shot less effective or dangerous? A shot was recorded but what is the quality of those shots?
In Soccer they put a stopwatch on possession and assign a percentage of time with the ball to determine who had it the longest. Could this not be done in Hockey? A recent discussion of the analytics in hockey ended up with the expert admitting that Hockey's fancy stats are still at the macro level with an inability to measure the mirco events that effect the game and possession. It revolved around the idea that FO% has little impact on possession. This is true, but not for the reasons that you may think. Possession is about time, whereas a FO is an event that establishes possession. They aren't synonymous.
Lost in the analysis of tracking possession are the events that led to the team gaining possession in the first place. The list of events that gain possession, while not inclusive, includes FO wins, Hits, stick or poke checks, errant passes or simple turnovers. If you look at the four best possession teams in 2013-14, Kings, Hawks, Sharks and Bruins, you will see a variety of playing styles and systems. How much does the system lead to the possession numbers?
Hits are said to have a poor correlation to possession because you cannot hit if you do not possess the puck. However, the top possession Kings led the league in hits, while Boston came in at #12 (http://www.nhl.com/ice/teamstats.htm?fetchKey=20142ALLSAAAll&sort=hits&viewName=realTimeStats) . Perhaps those two teams use hits as a means of establishing possession and their shot totals grow from there.
I end with the plea that we view both the possession stats and the eyes as co-partners in evaluating a players worth to a team and the understanding that teams need role players and not just a whole roster full of possession numbers. There is no good measure for intangibles or work. Look no further than Teddy Purcell as a prime example. He is the leading forward on the Lightning in terms of possession yet is a top candidate among fans to be moved. Why? There is a disconnect. Several fancy stats experts have recently stated that Teddy has value to the Bolts as well as many other teams in trade, especially if the Lightning were to retain some salary. That is a contradiction in terms. A recently seen quote says "Hard work beats talent when talent doesn't work hard." Fancy stats can't yet measure everything, but they sure do fill in the blanks.