HIRING DELLOW: THE BOTTOM LINE

740px-Bar_chart_mode_01.svg

The emergence and growth of analytics and advanced stats in the NHL has grabbed its share of headlines in recent months and that was the case again today with news the Edmonton Oilers had hired Toronto lawyer and avid fan Tyler Dellow to join their hockey ops department.

Senior columnist Mark Spector of Sportsnet, who, like several members of the MSM in Edmonton has engaged Dellow in debate over the years, wrote what I thought was a balanced and biting piece about that hire today. It’s a column that has since been pulled by Sportsnet.

I’m not sure why it was pulled, but Spector’s piece in many ways summed up my own reaction to the news Dellow, who has one of the sharpest minds and sharpest tongues in the analytics community — he has used both to get his points across over the years — is on the Oiler payroll. I’m going to run two snippets from the item that was spiked – one now, one later.

Wrote Spector: “Whatever your feelings on Tyler Dellow the Internet personality, they are the sidebar today, not the lead. 


“The lead is that a staunch Edmonton Oilers fan who made his resume with some groundbreaking advanced statistical work on his own website, then used Twitter to promote his expertise, has been hired by the Oilers to be their analytics guru.”

You can read the cached item in its entirety here.

BY THE NUMBERS

Dallas Eakins 2

I wrote an item on Oilersnation back on July 24 about analytics and advanced stats after the Toronto Maple Leafs hired Kyle Dubas as an assistant GM. It reads, in part:

The use of advanced stats has come a long way even since I was asked about them in 2008. It’s an evolving field of study with refinements being made on an almost daily basis. There’s a long way to go to sort the meaningful from the meaningless, but teams committed to doing so already and those in the process will have a leg up on stragglers who don’t.

There are a lot of people, many of them right here in Edmonton, doing the kind of work that pushed Dubas into the spotlight with the Maple Leafs this week. The Edmonton Oilers are among the growing number of teams taking advantage of that brain power. There are more hires to come.

That next hire didn’t take long. I feel today, in the wake of the Dellow news, as I did when Dubas was unveiled by the Maple Leafs. There’s no doubt in my mind that there’s real value to be found in advanced statistics as a tool in assessing players, especially regarding long-term trends.

The trick is separating the meaningful from the misleading. That’s a work in progress and one being taken on by a growing number of NHL teams. Teams that don’t explore and employ advanced stats to supplement their hockey ops departments are destined to be left in the dust. There’s information to be had, and every advantage in the big money game the NHL has become matters.

The Oilers have embraced the use of advanced stats in recent months. If they haven’t actually solicited and paid for Dellow’s work, they’ve most certainly been aware of it while employing other people in the analytics community. Coach Dallas Eakins is a believer in the numbers. So, too, GM Craig MacTavish. A team that has missed the playoffs for eight straight seasons as the Oilers have needs absolutely every edge it can get. Dellow has been hired to provide exactly that.

THE BOTTOM LINE

Twitter_bird_logo_2012.svg

One of the distractions for old MSM types like me – although I barely qualify as that anymore – is that Dellow is a polarizing figure. He has taken his share of pokes at a lot of members of the Edmonton media corps, many of them gratuitous, in my opinion. Like Spector, I blocked Dellow on Twitter a long time ago – that’s not a rarity for me – not because of the message he delivered but because of the sometimes condescending way he did it.

He’s also taken some pretty good runs at Edmonton management and coaching over the years, so nobody can accuse Dellow of sweet-talking his way into the job he just landed with the Oilers. He’s bitten them as often as he’s wagged his finger at MSM types slow to see the statistical light he’s been shining on the game for years.

It’ll be interesting to see how Dellow interacts with the hockey people and media folks he’s dealt with from afar in the past. The bottom line for me is that what Dellow does should stand or fall on its own merit. He’ll certainly get the chance to make a difference within Edmonton’s hockey ops department, even if he’ll do so mainly from Toronto. A lot of people are waiting to see how that goes.

For the final word on that, I’ll cite something from the Spector piece that evaporated today.

He knew everything about the game from his spot outside the circle. Now he is inside, with a chance to prove how smart he’s always said he is.
Godspeed, Tyler. This is where preparation and opportunity meet.

Listen to Robin Brownlee Wednesdays and Thursdays from 3 p.m. to 5 p.m. on the Jason Gregor Show on TSN 1260.

  • Max Powers - Team HME Evans

    Hiring Dellow is a good move by the Oilers. What I like the most about it is that they are getting outside the box in their work to improve the team.

    Will Dellow have an immediate and significant impact on the on ice product? Not likely. Will Dellow and his work have input into decisions made by management over the longer term? Yes, undoubtedly. Will this be beneficial? Yes, I believe it will have an impact on player selection relating to trades and acquisitions, potential line combinations and in many other ways.

    As long as his data and input are put to good use, this signing will help the Oilers and their management team and, personalities aside, that is all that matters.

  • Do what Weight did?

    AHHH- A second eye in the sky . Scenario : Thompson phones down to Eakins on what needs to be changed while Dellows disagrees and says something different . Confused , Eakins decides to go with each on next two shifts and his own on third shift . They’ll need another eye in the sky that can put all this info together into one fast enough to figure out the fourth shift . Meanwhile , the assistant coaches will be running around frantically trying to implement and piece it all together . A little info is good , but too much might not be ? We’ll see I guess . Can a coach realistically decipher and use all that info to their advantage during a game or between periods ? Getting to be like the racing forms – the more the info the more confusion , and you still end up losing more often than not .

    • v4ance

      Gee, that swath of fiction almost lives up to the Twilight movies.

      Anything that Dellow suggests will be formulated and recommended long before gametime. Analytics do not tend to support conclusions based on small sample sizes of a period or a few shifts. If Dellow suggests something, it’ll be part of a larger overall strategy, not a small in game tweak.

      Most analytic conclusions will be in the form of probabilities. Rocky might advocate something and it may work out but Dellow would point out that it worked, in spite of the expected trend because we got lucky.

      To use an analogy, if you overeat during one meal, that won’t make you fat but if you consistently have two helpings of dessert, over time the pattern of eating indicates that you’ll gain more weight, unless some other factor (like more exercise or genetics or luck) changes the results.

      If Rocky Thompson recommends using a bigger mauve plate for one steak so that it looks smaller & less appetizing so you’ll eat less that meal, maybe it’ll work. But it doesn’t change the longer term strategies that you should have done to ensure consistent results.

  • Do what Weight did?

    I enjoyed Dellow’s work on advanced stats and I believe it is a good hire by the Oilers. However I did not like his approach to people that questioned his work. Often his reaction was an over reaction. A few times I unfollowed him on Twitter because of his twitter comments. He reminded me of Nancy Grace on TV. That being said, some media personalities have that same persona in their retort to others (no names).

  • Do what Weight did?

    OFF TOPIC : You can still catch last nights game replay on Oiler site . Canada juniors 6-2 over Czechs . Watch Nurse and Chase have a good game along with several other you are familiar with .Even get to see Bennett and Lasar . Tonites and tomorrows game against Russia will also be broadcast live at Tsn and probably Oiler site . Tonites game at 5:00 P.M. I believe , and MacDavid should make his appearance . Last nights game featured a 162lbs. Czech crushing Ekblad with his head down on an offensive rush and putting him out of the game in period 3. Live hockey to entertain you for next couple of days .

  • IHeartHemsky

    Why would anyone not like this hire? He’s a bright, innovative guy in an emerging field. His input will help with player development and acquisitions. Good job MacT.

  • Quicksilver ballet

    Hey, there’s Jeffey calling yet another poster an Idiot.

    Who knew…..

    With school starting up in 3 weeks, it soon will fortunately limit his unsupervised internet time.

  • Gordie Wayne

    it astounds me how many people are against advanced stats. its not just about discovering that “Jose Bautista struggles to hit doubles on a Thirsday afternoon game on turf when playing on the west coast”. Its about finding if that trend is statistically significant. That process is actually quite complicated and requires years of statistical training. Human patterns at the macro levels are so predictable is spooky. Those who haven’t delved into statistics likely do not appreciate this. If you have a large enough sample, over a long enough time you can usually find a way to predict any behaviour with a degree of accuracy that allows real decisions to be made, and the consequences of these decisions to be measurable and tangible.

    • Lowe But Now High Expectations

      I disagree. I think that removed the human element. If this is true then why play the games? Just look at the stats and we should know who’s going to win.

      • look at the stats and you may have a more reliable prediction of who will win more games over the long term, but this loses predictive power on a single game basis…there is simply not enough sample size to predict and no statistician worth their credentials would ever set out to do that or put their name behind it. Statistical methods are designed on purpose to be overly conservative as their results can have huge real world implications on health and safety. When a trend is there, it thus deserves consideration. An AHL team may beat an NHL team some games, but over the entire season an NHL team will win more. That is something we don’t need stats to investigate, as it is so obvious to our brains. But if we were a complete outsider we would need to measure and analyze those two teams in order to prove that. That’s what advance stats does, it looks at things that may be very important but have yet to be, or cannot be, processed by our brains as we sit and eyeball the games. By adding another more computerized eyeball we can attempt to confirm or deny the hypothesis we have and beliefs we hold…in a hope to gain an edge that no one else has. The human element is key, as it forms the beliefs that need to be confirmed or torn apart and new the hypothesis that need to be tested.

  • Lowe But Now High Expectations

    If you’ve watched Moneyball, you know the Oakland A’s basically invented the use of advanced stats. But what the A’s mainly use them for is scouring other teams minor league systems looking for buried gems. There are very few homegrown players on the A’s roster, most have come in trades.

    That leads to the next point, how will this translate to hockey. Can it be used to search the AHL or European leagues to find players. Or is it something only coaches will utilize to determine who takes what faceoff or what d pair should come on the ice.

    • I think where you’ll find the use is in analyzing your own team.

      Before he took it down, Dellow did some breakdowns on face-offs and how teams did after winning/losing draws in the various ends of the ice.

      What originally made him do it was because he saw a team who was very good at generating shots after a face-off loss (Anaheim) and a team that was poor at generating them off a win (Edmonton) and went and looked at what the teams did tactically after the draws.

      Turns out the Oiler face off wins saw a lot of help from the wingers, which allowed them to win the draw, but didn’t put them in good positions to get the puck to the net.

      Conversely, when Anaheim lost a draw, they’d immediately send pressure onto the puck where a team like Toronto would immediately drop into a defensive shell.

      Determining what has happened is where the stats come in. Determining why it has happened is where the video comes in.

    • cabaj

      Thanks for the link.

      I can’t see why SN would pull it.

      It will be interesting to see how Dellow works within an organization. Hopefully he can add some insight to the Oil’s assessment of talent.

  • cabaj

    For the most part, condescension from Dellow has been in response to unsubstantiated criticism of his analysis from those who either do not understand or feel threatened by this new way of evaluating the game. (i.e., Mark Spector)

    Kinda like when a creationist gets all hot and bothered when someone is discussing the merits of natural selection.

    Spector’s article does little more than provide insight into his own bruised ego.

  • I like Dellow – his analysis usually offered a unique perspective and made me think. I even mostly enjoy his acerbic style.

    My concern with the hiring though, is not that mudcrutch is now on the Oilers staff (I like the hire) but that he’s reporting down the wrong channel. To me, the deeper value that analytics should really bring to the table is in player acquisitions and contract negotiations. Who do you pursue on July 1st? Who do you retain and who do you let walk? Are you getting good value on a multiyear deal or should you chase shorter term? Fancy stats works here because you have a broad sample set to underpin the analysis. But this is all GM domain and not coaching.

    I suspect the tipping point that led to Dellow’s hiring was his analysis of the Oiler’s o-zone draws when Gagner and Nuge were on the dot. It was good stuff that hilighted some tactical flaws and probably hit home for Eakins. The analysis worked after the season with some 30 games played before the tactical change and 50 odd games after. Applying fancy stats to tactical changes like this probably aren’t going to work well mid-season due to the lack of sample size. Beyond high level suggestions such as carry vs. dump & chase I worry that the benefit when reporting to Eakins will be somewhat limited. Tactical fancy stat analysis might even be detrimental when you’re relying on small sample sizes (which Tyler seems prone to do when the data supports his opinion and quick to discount when it doesn’t).

    I do like this hire. If MC79 was reporting direct to the GM (like Mehta in NJ or Dubas in TO) I’d love the hire.

    • "Frank the dog"

      To date, all that the Oilers have announced is Dellow’s hiring as a consultant.

      All talk @ who he reports to, or what his specific role is to be, remain speculative.

      Personally, I’d expect MC79 to be utilized in a variety of ways.

  • Maggie the Monkey

    Does this mean Eakins will finally be held accountable by someone in the organization, or was this just an organizational ploy to shut him up? If he is given the liberty to openly criticize the club on behalf of the fans this is a welcome change … why do I doubt that will happen here however?

  • Tyler Dellow only has 24 years to win a cup with the oilers to be more effective than “Saw him good” MSM crowd!

    Dubas in TOR does it in 46 years he will be better than the ballard, MLSE crowd!

    RB:
    “condescending way he did it.”

    Pot!
    Kettle!

    • Gordie Wayne

      “The bottom line for me is that what Dellow does should stand or fall on its own merit.”

      If you choose to ignore my point to fixate on the pot-kettle bit because it better suits your need to drop off a shot at me, that’s up to you.

      By the way, the MSM “saw him good crowd” has NOTHING to do with whether the Oilers win or lose. We observe. We write. We comment. We aren’t part of the hockey ops department.

  • Patrick!

    They looked at team structures of cup winners.

    CHI forward depth; LAK and BOS ZS forwards; dmen and Goalies.

    Edmonton have 5 1st line producers

    Hall; Eberle; Perron; RNH; Purcell

    2-3 2nd line producers

    Pouliot; Yakupov

    2 strong tough ZS Start pairs

    Hendricks-Gordon; Pouiot-Purcell

    6 strong Box protection d resulting in ELITE GA.

    Ference – Petry 1st comp 2.19 EVGA/60

    Nikitin (1.31)-Fayne (1.61) 2 best 2nd comp Box d in league!

    Marincin (2.08) 1st comp 3rd best shot rate in box in league.

    Aulie (1.48 brewer -1.89 Phanuef) 1st comp paired with Petry type.

    TOP 10 Save% inside box

    Scrivens #4

    So Mact has replaced 17 of 23 roster spots in 13 months.

    with structured goal in mind.

    Now he is bringing a person who is using and advancing dynamic play analysis. Pocession; Entry; Box protection; Bench use.

    Quite honestly. Those resistant to the analysis give the impression they are not strong enough in math to understand; driven by english language study (MSM) over science studies.

    Bitching about results from a team that was removed and replaced
    By a proactive GM (75% of roster.) retaining only one Pre Katz pick.(petty) is Silly.

    MacT cleaned house.

    • Are you suggesting that:
      Purcell is going to produce first line numbers,
      Ference and Petry are a number one pair,
      Aulie is a number 5 dman,
      Scrivens is going to be top ten in save %?

      I love what MacT has done as far as changing the roster. I understand the stats fine but think they are useless analysis and theories that add no real value. If those points I listed above are what you are actually trying to say then I you are in dream land. There is no possible way that those four things can happen in the real world.

      • Quicksilver ballet

        Are you suggesting that: Purcell is going to produce first line numbers, Ference and Petry are a number one pair, Aulie is a number 5 dman, Scrivens is going to be top ten in save %?

        ————————————————————————–

        Can’t you let Rick have his moment in the sun? It is summer afterall. Reality will set in soon enough for the Oilers come October.

        Have at’er Ricki.

        • Quicksilver ballet

          Don’t tell me you and I actually agree on something? I didn’t see that coming!

          I am a die hard and I have supported the team through its best and worst days…but I think reality is going to hit a lot of people hard this year when they see very little improvement. Our defense is better but it’s not hard to upgrade on Philip Larsen. We still have a gaping hole at 2C and our goaltending is better but again…how hard is it to look good after what we got from Dubnyk last year? Reality is going to be harsh for a lot of these people.

        • Quicksilver ballet

          Reality will be a cruel mistress. Like the monotonous sting of a million little needle pokes due to type II diabetes caused by a summer spent drinking too much Kool Aid.

  • "Frank the dog"

    There’s only so much you can do with naked eye analysis of player and puck position, shot counts, etc.
    But this is still at the send people to the moon with slide rules stuff. The next generation of accurate, perhaps even real time statistical analysis will come when they use rfid’s in the pucks and in the players skates, place an electronic grid under the ice, and then map id’s of the players and puck against time and location on the ice down to 1/10th of a second and a 1cm location or better.
    That will provide tons and tons of data that will lay the game completely bare to competent analysts like Dellow.
    The investment may be prohibitive right now but the teams that get this right first will gain a huge edge over the others until they catch up.

  • Quicksilver ballet

    Man, 4 months away from the site and it’s still the same bitching about everything to do with the team. Didn’t miss a damn thing!

    The stats are information. Information in and of itself is inconsequential. It’s what you DO with the information that makes a difference. Moneypuck? We won’t know until the season.

  • The only reason the Oilers would hire him is to shut him up and prevent him from exposing all the problems that Lowe and Mac T have created and are going to continue to create with their poor management of this team.

    I doubt they will listen to him at all, because all his previous posts clearly and frankly exposed their incompetence.

    Now the website is down.. Too bad..

    If this team is to really improve, He should be GM.

  • It’s a funny thing Brownlee and Spector blocked Dellow…..(who i agree maybe does/did go too far at times)

    because i stopped following both Brownlee and Spector on twitter because i find their opinions (Spec much moreso) are often blatantly biased with glaring back handed remarks that are often unnecessary.

    Example from Spec’s article: “Dellow was The Voice of analytics online and has spawned protégés. There are others who do what Dellow did when it comes to championing analytics, but nobody did it like he did, slathered in smarm and condescension, with an antagonistic side dish of provocation.”

    • “Example from Spec’s article: “Dellow was The Voice of analytics online and has spawned protégés. There are others who do what Dellow did when it comes to championing analytics, but nobody did it like he did, slathered in smarm and condescension, with an antagonistic side dish of provocation.”

      That’s a terrific turn of phrase by Spector and it decribes Dellow’s chosen approach pretty accurately.

      • when someone is so obviously wrong AND completely arrogant on a consistent and ongoing basis, its pretty easy to not respect them. There are certain people in this world that have a gift for identifying when people need a dose of their own medicine and then delivering it with fantastic efficiency. Dellow gave Spector his dose almost daily, as needed, to the delight of many many people.

        • You’re quick to afford Dellow an almost-free pass, and that’s fine, but let’s call it was it is.

          If your argument is being right and arrogant is a lesser evil than being wrong and arrogant, you might have a point. To suggest Dellow only came across that way when somebody needed “a dose of their own medicine” is fantasy. An arrogant streak is part of Dellow’s character, and that’s OK. It’s an aside to his obvious talents, but it’s there.