I call this graph… HBO is killing it.
Click to biggen.
I call this graph… HBO is killing it.
Click to biggen.
Abstract: Based on the analysis of the relationship between common fantasy hockey stats and actual team success, I devise a new fantasy hockey scoring system that leaves out stats that do not have an obvious contribution to success. The result is a simple system that will lend itself to new fantasy players.
A hockey game, like most sports, is won or lost because of one thing: goals. Your opponent can out-hit, out-shoot, and out-skate you, but the only category that matters in the standings is how many times you and the other team put the puck in the net.
In our age of non-stop stats though, this gets lost. Turn on any given game on TV and you will be informed of power play percentage, how many shots each player has blocked, or how long any given player has spent on the ice. We largely fail to realize how much any of these numbers actually leads to goals and ultimately the team points that determine the standings.
This is no different in fantasy hockey, where hits, penalty minutes, and shorthanded points all matter more than perhaps they should. Leagues that use category scoring take one step forward by dividing stats up, but two steps back by giving goals as much weight as faceoff wins—both are important, but they are not equally important.
I wanted to create a scoring system in which stats are weighted based on how often they actually occur relative to each other, leave out all the stats that lack a strong connection to team points, and is simple.
To get a better sense of what we are looking at, let’s start with goalies. A goalie’s job is simple, at least to describe: Stop the puck from going into the net. This is why our new scoring system has just two categories: Goals Allowed and Saves.
Most leagues use goalie wins, shutouts, and losses too. Aren’t those important?
Wins and losses are important, but they are largely beyond a goalie’s control. Like wins and losses for pitchers in baseball, goalies are at the mercy of their team’s offense if they want to earn a win. A goalie who allows two goals on forty shots had a great night regardless of whether his team won or lost, so why should he receive extra points (or have them taken away) because of how his team’s offense (and the other goalie) performed at the other end of the rink?
I’m still not sold, isn’t winning the game the point, regardless of how many goals are allowed? Why is this better than current systems?
The standard goalie points system on ESPN will give a goalie .2 points per save, five points for a win, three for a shutout, and -1 for each goal allowed. This means an average night for a goalie (3 goals and 27 saves) will be worth either 2.4 or 7.4 points. That five-point swing has nothing to do with anything the goalie did (or failed to do). In comparison, an offensive player would get five points for a goal and an assist, so we are giving/docking a goalie the equivalent of a good night on offense for things beyond his control. You might as well flip a coin for his five points.*
Note: It is not quite a coin flip over the whole season. It could be argued that goalies on better teams will get more wins, so a team’s offensive quality should be taken into consideration when drafting a goalie. This is true, but it works against the reason many play fantasy sports, where individual players and not teams are drafted. A player’s team will always affect his stats in some way, but we should work to diminish these effects, not amplify them. Simply put, it is not fair to punish a goalie because his teammates cannot score.
Under ESPN’s settings if a goalie stops all 30 shots they face during a game, they will not only get the six points for their saves (a two-goal night for offensive players), they’ll get five more points for the win, and three more on top of that for the shutout. That’s a 14 point night. A shutout for a goalie is great, but for an offensive player to match that he would need to score four goals and have an assist. There are far more shutouts than there are four goals, one assist games. A good goalie is a valuable asset to a team, but few, if any, are worth more than the top offensive stars in the game.
So what will goalie points be worth?
On an average night an NHL team will score 2.7 goals on 30 shots. Therefore, win or loss, the average goalie performance is three goals allowed (we’ll round up, which will benefit the goalies in the long run) and 27 saves. We can set such a game equal to zero fantasy points, which conveniently means that we can make each save worth one point and each goal allowed worth -9.
That may appear drastic at first, but remember, nothing hurts your chances to win a game more than allowing a goal, and it is after all, the goalie’s sole job. It also means a shutout does not need an extra bonus—without any points taken away, the goalie is going to earn a lot of points. A bad night, say five goals allowed, is not going to be pretty. On the other hand, a goalie on a team that gives up a lot of shots will reap the benefits of having a good game with so many extra saves.
If you don’t understand the intuition behind why goalies stats are what they are, you should probably just stop reading now before the offensive stats make your head explode.
Offensive stats are much more difficult to value for a number of reasons. You can analyze a baseball game batter-by-batter and if you have enough games, you can figure out how much each play contributed to a team’s likelihood to win the game. And those numbers can be used to create a fantasy league with points based on reality. Hockey is much tougher to break down play-by-play. Does a player help his team more by dumping a puck in or trying to pass it to his teammate? There are too many questions than we can currently track to come up with a good answer—what is the score, how much time is left, does his team need a line change, what is the chance the pass is picked off?
What Should Count for Offensive Players?
Before I started handing out points for anything, I wanted to see how strong of a connection there is between team success and the common fantasy stats are (see my detailed breakdown here). Scoring goals, for example, is good. Other than point differential, scoring a goal has stronger correlation to team points than any other common stat. Which is, of course, exactly what you would expect. Goals against are second.
Perhaps more interesting is the lack of connection we find between team points and hits, blocked shots, or penalty minutes—which are all common fantasy stats. This does not mean that taking a penalty is good, but taking more penalties will not automatically make you a bad team. There is also a very weak correlation between blocked shots and goals allowed. Announcers often praise blocking shots—and blocking a shot is better than not blocking it—but the best teams are not blocking shots, they are preventing the other team from getting a shot off in the first place.
Are You Really Saying Blocked Shots and Hits Don’t Help a Team?
Not at all. I am saying that hockey players do things, the value of which cannot be quantified accurately or with exactness. Team strategy plays a huge role in what a player’s stats will look like. If a team never hit anyone, took a penalty, or blocked a shot… well, it would be interesting. At first glance I would not expect such a team to do that well, but it is possible they would be masters of puck control and a success. I would love to see someone try it.
So for this league, we are going to stick with four categories we know have a stronger correlation to a team’s chance of winning: Goals, assists, faceoff wins, and shots.* We will also be docking players for each game they play. There is more explanation on all that in a moment. Before that, let’s talk point values.
Note: Correlation doesn’t mean causation, but I would be shocked if any of these do not cause a team to be more successful.
OK, Let’s Talk About Point Values…
Coming up with point values is not an exact science, but relative to each other they fit pretty well. We made a goal against -9 points for goalies, and we will stick to nine points per goal for offensive players. We will work with that nine-point baseline for everything else.
Scoring is not evenly distributed throughout all of the players. There were almost 18,000 goals and assists in 2013-14, but half of them came from the top 20% of the players. Coincidentally the top 20%, or about 160 players, will be about the size of our fantasy league.
Because our league will be so reliant on goals and assists, it will be forward dominated. This is a big reason why most leagues include blocked shots, hits, and PIMs in the first place: without them, most defensemen would be valueless, which is more or less what they will be in this league. In reality, defensemen are of great importance, but that importance simply does not translate to stats very well. As I mentioned, the best teams are preventing shots from being taken in the first place, not blocking them. But there is no stat that tells us how many shots a player has prevented.
Rather than making up arbitrary values for them,* we are going to leave them out. So throwing out the defensemen leaves us with 12 forwards per game. An average game (3 goals on 30 shots) means that the average game for a forward will consist of.25 goals on 2.5 shots with .425 assists.
Note: By including blocked shots, hits, and PIMs we may actually be skewing their importance in reality. Again, the correlation between these stats and team success is non-existent.
Because there are 1.7 times more assists than goals, we’ll make goals 1.7 times the value of assists: That’s 5.2 points per assist… ok, we’ll make it five points because round numbers. Like goals, shots will retain the same value they have against goalies, which is one point each. There are twice as many faceoffs in an average game (60) than there are shots (30), so we’ll make a faceoff win 0.5 points and a faceoff loss 0.5 points (those numbers may sound off, but that’s shots on goal, not shots attempted).
Remember our average game for goalies was worth zero points, which means that our aforementioned average game for an offensive player should be the same. This means that we need to add our average values: 0.25 goals, 0.425 assists, and 2.5 for shots to get an average game of 3.175 points… er, three because round numbers. (Faceoffs are left out because a 50% game would be average and worth zero points.) Therefore each game played will be worth -3 points.
Because games can have such dramatic swings in points, particularly for goalies, it may not be best to do a head-to-head weekly format with these point values. I prefer roto leagues, as all teams finish based on their point totals rather than who they happened to play on any given week, which can lead to fluky (and frustrating) outcomes.
A team consisting of about 12 offensive players and 2 goalies (the typical NHL team size) should make things simple enough for newer players to grasp and it will mean all of the players will have positive point totals on the year.
This fantasy league would obviously not be right for everyone, but like Reese’s Peanut Butter Cups, there is no right way to fantasy sport. Some defensemen will still be worth having on your team, but they will largely be irrelevant. The positive is that it is simple to understand and the point values are based more in reality than most leagues, which may mean newcomers to fantasy hockey and those with more knowledge of offensive stars will enjoy it more than traditional systems.
Welcome to A Quote and a List, in which a quote and a list are presented, free of commentary, for your consideration.
“The president [Barack Obama] is the most racist president there has ever been in America. He is purposely trying to use race to divide Americans.”
– Ben Stein
And A List… of US Presidents Who Owned Slaves
Lots of people are getting MacBooks, but I’m sticking with a PC because I like to think different.