USATSI 7514294 154511096 lowres


The Contrarian takes aim at so called "advanced statistics."

My article this week is not going to focus on any one single piece but on the overall assumptions and perceptions of statistics. Especially those that are considered advanced stats.

 

Accuracy and Importance


Some statistical categories are more important and thus are very accurately kept by the league. Goals, assists, penalty minutes and indirectly plus/minus are on the official game score sheets.

 

If something was missing or not accurate there would be complaints from many.

 

Other categories, blocked shots, face-off wins and hits for example can be found on other secondary reports. These stats are compiled but are not entirely necessary or even accurate. There is an element of subjectivity to them.

 

In a subsection of an article that Steve Simmons wrote called Stats Sometimes Lie he explains how an NHL GM turned down a group selling advanced stats analytics because the compiled data could vary from city to city. “What’s a giveaway in Toronto isn’t a giveaway in Columbus or New York, for example”.

 

The GM said to Mr. Simmons that they compile their own stats so they could trust the consistency of the data.

 

To further the point about the importance of these categories, can you see a player chase down an office official to complain that he should have been credited with an extra shot in the middle of the second period or a visiting general manager argue that the home team gets credited with more hits than his team?

 

They do not care because the important stats are accurate. Fantasy owners in leagues that use those categories are the ones that would care but as long as they got something then that is good enough. The illusion remains intact. Corsi and Fenwick advanced stat calculations are built on a weak foundation and do not even know it. See article by Tyler Dellow titled "Advanced stats already telling in young season".

 

What some fantasy geek should do is figure out which cities mark down more hits, blocked shots, etc. and if there is an additional bias towards the home team.

 

The Past and Forecasting the Future

 

We use statistics to back up our forecasts. It is logical to defend your stated position but by no means does it mean that we are guaranteed of the future outcome. These are our opinions and best guesses regardless of the values backing up the claim. If anyone of us knew for certain what will happen then those special people would be rich.

 

It is when people, including myself, say things like “Player X got off to an unexpectedly great start to the season so expect him to slow down because the law of averages will bring his numbers back to reality.”

 

It sounds right but it is not. This has a name and it is called the Gambler’s Fallacy. Take a fair coin and flip it hypothetically five times. Assume it comes up heads each of those five times.

 

On the sixth toss is it more or less likely to turn up heads again? The answer is neither, the odds remain the same. A head has a 50% chance of being flipped as does a tail on the next flip. Strictly, the past results do not influence the future ones yet in our heads we make the bad assumption that because there was a run of heads that means that via the law of averages that the next toss would produce a tail. There is no real reason behind the expectation.

 

Unlike justifications similar to, “Team Y will start to lose now because a few key players got injured” or “Player X has a difficult time playing on the road and the team is going away for 11 games”. These are based on a change to the situation. The fair coin has been bent in a manner that makes is less likely that a head will be tossed. The predicted result is still not guaranteed though.

 

Stats Do Not Tell the Whole Story

 

One of the things that stats do best is tell the story of the past but even so they miss some things that cannot be quantified.

 

If you have seen the movie Moneyball you know all about Billy Beane and his struggles to implement those advanced stat theories in baseball. He had a few obstacles in his way before he could show that there was some method to his madness.

 

He got all these players that the numbers said would do well but they were losing games at the start of the 2002 season. Why?

 

First thing was the players were not used in the manner they needed to be in order to be effective. So when a player signs with a new club or a coach is fired you do not know how these changes will truly work out until the games are played. The second thing is that there is still an intangible element. Jeremy Giambi is the case in point. His numbers said that he would be valuable to the team. The Oakland A scouts did not like him very much.

 

They said that he was a trouble maker and would be a bad fit. Billy Beane though was looking at his on base percentage and the cost of his contract. The other stuff was irrelevant to him at first. Then as the season played on he noticed exactly what his scouts warned him about. Numbers or not, he traded Jeremy Giambi away. The losses then started to become wins.

 

So how does a stat tell another person about how they will gel with their teammates? How can you effectively predict if players will amplify each other’s strengths and shore up their weaknesses instead of weakening each other and exposing their faults? Sometimes fantasy owners have to go with your gut but be prepared to be wrong too.

 

Info vs. Statistics

 

In another movie, Wall Street, character Gordon Gekko says, “I got 20 other brokers out there, analyzing charts. I don’t need another one.” and then in response to Bud Fox’s pleading, “You want one more chance? Then stop sending me information and start getting me some”. What he is saying here is that raw information is more important than calculated values. Bud Fox later on goes to get inside information which he then uses to make money in stock trades.

 

For fantasy owners, immediately you start thinking about equivalents like finding out if a player is injured, going to be playing on the power play, pair up with Crosby, and so on. Pretty obvious stuff.

 

What about things like knowing a player’s agent? What am I talking about! How could that be of importance to fantasy owners?

 

Take a look at Jay Grossman client list. He has quite a few but of special note, Alexei Morozov, Nikolai Zherdev, Alexander Radulov, Stanislav Chistov and Ilya Kovalchuk.

 

Hmm, would have been nice to know since all these guys left for the KHL. Maybe you would have stayed away from picking them in the first place or considered trading them for other players instead before it was too late.

 

Is that an advanced stat or is it information? More importantly, is there value in knowing. I think so.

 

For what it is worth, Mikhail Grigorenko and Evgeny Kuznetsov are his clients too.

 

Access, Advantage and Innovation

 

Long time ago, access to information was limited. Once a week you checked out the Tuesday stat pages or you got a copy of The Hockey News delivered. Later on in the early 90’s some hockey fan at the University of Hawaii started to collect the nightly box scores and updated an ASCII file with the stats.

 

Now stats are everywhere and more than just the core stats, hence all this advance stats talk.

 

So if all this info is out there and everyone has the ability to access it how do you get the advantage?

 

In the special feature portion of the Moneyball DVD, it explains that the people in baseball have adopted some form of analytical analysis evaluating players now. The advantage that Oakland once had no longer exists because others are using similar tools and they still have more money to spend.

 

That means that teams like Oakland have to search for further innovation. Crunching those same numbers that other teams have access to will not help very much. Innovation brings to light new raw information and exposes gaps.

 

Back to hockey, whether it is video analysis introduced by Roger Neilson, the Soviet Red Army implementing five man units and puck possession play, or the Montreal Canadiens defensive system which was a precursor to the left-wing lock. Innovation is not limited to game stats.

 

Newer examples include, the Vancouver Canucks doing a sleep study on their players, purchasing a hyperbaric chamber to help heal injuries and recently Dallas Eakins on player training data.

 

What do fantasy owners do to get the advantage over their competitors? Please tell me it is not choosing between Corsi, Fenwick or some other advanced stat.

 

Like Gekko did to Fox, I’ll recommend reading the Art of War by Sun Tzu. As Bud states, “All warfare is based on deception. Sun Tzu says, If your enemy is superior, evade him, if angry, irritate him, if equally matched, fight… if not, split and reevaluate.”

 

Now go watch how Billy Beane made deals with his fellow general managers in Moneyball. There is more to it than just advanced statistics.


Write comment
Comments (0)add comment
You must be logged in to post a comment. Please register if you do not have an account yet.

busy