I have almost nothing to say about this. It’s virtually beyond mockery. The part that really hits me though, is that only an organization with a rock solid expectation of the local media dealing with stories in an Oiler friendly manner would be insane enough to run a press conference like this. Also, if you’re going to release a statement as vague as “MacTavish isn’t going anywhere” and then take questions, you might want to come up with an answer to “Will he be back as coach?” Plus, the bit where he hammered CHED reminded me of Michael yelling “In my home! In my bedroom, were my wife sleeps! Where my children come and play with their toys. In my home!” in The Godfather, Part II after Fredo set him up.
Also: Katz thinks that the Oilers are in the middle of a playoff race? I guess they don’t have www.sportsclubstats.com in the Bat Cave.
Oh, and what’s Terry Jones’ obsession with speaking directly to Katz? It’s kind of weird. Also worth mentioning: Jones pointed out that Lowe was initially to speak first, followed by MacTavish. One would think that they told MacTavish what was going to be said and MacT indicated he’d much rather not to have to follow that statement.
Enough with that. Craig MacTavish has been getting hammered for his decision to try to tag Teemu Selanne with the illegal stick penalty with 2:11 remaining in a 4-3 game. I’ve been watching hockey for a long time and I’ve never seen a hockey game end like that. It’s a loss that’s going to go into the pantheon of famous Oilers losses, somewhere below the Steve Smith game and G1 of the 2006 Stanley Cup Finals but certainly somewhere above the famous HNiC loss to Los Angeles in 2004. A memorable way to lose a hockey game and, if this is the end of MacT as coach, an unusual capstone to his run as Oilers head coach.
I caught a bit of Bob Stauffer and Dan Tencer after the game and Tencer, oddly, suggested that he figured the reward wasn’t worth the risk. He also suggested he’s got some facility with odds. I can’t reconcile those two statements. As I understood his reasoning, he didn’t think that the risk of the Ducks having an empty net to shoot at and no risk of icing made any sense and he preferred the relative safety of the Ducks having to risk an icing to get the empty net chance.
Tencer’s premise kind of strikes me as insane. If you accept that the goalie was coming out regardless, the question is how much risk is added by the Ducks being able to shoot at the empty net from the wrong side of centre without risk of an icing. Given that they’re going to be down two men and not have time to take aim, I don’t see this as being a serious issue. In any event, in this situation, a coach should be willing to take on more risk in exchange for a higher potential of a goal. The goal against costs you virtually nothing, as a loss is almost certain.
The jey question though is how much offence would be added by the 6 on 4. We don’t have data on 6 on 4 versus 6 on 5. There is, however, considerable data available about 5 v 4 versus 5 v 3. It shows that the effect is considerable. JavaGeek’s data is typical of the numbers I’ve seen: 6 GF/60 at 5 v 4, 23 GF/60 at 5 v 3. It’s an incredible swing. My rule of thumb has always been that going to a 5 v 3 makes you about three times more likely to score a goal, say a 6.5 GF/60 to a 20 GF/60, or thereabouts. You’ll appreciate that those numbers are more conservative. 4 v 3, by the way, slots in between 5 v 3 and 5 v 4, at around 10.5 GF/60 last I looked at this year’s numbers.
I’ve thrown all of this into a chart, so that you can kind of get a sense of what I’m talking about here. Five key numbers are listed there – the probability of goals for at 4 v 5, 5 v 5, 5 v 4, 4 v 3 and 5 v 3.
MacT was faced with four options and six potential outcomes.
- Don’t pull the goalie at all, don’t call the stick measurement and play out the game at 5 v 5.
- Pull the goalie, don’t call the stick measurement and play out the game at 6 v 5.
- Pull the goalie and call for the stick measurement, with two potential outcomes: Ducks penalty or Oilers penalty.
- Don’t pull the goalie and call for the stick measurement, with two potential outcomes: Ducks penalty or Oilers penalty.
It seems pretty clear that the goalie was going to be coming out and, based on the academic research into when that should be done, that’s probably the right call. The decision I’m interested in though is the decision to call for the stick measurement. MacT indicated after the game both that he was pretty sure the stick was illegal and that he thought he had good information but also that he was of the view that he had to be certain before making the call. I’m not so sure that that’s true.
I’ve put together a decision chart that considers how “sure” the coach has to be that he’ll get the call in order for it to make sense. To evaluate this, you need to know three things: 1) the rate at which teams score 6 v 5, the rate at which they score 6 v 4 and the rate at which they score 4 v 5. Those are the potential outcomes. I don’t have solid information about how teams score 6 v 4 versus how they score 6 v 5 but I would suspect that 6 v 5 slots in somewhere below the 6.5 GF/60 5 v 4 rate and above the 2.5 GF/60 5 v 5 GF rate, while 6 v 4 slots in somewhere above the 6.5 GF/60 5 v 4 rate. I set the Oilers SH GF number at .75 GF/60, which is below the league average.
In effect, what I’m asking is what the expected value of the decision to call for the penalty is at various levels of certainty that Selanne’s stick was illegal. Where the probability of a goal, accounting for the risk of a penalty is increased by the decision, the decision to take the risk makes sense.
Assume, for example, that the average team scores 5.0 GF/60 at 6 v 5. Further assume that they score 12.5 GF/60 at 6 v 4 and 0.75 GF/60 at 4 v 5. If you apply Ryder’s poisson method, the probability of a goal in two minutes of 6 v 5 is 15.4%. The probability of a goal in two minutes of 6 v 4 is 34.1%. The probability of a goal in two minutes of 5 v 4 is 2.5%. In order to determine the breakeven point, you need to solve for x*.341+(1-x)*.025=.154, where x equals the chance that you’re right about Selanne’s stick being illegal. As it so happens, in that case, x is 40%. If MacT is 40% sure and makes the call, the Oilers probability of a goal is unchanged by the call.
I ran that calculation for various scoring rates at 6 v 5 and 6 v 4, using a constant scoring rate of 0.75 GF/60 at 4 v 5. Here’s what the chart looks like.
In effect, you can basically decide for yourself what you think the Oilers 6 v 5 scoring rate is, what their 6 v 4 scoring rate is and what degree of certainty MacTavish needed to make that call. I’m inclined to think that the numbers referenced above are something near correct (5.0 GF/60 at 6 v 5, 12.5 GF/60 at 6 v 4) and that he didn’t need to be particularly certain in order for it to be the right call.
One caveat for my chart – I didn’t consider the possibility of scoring a goal at 5 v 5 with the goalie pulled, which would presumably drop the necessary certainty level because I would think that a team is more likely to score a goal at 5 v 5 with the goalie pulled than they are at 4 v 5 with the goalie in and, therefore, the cost of getting it wrong is lower than it would seem. You can run this a number of different ways, and if I was the Oilers employee in charge of putting together information for the coaching staff dealing with the management of the probabilities of various situations, I would, but for my purposes, this seems pretty cut and dry: if MacT was as confident in his information as he seemed to be after the game, he should make the decision again. The issue he should be worrying about here isn’t the decision but, rather, his information gathering systems and whether they are as reliable as he thought they were.
The fact of the matter is, scoring another goal against the Ducks was an unlikely proposition, even with the goalie pulled. The coach’s job is to do things that drive the probability of winning up. The fact that those might lead to the effective loss of the game instantly if the gamble blows up doesn’t matter – if taking the risk has a better expected outcome than not taking the risk, the coach should take the risk. If MacT had been offered the chance to flip a coin to decide the game at that point, and the coin was weighted 80% Ducks and 20% Oilers, taking the coin flip would have been the right thing to do. If he was as comfortable with his information as he seemed to be, gambling on this call was the right thing to do.
If it blows up as it did, it looks terrible and it’s easy for those not faced with the choice to criticize but doing nothing is a decision in and of itself. If he was 80% sure that Selanne’s stick was illegal, a decision to do nothing would not be maximizing the Oilers’ chances of winning the game. Personally, I like a coach who’s willing to risk some bad headlines on the right decision. This decision may well prove to be part of his epitaph as Oilers’ head coach, right beside “Didn’t have Jussi Markkanen on the bench for G1 of the Stanley Cup Finals” but unless someone can point to some flaw in his information gathering, it strikes me as probably being an unimpeachable decision.