Wednesday, September 4, 2013

How Advanced Stats Can Help You Win Your Fantasy Hockey League

jordan eberle oilers advanced stats
Unless you're playing with a bunch of neglectful owners (the absentee parents of fantasy hockey) you will probably need to find an extra edge to outsmart your opponents and win your fantasy hockey league.

Usually, that means finding the best sleepers. But with a bunch of knowledgeable opponents who have all done research and checked the Internet's sleeper lists, you're all likely going after the same players. And when you're all going after the same players it means you're going to have to overpay to get some sleepers, which somewhat defeats the purpose of a sleeper. And if you overpay for a sleeper you better be sure your investment pays off.

For the owner looking for that extra edge, that extra bit of knowledge about a player, which can help you decide whether the sleeper is worth it or whether last year's breakout star can repeat, advanced stats can help.


An acronym that stands for nothing, PDO is the combination of a team's save percentage and shooting percentage at even strength when a player is on the ice. This number tends to regress to 100 over time. Accordingly, players that have a very high PDO over 100 are thought to have been lucky, whereas players with a low PDO below 100 are thought to be unlucky.

If one of the players you're scouting has a very low PDO he could be a good bounce-back candidate. If a player has a high PDO that should set off red flags.

Let's look at the two stats that make up PDO in a little more detail.

On-ice save percentage

A goalie's save percentage when a player is on the ice will dramatically affect how good you think that player is defensively. More importantly, for fantasy purposes, how well a goalie plays behind a player will have a big effect on a player's plus/minus.

Over the last five years the average save percentage at even strength has been around .921. Players that were lucky enough to play in front of a goalie with a sky-high save percentage above .921 aren't likely to be as fortunate next season, meaning more goals against and a worse plus/minus. In general, numbers tend to regress to the mean over time. It's incredibly difficult to sustain an elevated performance. It's also difficult to be astronomically bad for prolonged periods of time, even though Vesa Toskala certainly tested that theory. So, players unlucky enough to play in front of a goalie with a rock bottom save percentage are likely to get a better goaltending performance behind them the next season. That means a better plus/minus.

And lest you think a goalie's save percentage can be substantially boosted by premier defensive players, let me stop you right there. The defencemen with the highest on-ice save percentage over the past three seasons are Michael Sauer, Jeff Woywitka, Keith Ballard, Kent Huskins, and Jared Spurgeon. The Norris Trophy should be safe.

Example: Kevin Shattenkirk

Shattenkirk led all full-time NHL defencemen with a .947 on-ice save percentage in 2011-12, which helped him register a nice plus-20 rating. The next year he dropped all the way down to 77th with a still above-average mark of .927. However, a dip of just 2 percentage points meant Shattenkirk's plus/minus dropped 18 goals to plus-2.

On-ice shooting percentage

But plus/minus is such a small part of fantasy hockey. What you really want to know about is whether a player is going to put up points next season. Well, knowing a player's on-ice shooting percentage in one year will help figure out where their assist total is heading the next year.

On-ice shooting percentage captures how well a team scores when a certain player is on the ice. Over the last three years teams have scored on approximately 7.9% of their shots. For players who have an on-ice shooting percent well above this mark you can reasonably expect this number to fall closer to the league average over time. Likewise, players whose on-ice shooting percentage is substantially lower than the average should see a positive regression closer to the mean over time.

Sometimes the bounces are going your way and sometimes you aren't as fortunate. If a team is scoring on a large percentage of shots when a certain player is on the ice you can bet their point total will reflect that fact. The opposite for those whose team fails to convert on their chances.

Example: Valtteri Filppula

In 2011-10, Valtteri Filppula benefitted from a 11.50 on-ice shooting percentage, helping him score a career-high 66 points. He notched 43 assists, 15 more than his previous career high. The bounces didn't quite go Filppula's way the next season, as his on-ice shooting percentage plummeted the other way—all the way to a dismal 5.94%. That contributed big time to his disappointing 17 points in 41 games (which would only be 34 over a full season).

Individual shooting percentage

If you want to know whether a player is going to keep up his goal scoring binge from one season to the next, the first place to look is at his own shooting percentage. Unless your name is Steven Stamkos, it is incredibly hard to maintain an elevated shooting percentage from one year to the next. In 2011-12, 20 full-time players converted at least 16.5% of their shots into goals. How many of those players did so again the next season? Only four. That at least shows it isn't impossible, but it is highly unlikely. If you're betting on Steven Stamkos doing it again, go ahead. Curtis Glencross? Not so much.

But you shouldn't assume a player will shoot close to the league average the next season. Especially in fantasy hockey, that will most likely undervalue a player, since we're talking about the better players in the league. It's not a radical idea to think Sidney Crosby will be a better-than-average goal scorer. In general, you should look at a player's career shooting percentage for an idea of how well they will convert shots to goals next season.

One caveat: Players can avoid a major decline in goal scoring even with a decline in shooting percentage if they are able to boost the amount of shots they take. And this doesn't mean they have to play like Jason Blake and fire weak wristers from the half-boards. It could happen if they've been traded to a better team, or will be playing with better linemates, are given more minutes, or even that they improve and can create more chances for themselves.

Example: Jordan Eberle

After scoring 34 goals in 2011-12, Eberle scored at a pace that would have netted him 27 goals over a full 82-game season in 2012-13. If you drafted Eberle high thinking 2011-12 was the beginning of a breakout for the young Oiler, and that 40 goals was possible, you would have been disappointed. But if you recognized Eberle's shooting percentage for what it was—an outlier—his 16 goals would have been right in line with what you would have expected based on his career shooting percentage and amount of shots he generates.

Contextual Stats

The role a coach gives a player will dramatically affect how productive he can be. Obviously, power play time is a huge factor in who will generate high point totals, but how a player is deployed at even strength will also have a big impact.

You should always be aware of the linemates a player is given, and whether the coach is going to use a player in an offensive or defensive role. Usually, the best offensive players and those you will be focusing on in fantasy hockey are going to be given offensive roles, but there are some examples (e.g., Patrice Bergeron, Pavel Datsyuk) who are deployed like checking centres, even though they are also excellent offensive options.

This doesn't mean you should avoid all players used in defensive roles, because you'd be crazy to ignore a player like Datsyuk, but it means if you're wavering between to players, you should probably pick the one being given easier offensive minutes.

Below are a list of contextual stats that can help you determine a player's role.

Corsi Rel QoC

The quality of a player's competition, as measured by his opponent's Relative Corsi. The higher the number, the harder the competition.

Corsi Rel QoT

The quality of a player's teammates, as measured by his teammates' Relative Corsi. The higher the number, the better the teammates.


The percentage of shifts a player starts in the offensive zone. Players in an offensive role will have a higher percentage of their shifts starting in the offensive zone.

An easy way of determining a player's role is to look at Hockey Abstract's Player Usage Charts.

Example: Mikhail Grabovski

Under Ron Wilson, Grabovski was used as a two-way centre, starting roughly 50% of his shifts in the offensive zone and not being exclusively given the Leafs' shutdown assignments. He responded with a 58-point season and a 51-point season, making him at least a viable backup centre on your fantasy team.

Under Randy Carlyle in 2013, Grabovski's role completely changed, and he was buried in the defensive zone (36.7 OZ%) and given incredible hard defensive duties (1.457 Corsi Rel QoC). He responded with 16 points in 48 games (which works out to 27 over a full season).

Now on Washington, and likely to be deployed like he was under Ron Wilson, Grabovski is more likely to score around 50 points, rather than 25.

Even-strength save percentage

Finding a goalie sleeper is probably the most difficult part of fantasy hockey. Unless you've drafted Henrik Lundqvist or Tuukka Rask, the only two goalies I would considered locks to be elite, figuring out who is going to have a great season is difficult. You're better off scouting teams rather than goalies. Picking a goalie from a powerhouse team with a good defence means your goalie will rack up wins and a low GAA, regardless how well he plays.

But once all the starting goalies from the best teams are gone you're left wondering who from the bad teams will overcome their teammates and post a solid fantasy season. Goalies with above-average save percentages will be able to cut their GAA, even if behind a bad defensive team that gives up a ton of shots, and will give their team a chance to win on most nights.

But save percentage fluctuates wildly from one year to the next, largely because save percentage on the penalty kill has no real stability from year to year. Even-strength save percentage has more stability, and can help you identify the goalies most likely to turn in a strong fantasy performance next season. Looking at multi-year samples is also a better way of determining who is the better goalie, rather than the one who got hot in one season.

You might be surprised to know that since 2008-09, James Reimer has been a better goalie at even strength than Jimmy Howard, Mike Smith, Cam Ward, Craig Anderson, Jonathan Quick, and Carey Price.

Example: Martin Brodeur

In 2011-12, a 39-year-old Martin Brodeur posted a .908 save percentage after looking done the year before (he had a .903 save percentage in 2010-11). Combined with a trip to the Stanley Cup Final in which he posted a .917 save percentage, Brodeur was seemingly a viable starting netminder for fantasy hockey in 2013.

However, his higher 2011-12 save percentage was driven almost entirely by his performance on the penalty kill. Brodeur posted a .911 save percentage when the Devils were short-handed, identical to his .911 mark at even strength. His .911 even strength save percentage in 2011-12 was actually worse than his 2010-11 even-strength save percentage (.912).

If you bet Brodeur would have a hard time maintaining his .911 save percentage on the pentalty kill in 2013, you would be correct. It dropped all the way to .811, dragging down his overall save percentage to .901, the lowest of his career as a starter, even though his performance at even strength was better.

For more information on advanced stats in hockey check out Rob Vollman's Hockey Abstract, which is a great introduction to advanced stats and has additional stats you may want to consider using in your fantasy hockey research.

No comments:

Post a Comment

Related Posts Plugin for WordPress, Blogger...