Skip to content

On ‘score effects,’ and the challenges of playing with the lead

Apr 24, 2014, 12:33 PM EDT

If you haven’t heard the term “score effects” before, you’re getting a great example of the phenomenon in the Pittsburgh-Columbus series, where it seems the first key to winning is falling behind by two or three goals.

Per Extra Skater, here are how Games 3 and 4 played out in terms of shot attempts:

source:  source:

It’s not hard to see what happened in both games. The team that fell behind was the one that attacked and dominated possession in the latter stages, while the team that took the lead went into a defensive shell.

This is typical in hockey. We see the same sort of thing in football, when a team with the lead employs a prevent defense and allows the opposition to do pretty much anything but score. Hang back. Don’t blitz. Hope the clock runs out.

But is it a good strategy? Or, in other words, as much as it’s intuitive to avoid things like turnovers at the opposition blue line when you’re playing with a lead, is dumping it deep — essentially handing possession back over — and making sure you’re in good defensive position really the smart play?

Another question: even if a coach tells his players to keep attacking and avoid playing scared — and coaches definitely tell their players this at times– how big a psychological challenge is it for players to actually do it? After all, nobody wants to be the guy who makes the boneheaded turnover that costs his team the game. Best to just dump it in. Or if you’re in your own end, chip it out. Nothing wrong with living to fight another day. Right?

Obviously, this is not a new debate in hockey. How to play with the lead has been a consideration since, well, probably since the first lead was taken. What’s different now, though, compared to maybe 10 years ago is there’s all sorts of statistical evidence to show coaches exactly what happens when a team goes into a defensive shell.

And make no mistake, there are NHL coaches and general managers who look at these stats. Not all of them maybe, but definitely some. Take the Minnesota Wild, who reportedly endeavored this season to stop dumping the puck in so much after considering some statistical analysis which concluded that carrying the puck in — even if it brought more risk into the equation — was the optimal way to enter the zone.

“You don’t need to take any chances when you’ve got a lead,” said Pittsburgh forward Craig Adams. “At the same time, you can’t sit back. That tends to give you more problems than if you just stick with your game. We did that a bit [Wednesday], especially in the second period. We were in our end way too much. We still want to get a lead.”

Game 5 goes Saturday in Pittsburgh.

  1. hockey412 - Apr 24, 2014 at 12:49 PM

    GREAT piece. No, it doesn’t make sense, especially for a team built like the Pens (which is built to attack, or at least keep the puck in the offensive zone), to stop attacking. If they want to play a defensive game, there’s going to be some coaching involved that just isn’t there right now. Last night they could only chase the Blue Jackets around as they cycled, hoping for a screw up, and when they got the puck, they coudln’t get it out of the zone without skying it. Pretty uninspired. Same can be said for Columbus when the Pens came back, though they only faltered for a few minutes.

    • sanfranbruinsfan - Apr 24, 2014 at 2:19 PM

      Very interesting article

  2. stakex - Apr 24, 2014 at 12:53 PM

    The only thing a prevent defense does is prevent a team from winning… and that’s true both in hockey and football. Its especially true when you have two teams like Columbus and Pittsburgh, neither of which have a very good defense, or have been getting good goaltending.

  3. statsdolie - Apr 24, 2014 at 1:00 PM

    Every team practices a specific approach built for their roster to give them the best chance to beat an opponent. The thought process behind changing up how you play a game based on how much of a lead you have especially in playoff series utterly baffles me.

  4. 7mantel - Apr 24, 2014 at 1:01 PM

    With that I’ll be rooting for the Pen’s to take the lead in Saturdays game !

  5. patthehockeyfan - Apr 24, 2014 at 2:12 PM

    Fascinating! And, thanks, Brough. It’s a shame that more coaches and GMs don’t employ analytics. One glance at the two graphs, and you know what you’re doing wrong. In other words, a “DUH!” moment.

    This is also a bit of a slap-in-the-face to the commenters to the analytics story a couple of weeks ago who decried using them, saying either that they’re useless or that hockey can’t be broken down into numbers. Nearly everything can be explained with numbers. Ask Leonardo Fibonacci.

    Good post.

  6. muckleflugga - Apr 24, 2014 at 3:20 PM

    analytics are great talking points for the likes of elliotte friedman, after forty minutes or after hours … and by people who don’t play the game or who play sports book

    i’ve known a number of jockeys over the years and to a man, none use a racing form for anything other than trying to outguess another jockey as to race pace, particularly out of the gate … none of them bet, even when on what is believed to be a sure thing … why, because of unpredictable variables encompassed in the term racing luck

    hockey, involving a total of sixteen bodies moving at different speeds with individual purpose chasing a puck on a slick surface, is no different … beyond basic understandings when over-matched teams meet better teams in the first round, hockey is still a guessing game

    statistics are by definition, lagging indicators …

    depending on lagging indicators to predict future outcome makes bookmakers happy … predictive analysis typically relies on interpolation and extrapolation, with missing numbers found through manifestations of linear regression … there’s too much reliance on unknown factors

    variance can be wide and varied, but any of emotion, general team fitness or fitness of individual players, travel, anaerobic and aerobic capacity, in-room dysfunction carried onto ice or even brooding hate and aggression are tiny part of elements that expand exponentially when weighed against the same numbers coming from an opponent

    everyone who could accurately predict the moment when matt cooke fell off the rails again, raise your hand … this while understanding that based on past performance with a reasonable sampling, cooke was due to reoffend …

    now tell us how you are able to predict with accuracy, the extent to which cooke’s behaviour and subsequent suspension will affect the wild … ditto tyson barrie’s absence from the avalanche … tell us how to use analytics to predict in-game relationships between opposing wild and avalanche players as the series intensifies toward closing, given cooke’s idiocy

    take it a step further … tell us how analytics were able to predict a puck would turn on edge, then happily skip-over marc-andré fleury’s stick, leading to the the tying goal even as five teammates were unable to predict the inevitable outcome while joining the other twenty thousand spectators as the play evolved

    missing in the analytics hysteria is inability of proponents to bring in-game numbers, beyond basic statistics such as face-off performance or line match-ups, into play as each evolve while the game progresses … made maddeningly complex by coaching strategies sensitive to abstract sense about game flow and team strength, often disguised in game-play or hidden outright

    analytics provide basic understandings based on a game past, valuable in laying-out basics in strategy for a coming game, and for providing background to changing strategy as the game itself evolves in the moment … nothing else

    slaves to analytics have bronze statues built in their memories

    waving white flags in surrender

    • dawgbone98 - Apr 25, 2014 at 9:56 AM

      If you are expecting analytics to tell you what will happen at what precise second, you are barking up the wrong tree.

      They aren’t even good at telling you what will happen the next game or over a 7 game series.

      This doesn’t make them useless though.

      When Toronto was winning all kinds of games early on in the season, the analytics crowd said it most likely wouldn’t continue. Why? Because teams who get out shot as bad as the Leafs were tend not to have long term success. Banking on career high shooting % and career high sv% from goaltenders tends not to last. So while you can have some success and maybe even prolonged success over many games (if you include the lockout shortened year, the Leafs played about 65 games at a 100+ point pace), analytics predicts that it will eventually crash.

      Now if you are a coach or gm and you see your team is off to a 16-4 start but you are getting outshot by 5 every game, you can be like the Leafs (or the Wild, Avalanche and Stars in previous years) and think you’ve found the secret sauce no one else has discovered (and they all inevitably think that), or you can say “we need to make some adjustments because it’s not likely we are going to keep shooting 11.5% from a group of shooters who typically shoots around 9% and get .940 goaltending from a group of goaltenders who is typically around .920”. You can at least attempt to identify and fix your problems while you are banking points, rather than scratching for answers in the midst of a 3-10 stretch.

      So yes, analytics provides basic understandings of games past, but they also are a good indicator of how games are going to go in the future (not specific games, not a time table, but eventually). Things like Corsi and Fenwick at the 20 game mark have a much better track record at predicting how your next 60 games will go than your win loss record does.

  7. nothanksimdriving123 - Apr 24, 2014 at 3:37 PM

    The Gretzky Oilers had the best plan for protecting a 3-1 lead: double it. Yeah, goalies and their gear were half the size in those days, but the same premise still applies. If you have possession in my zone, I can’t score. The point in the game to ease off the gas pedal is: 00:00.0 of the 3rd period.
    This also applies to effective penalty killing. If you have the option of retaining possession without being seriously challenged, do not give them back the puck by clearing it and rushing to be replaced. If you can skate it up ice to their zone and can keep it down there an extra 5-10 or more seconds, you do not need to be replaced. Make them take it from you and if/when they do, then get replaced on your way back to your zone.

  8. muckleflugga - Apr 24, 2014 at 4:05 PM


    perfect, in a nutshell …

    the russians came to canada in 1972 and gave us all a lesson in the possession game …

    they still do

  9. titansbro - Apr 24, 2014 at 6:53 PM

    You can’t build a lead by sitting back & defending. You also often cannot defend a lead this way. “Do what got you there.”

  10. maalea - Apr 24, 2014 at 8:57 PM

    The race is not to the swift or the battle to the strong. But a wise man once said that is the side to bet.

    Numbers hold true over the long season but a 7 game sample is too small.

Top 10 NHL Player Searches
  1. P. Kessel (1801)
  2. P. Kane (1446)
  3. M. Richards (1301)
  4. P. Datsyuk (1294)
  5. N. Backstrom (1150)