mc79hockey.com

Tom Benjamin wrote an interesting piece the other day speculating on whether or not the Sabres might be operating with a self imposed cap of $36MM. The idea is that teams want to maximize their income. For some teams, the best way to do that might not be by investing in their on-ice product.

A lot of the commenters in the thread appear to be operating on the assumption that going over a certain figure in spending disqualifies you from receiving revenue sharing. I was under that impression myself until I actually looked at the CBA. It doesnt’ appear that that’s the case. I think that the confusion stems from the fact that the disbursement of funds taken from the players and held in escrow is tied to the amount that a team spends on player salary. This phase of NHL welfare doesn’t take place until after the revenue sharing phase though. I’m not sure how important this phase actually will prove to be - it depends on how much money was escrowed that the league is entitled to keep AND how much money is available after the revenue sharing phase. We’ll have to wait to see how important this phase of proves to be.

Eligibility to receive revenue sharing is determined by criteria set out in 49.3 (b) of the CBA.

Clubs Ineligible to Receive Player Compensation Cost Redistribution Distributions. A club shall be an Ineligible Club for a particular League Year if, for such League Year, it meets any of the following criteria:

(i) If the Club is in the top half of the Master List (i.e., the Club is among the fifteen (15) highest Clubs in terms of its Club Gross Preseason and Regular Season Revenues) for such League Year; or

(ii) If the Club is in a DMA (or the equivalent BBM market) with a value of greater than or equal to 2.5 million households; or
(iii) If the Club has Available Team Player Compensation for the League Year that exceeds the Targeted Team Player Compensation for such League Year.

There’s nothing there that refers to the size of a team’s payroll. It’s all about revenue and market size. 49.1(c) goes on to say that any club that is not an Ineligible Club may be a Recipient Club and may receive a “full share” Distribution, subject to factors related to how they’re doing in improving their revenues.

Taking the DMA criteria first, I think that you can eliminate LA, the Rangers, the Islanders and Chicago. I’m not sure whether there are any other teams eliminated by this - New Jersey, Toronto, Boston and Philadelphia would seem like possibilities as well.

The revenue criteria is a little trickier as the NHL doesn’t disclose that information. We do have loose rankings from the 2003-04 season to work with though, courtesy of the NHLPA’s December 9 proposal. Some of you may remember that they proposed revenue sharing based on team revenues and then had an exhibit appended to the proposal explaining the effect of their revenue sharing proposal on each team. If you tossed the rankings into a spreadsheet, you could figure out where each team ranked in terms of revenue.

I’ve done that, provided their 2003-04 ranks and then made what I think is a pretty educated guess at whether or not a team will be in the top 15 in revenue in 2006-07. Thirteen of these are no brainers, I think. The top ten from 2003-04 look to be pretty safe to repeat in those positions to me with the possible exception of Boston. None of the Canadian teams received any revenue sharing dollars in 2005-06, which leads me to think that Edmonton, Ottawa and Calgary benefitted from the rather large increase that the Canadian dollar has undergone. With all of those teams having had good seasons in 2005-06 and the continued strength of the dollar, I can’t see any reason to expect them to be in the bottom 15 in 2006-07 revenues - they’re in my top 15. That leaves me with two spots and some guesswork. I think that those two will come from a group that includes Los Angeles, San Jose, Carolina and Buffalo.

The third criteria is a bit more complicated. Teams can be eliminated from eligibility for revenue sharing if they have Available Team Player Compensation for the league year that exceeds the Targeted Team Player Compensation for the year. What does this bit of jargon mean?

Available Team Player Compensation is defined in 49.1(b) of the CBA. It’s a simple calculation - it’s the result obtained by multiplying the a club’s gross preseason and regular season revenues for the league year by the percentage of hockey related revenues that the players are supposed to get in the league year. So if the players are entitled to 54% of revenue and the team has gross preseason and regular season revenue of $58MM, the Available Team Player Compensation is $31.32MM.

Targeted Team Player Compensation is defined in 49.1(s) of the CBA. It’s the sum of Targeted Team Player Payroll plus Pro Rata Benefits. Targeted Team Player Payroll is defined as well - see 49(1)(r) of the CBA. The NHL has discretion to set this provided that it’s not less than an amount equal to the salary floor ($28MM for this year) + 25% of the range. It can be no more than the floor + 50% of the range. The determination is based on the total funds available - the more money there is available, the higher the Targeted Team Player Compensation would be.

So the point here is that you can fall below the top 15 in revenue and still be ineligible for revenue sharing dollars if you have a certain amount of revenue. We don’t have enough finely grained financial data to make a call at the margins about who’s going to be affected by this. I would guess that at least some teams will be caught by this but I’ve got no idea who.

Who are the teams who are going to be eligible for revenue sharing? I’d guess that you’re looking at possibly the entire South East Division (we’ll see how Carolina does), Phoenix, Pittsburgh, Nashville, Anaheim and St. Louis. What conclusions should we draw from this? Some of those teams are probably real drags on the league. St. Louis is special because they’re a team who’ve produced revenues in the past I think and they’re at the bottom end of the success cycle at the moment. Washington may be unique as well - I don’t know what their revenues were like historically. Pittsburgh has the arena situation. The other markets though, at some point you have to wonder what the point of the NHL being there is.

I’m a bit of an atheist when it comes to revenue sharing - I don’t know what’s best. On the one hand, if you’re in the business of operating a sports league, not every team can win. You can’t have the good teams without the bad teams. While managerial incompetence can be and is part of the reason that teams are bad, some teams have to be bad. That’s just the way it is. It appears that in most markets, the reaction to bad teams is that many people stop supporting the team by attending games. If bad teams are necessary result of the operation of a league and they will of necessity result in teams taking a beating in terms of revenue, then it seems reasonable to share the revenue created in markets where the team is winning with the markets where the teams are struggling.

Benjamin likes to talk about the idea of teams being very cheap when they’re at the bottom of the success cycle and then becoming expensive as they become more successful. If I understand him correctly, his idea is that as the players become more expensive, the team’s revenues should be growing as well. The problem with this as I see it is that you have markets of different size in the NHL. For a team like Edmonton, maybe their revenues range from $40MM US at the low end of the cycle to $90MM US at the top end of the cycle. In Toronto, the low end might be $90MM and the top end of the cycle might be $150MM. The market is affected by teams with significantly highs and lows on their revenue cycle, which can make sustaining a really good team difficult - if the Oilers put together a good team and the players want to be paid like they’re on a good team in Toronto, Edmonton has problems. Revenue sharing can address this.

The flipside of this is that I do believe in incentives and I’m not necessarily sure that revenue sharing is in the best interests of a hockey fan in a given market in terms of the incentive that it gives the team. If you know that you’re going to get a revenue sharing cheque if your revenues don’t hit a certain level, it affects your willingness to invest more in your on-ice product. In order for it to be rational to invest in the product that you’re putting on the ice, you need to be able to rationally expect greater revenues from doing so than you can achieve by putting the cheapest team possible on the ice.

Consider the example of a team that has $50MM in revenue. The salary cap is $28MM. Assume that the league will provide revenue sharing to bump them to $36MM in available player compensation. That’s an $8MM cheque. They’ve got a chance to grab a player who makes $4MM. In order for this move to make sense, that player has to be expected to produce $6.51MM in revenue.

The table to the right graphs out how this works. All numbers are in millions of dollars (MM$). P means payroll. R means revenue. RS means revenue sharing. RAP means revenue after payroll. This is the number I’m judging the decision on - I’m assuming that signing my hypothetical player doesn’t impose any other costs on me. The first line in the chart shows the RAP for a $28MM team with $50MM in revenue. You can see that it isn’t until revenue exceeds $57MM that this investment in a player produces a greater return than not making the investment. If there’s no revenue sharing, this becomes a smart move to make when the expected return from the player exceeds $4MM, the cost of employing him.

It seems to me that if you want to remove the incentive for teams to make decisions with an eye on the impact of those decisions on revenue sharing, you need to make the revenue sharing cheques as disconnected from factors within the team’s control as possible or try to connect it to factors that encourage investment in revenue growth, as perverse as that sounds. Otherwise, you end up with a form of league subsidized welfare that discourages taking risk to grow revenues. The decision not to tie revenue sharing to payroll is a good one in a certain sense - nothing would deter investment in the on-ice product like penalizing teams for doing so.

I’m sure that there will be a lot of discussion about this over the life of this CBA as we see how it plays out and who the constant recipients of revenue sharing are. It should be noted that you can only receive the maximum in revenue sharing so many times before you start getting docked if you aren’t achieveing certain targets. That will affect the calculations presented here to a certain extent.

To tie this back to the issue of the Sabres and their payroll, I don’t think Buffalo will be driven by a desire for eligiblity to collect revenue sharing because I doubt it’s going to be significant for them in terms of it’s effect on their revenue after payroll. They might have a shot at collecting money from the escrow account based on having a payroll below Targeted Team Payroll but I doubt that it makes economic sense for the Sabres to chase this because they’ve got a good team and their superior expected revenue number probably involves trying to win. There are teams out there though for whom it makes sense to keep the payroll as low as possible because of the revenue sharing. I’ll be taking a look at the payrolls of those teams around the start of the season to see if it looks like they’re functioning with that in mind.

In light of all of the big numbers coming out, there’s been a lot written about the arbitration process recently. Al Maki had a piece in the Globe and Mail today (behind the subscriber wall) vaguely lamenting the unfairness of it all. Mirtle’s written about it. Fenwick’s written about it. It’s an interesting topic.

It’s unfortunate that Maki’s story is behind the suscriber wall because it hits all of the traditional high points that one expects from this topic. I’ve noticed some debate on the internet over what exactly the process is - it was widely reported at the time of the deal that they had switched to final offer arbitration from a system where the arbitrator heard each side and then just chose whatever number he wanted. As Matt noticed though, there’s nothing in the deal stating this. Based on comments in Maki’s article with the same whining we’ve heard from NHL types for the past ten years about how it’s not right that the arbitrator doesn’t have to just pick one of the two numbers, I’d say that we can officially throw dirt on this idea now.

I’m not as inclined to agree with that position as I was before the salary cap. Consider the Devils situation with Gomez in a final offer arbitration environment. There’s a perverse incentive for the Devils to take a long shot at bamboozling the arbitrator. Say that they were offering $3MM and Gomez asked for $7MM. From the Devils perspective, they can’t really afford Gomez above a certain point due to their cap constraints so all that they’d be doing is haggling over his value in a trade that’s likely going to return futures. BY refusing to budge and shooting for the miracle in arbitration, maybe they can keep him or come up with players to move who are less important. It’s a subtle point but I’d argue that the existence of the salary cap might destroy the desire to reach a market value agreement, which is what arbitration is theoretically driving people towards.

The whole idea of market value is pretty contentious as well, if the comments in various threads around the internet are anything to go by. There’s a fair argument that it’s ridiculous to say that market value for a player can be determined by looking at others who have similar statistics as that doesn’t consider the fact that their really is no market. This too is readily answered, I think. The players don’t get to pick where they’re playing - the tradeoff so that they don’t get really badly hurt financially by this is that a fictitious market is created. Is it perfect? No. Is it more fair than the alternative, which is that the player takes what he’s given? It’s hard to say that it isn’t. If this is a legitimate concern, I’d think that there’d be ways to minimize the effect by expanding factors that are considered. I suspect that what this argument really means though is that a team has other players who they want to give their money too (or they just don’t want to spend it). Teams essentially want to be able to own the players AND pay them whatever works best given their own situation. It’s hard to blame the players for not being interested in this.

The high point of Maki’s article from my perspective was his reference to Mike Barnett. Barnett, who basically has his job because he’s an F.O.G. (incidentally, you can’t spell “goof” without F.O.G but none of those letters appear in “competent”) was talking about how it’s unfair that Mike Comrie and Ladislav Nagy are looking for big raises when they play on a non-playoff team. It’s hard to even know how to respond to this. Barnett, who has three lousy backup goalies under contract for next season, apparently can’t wrap his head around the logic of the entire thing. If Comrie and Nagy aren’t being paid market value for their individual contributions, then the Coyotes should invest some money in showing so. Does Comrie play really soft minutes? Is he terrible defensively? Put together the data and show what makes other 30 goal scorers better than him and poor comparables. It might require investment in developing data sources but with the dollars in dispute in arbitration, it’s surely an investment that would pay for itself.

Of course, that investment might reveal something that Barnett, who has a lousy starting goalie under contract for $2MM next year while someone who’s been a superior goalie for the past 3 years or so wanders the NHL begging for a job, doesn’t want to know. Maybe Comrie and Nagy actually made similar contributions to those made by the guys that they’re citing as comparables, guys who are on teams that made the playoffs. What does he have to toss at them then? Their decision to hire Wayne Gretzky’s brother as head of scouting? Their decision to acquire Oleg Saprykin? Their decision to sign Petr Nedved? Their decision to sign Brett Hull?

Of course not. That’s Barnett’s responsibility. The alternative is that the information he developed would suggest a lesser value but that the arbitrator wouldn’t buy it. In that case, if they’re still cheaper than players of comparable value on the free market, his complaint boils down to “I’m not getting enough of a discount” and if they aren’t cheaper than players of comparable value, he ought to trade them or walk away from them and spend his money elsewhere. Of course, if the arbitration system systematically overpays players on teams who miss the playoffs (as is implied by Barnett’s statement) then players on teams who make the playoffs are getting systematically underpaid relative to the market.

My point in all of this is that this complaint about the system is specious. Barnett can’t change the system so he might as well learn how to work it more effectively. Or sign Georges Laraque for two years and $2.4MM with a no trade clause. Either or.

Mike W.’s point over at Mirtle’s about teams not getting the bang for their player investment dollar seems like a good one if you believe that teams deserve the outcome that they get from their drafting. I’ve got serious doubts about that myself. I’d guess that for many teams, they can do everything right and still have years of drafting like the Oilers in the 80’s. I think that the unknowns are so great that it’s hard to say that a team ought to be rewarded for their drafting. Where you draw the line is a matter of personal preference of course - I just don’t know of any valid reason to say that it ought to be later other than that’s the way that it used to be.

In the end, there are only so many dollars available for players because of the cap. There really is little point about getting worked up about arbitration because every large award is just a few dollars taken away from someone else. So long as arbitration decisions continue to reflect the NHL labour market as a whole, I don’t see any grounds on which you can really have a strong grievance with it. You can argue that it should be improved but I think you have to acknowledge the need for arbitration decisions to reflect the type of contracts that players who’ve made similar contributions have received.

Like the swallows returning to Capistrano, it’s inevitable. Every year, Kevin Lowe cuts a deal that has the guys who follow the numbers on the Oilers screaming. There were some mutterings after the Roloson and Pisani deals which I agreed with but it’s harder to criticize the front office on the Jarrett Stoll contract.

RiversQ and speeds of IOF are the two Oilers fans who are most on top of this stuff. The two names I’ve heard RiversQ mention as comparables for Stoll most often were Chuck Kobasew and Pierre-Marc Bouchard. I’ve got serious doubts about using Kobasew as a comparable but Pierre-Marc Bouchard is a reasonable one. I’ve taken a look for other possible comparables and they’re tough to find. Stoll entered the league after signing a contract coming out of the 2002 draft, which was a terrible draft year (Cam F. Ward notwithstanding). Very few players from that year have made any sort of an impact in the NHL to this point in time, which makes finding comparables difficult. Moreover, as I wrote in my piece about the Havlat deal, you can’t make reference to any salary arbitration award issued in 2005-06 or any reference to salary or compensation in a salary arbitration decision other than one under this CBA. This really limited the possibilities for the Oilers in terms of finding other comparables that they could use.

Bouchard is signed for $1.9MM for the upcoming season; Kobasew for $1.2MM. No matter how you examine things, Stoll had a stronger case to put in front of the arbitrator than did Bouchard or Kobasew. In terms of soft factors, Stoll has a decided edge. The Oilers went to the Stanley Cup Finals this year, the Wild missed the playoffs and the Flames had a shameful first round failure that was celebrated by their fans. An obvious edge for Stoll.

In terms of their numbers, it’s hard to find an edge for Bouchard that I’d think an arbitrator would take seriously. First, their TOI for the 2005-06 season:

Stoll was obviously the most important of the three to his team: he not only had more ice time in each possible game state but Kobasew and Bouchard basically didn’t kill any penalties. Stoll’s 203 minutes of short handed time, all other things being equal, should see him earning more than Bouchard and Kobasew as should his edge over them at ES and on the PP. This is probably in large part due to his efficacy in the faceoff circle, something that the arbitrator will also consider.

Kobasew’s EV-/hr number almost defies belief. Only three other Flames managed to come in at worse than 2.01 EV-/hr; Richie Regehr came in at 2.15, Andrew Ference at 2.30 and Matthew Lombardi at 2.33 EV-/hr. You have to wonder how he managed that - it’s unbelievable. Either he had terrible luck when he was on the ice, played tough minutes or he’s just terrible defensively. I have a hard time thinking that he played any kind of quality minutes. In fairness to Kobasew though, it should probably be noted that he scored more than 50% of the ES goals when he was on the ice - the fact that the Flames did so poorly when he was on the ice has to be due at least in part to the fact that the team was made up in large part of players who wish that they had the finish of Todd Marchant.

Stoll and Bouchard look reasonably comparable in terms of their EV+/EV-. Keep in mind though, that Bouchard had far better goaltending behind him than did Stoll. While the quality of goaltending probably doesn’t matter that much from the perspective of who will win in front of the arbitrator, it is relevant to the question of which is the better player. Stoll put up the superior scoring numbers, which I’d imagine to be something that the arbitrator weighs pretty heavily. They were pretty close though - if placed in a neutral context, I’d expect Bouchard to put up the better offensive numbers, although I’d also think that Stoll would get a serious bounce from having real goaltending. I don’t have the math to back this up, but logically, I’d guess that the offensive context that Bouchard played in was closer to the mean than the goaltending context that Stoll played in. My point is that I think Stoll is probably the better ES player at this point but Bouchard is better offensively. I assume that Bouchard didn’t play tough minutes - I actually met him at the draft in 2002 and he was both shorter and more slender than me. I’d imagine he’s bulked up some but he’s hardly going to be the type of player who you want on the ice doing dirty work in his own end.

As for the PP, it should be immediately obvious that Stoll and Bouchard are going to have very similar counting numbers. Bouchard has a slight edge in terms of PPP/60 but Stoll had the edge in ice time. The arbitrator isn’t going to find much there. Personally, I pay attention to the % of events a guy is in on. Stoll was in on a lower percentage of events than were Kobasew and Bouchard. It’s not included in here but Stoll also contributed a lot of second assists - he had 12 to Bouchard’s 11 and 8 first assists to Bouchard’s 13. Stoll was a significantly better goal scorer than Bouchard. If I was looking at the whole thing, I’d be inclined to say that Bouchard is the better PP player but again, that requires a belief that a player who scores a higher percentage of points on a lower scoring team is a more valuable and contributing player than one with a lower percentage of points on a higher scoring team. I doubt that you could sell this to an arbitrator without some compelling evidence, which I’ve never seen. This is probably pretty close to a saw-off.

Looking at it all together, I don’t see the case that Stoll should make less than Bouchard. From there though, it’s a matter of degree. How much more should Stoll make? $100K? $200K? It’s tough to put a finger on exactly. I can understand why the Oilers decided to settle this rather than go to arbitration. They got away with paying him just $300K more. If you take Bouchard as a floor for what he was going to make and figure that he was asking for significantly more, there doesn’t seem to be much of a margin for going to arbitration.

Say the Oilers thought he was worth $2MM (market value, not actual hockey value). Stoll is asking for $2.5MM or so. It has to be tempting to settle, depending on the degree of risk that you perceive in going to arbitration. It’s hard to say in the abstract, without knowledge of the specifics whether or not the risk was there but given the paucity of comparables of his own age range, you start to look at older players with similar stats and roles. Those players are only going to be more expensive, which means that they aren’t great comparables.

In terms of value for the dollar, I don’t necessarily know that Stoll is providing it - I’d guess that Horcoff brings significantly more. In terms of where the market is for the two players identified as his comparables, this contract doesn’t seem to be worth screaming about. He has indisputably better numbers then the other two. With the Bouchard contract being what it is, it’s hard to say that this isn’t fair. Without knowing further specifics of the negotiation, I’m hesitant to condemn Lowe. This isn’t a bad deal relative to the market and it’s certainly not the failure to read the market that the Roloson deal was.

As much as the cards were stacked against Kevin Lowe with the Stoll deal, it’s not always like that when dealing with RFA’s. One guy who should be getting a raise of nothing is Joffrey Lupul. If ever there’s a situation where Lowe has the hammer over a guy, it’s here. Lupul isn’t eligible for arbitration. His qualifying offer has expired, meaning that he can’t accept it and be eligible for arbitration next year. The Oilers have Horcoff, Hemsky, Smyth, Torres, Stoll, Moreau, and Pisani who have played top 9 minutes in the past. There are good forwards out there available at a low price, including Radek Dvorak. Lupul is a scorer who apparently doesn’t bring much defensively - it’s not like replacing the goals of someone who keeps the puck out of his own net as well. The Oilers had a good PP last year without him and can likely cobble a good PP together this year without him, assuming that they get someone signed to play the point on the first unit. That player isn’t going to be Lupul in any event - if anything, he’ll be bumping Horcoff, a good PP player, off of the first unit. Nobody ever makes contract offers to guys in his position. If Lowe can’t get Lupul signed to a deal that he likes, he goes out and buys a forward on a free market where it looks like many teams are about to be unable to spend any more. He’s in an excellent position.

At this point, Lupul does not have much in the way of a best alternative to a negotiated agreement. If he can’t come to a deal with the Oilers, he goes to Europe and almost certainly makes less than the NHL minimum salary. There’s not going to be an offer made for him there. He can’t become eligible for salary arbitration if he doesn’t play in the NHL or in an affiliated league under an SPC. In short, if he can’t come to an agreement with the Oilers, his best alternative is to play in Europe until such time as he becomes an unrestricted free agent.

Lupul has three years of professional experience. He needs one more to qualify for arbitration. If Lowe offered Lupul a four year contract worth a total of $1.9MM, that’s almost certainly better than Lupul’s alternative of spending those years in Europe and then returning to North America as a UFA. Now, Lupul is probably not inclined to accept an offer on those terms and one thinks that there would be some give in Lowe’s position. The job of a GM is a complicated one at this point - determining the threshold here is not simple. At what point does it make more sense for the Oilers to refuse to sign Lupul to a contract as opposed to signing him? There are a lot of moving pieces to consider: the cost of signing a Dvorak, the difference in team performance, if any, associated with the players, the “cost” of signing Lupul to a one year deal and making him eligible for arbitration…it’s a complex equation. I’d venture to say that Lowe is in a far better position than most GM’s dealing with 28 goal scorers are though, given the Oilers circumstances and strength at forward.

NHL GM’s haven’t made particularly good use of leverage when they do have it, so I don’t expect much - I expect that the Lupul contract will be for far money and a shorter term than it ought to be, given the strength of Lowe’s position and the weakness of Lupul’s. I don’t know why NHL GM’s aren’t tougher on this or why owners don’t make a show of publicly backing them: if the EIG announced that they had a club policy of extracting deals that ran until free agency at minimum wage from players who declined to accept their qualifying offers, unless a longer term deal was signed prior to the expiry of the qualifying offer, it would make these negotiations a lot easier on Lowe - he could tell the agent that his hands were tied and that the player would be well advised to accept his qualifier. The player wouldn’t really have anyone to blame except for the PA, who negotiated an agreement giving the clubs this type of power. The leverage is there - it ought to be more frequently used.

The Hemsky deal has gotten finished as well since I first wrote and never got around to posting this. The consensus seems to be that it’s a great, great deal. I have a hard time disagreeing.

If you’re going to spend money, you might as well spend it on elite players. While he’s not an elite ES offensive player yet, he absolutely is on the PP. Hemsky finished 12th on the list of PPP/60 amongst players who spent at least 100 minutes on the ice at PP and who didn’t screw me up by switching teams or numbers. He’s not just out there collecting points off other players either - amongst players on the ice for at least 20 PP goals (and who didn’t screw me by switching teams or numbers) only 23 were in on a higher percentage of those goals than Hemsky. Only four of those players were on the ice for more PP goals than Hemsky. He’s serious business on the PP.

It’s also interesting to me how young Hemsky is. I think, with some evidence, that young players don’t drive the PP. That’s part of the reason I was so dismissive of Rob Schremp, PP solution for Los Oilers in the finals. Of the top 20 this year in PPP/60, only Andrej Meszaros, Jason Spezza, Ryan Getzlaf, Hemsky, Sidney Crosby and the Sedins are under the age of 25. Hmm. On further thought, that may not be statistically significant. I’d venture that Meszaros and Getzlaf weren’t driving the bus though.

My point is this - Hemsky’s really good on the PP. He’s young enough that it’s fair to think that he might get better on the PP. Even if he gets no better, he’s one of the best. This isn’t a case of buying a guy like Michael Nylander, who got a lot of PP time, improved his counting stats because of it but was actually horrible.

Apropos of nothing, only one player had more PP TOI than Ilya Kovalchuk this year - Mikkaa Kiprusoff. Yes, I’m including goalies. Kiprusoff only beat him by 25 minutes too - if Atlanta didn’t have such a good PP last year, Kovalchuk may well have led the league. Possibly the most incredible statistical nugget from the 2005-06 season.

Anyway, back to Hemsky. As I said, he’s not quite as good an offensive player at ES - he finished 100th in the league or so in ESP/60 this year. If he can make the leap at ES, he’s an absolute steal. As it is, he’s likely going to be a bargain over the next few years.

Why does Hemsky make a deal likw this? I’d guess that he wants security - maybe getting your neck slashed with a beer bottle convinces you of the fragility of your career or something. I really am surprised at the deal though. If Hemsky had gone to arbitration, he’d have got at least $3MM for this season. His salary number was only going to go up from there, even if the Oilers had elected a two year term. We’ll see what the breakdown is on the contract but if the deal was anywhere near even, he makes $8MM over the next two years instead of $6MM if he’d gone to arbitration. Would he make at least $2MM more over the last four years if he’d gone the arbitration? I’d have to think so.

The other way to look at it of course is that at the end of this contract, Ales will be 29 and set financially, whether he turns out to be a Peter Forsberg of the wing or he blows out his knee tomorrow. If he’s Peter Forsberg, he’s going to end up significantly richer. If his knee explodes, he’s have to satisfy himself with just being obscenely rich. He’s now hedged against the possibility that it all goes wrong. If Hemsky is a lot more risk adverse financially than he is when passing the puck, this deal makes more sense from his perspective. If I was him though, I’d probably not have hedged this risk but I may well have greater tolerance for risk.

Interesting thing from the Oilers perspective too is the way that the deal unfolds. He starts out at $3.5MM and ends up at $5MM. If the Oilers want to trade Hemsky at some point, he’s going to be extremely enticing to some of the teams who are tight to the cap but flush with cash, as you’d have to figure that the odds are pretty good Hemsky is a value proposition to begin with towards the end of this contract plus he’s going to come with a lower cap number than dollar figure. It’s a nice looking deal that way as well.

Lowe’s taken some well deserved heat in the past for contracts and he didn’t get off to a fantastic start in the off-season with Dwayne Roloson and Fernando Pisani. He and his guys have done quite well with their RFA’s this year though - Horcoff, Hemsky and Stoll are all signed to good deals. It’s going to pay dividends at some point, either when the rest of the arbitration figures are in or during the regular season when he needs to make some moves. It’s only fair to give these guys some credit when they do some nice work.

Inspired by Matt’s excellent post on stats, Vic’s good comment to that post about posting macro code and a plaintive email from Andy admitting that he has no idea what macros are and suggesting a post on the topic, I figured I’d post the most important spreadsheet that I’ve got for those who are interested in at least one way of sucking data off of NHL.com.

I have to admit, I got a bit of a laugh from Matt and Vic in that post. First from Matt:

It amounts to reading every game sheet and marking it down; he has software that he has made/modified to do this for him, but he still has to collect the game sheets and run his software on each one, not a task that’s overly speedy or that any jackass can duplicate in 5 minutes of spare time.

Then Vic:

I’m not a programmer BTW, typically I just use “Record Macro” to do something for one game, then tweak the code to make it loop through and repeat the process for all 1260 games.

Software? “Record Macro”? Please. No, what you need to do is have a brother (Happy Birthday Colin!) who’s much smarter than you and a good read on the extent to which you can impose on him to help you do stuff like this in terms of setting you up with some tools even though all interest in the game was beaten out of him by being dragged along to watch his siblings play hockey games in small towns at a young age. That way, you can avoid any of the tedious learning of what do, although I’ve grudingly started to make concessions on that front (teaching myself MySQL).

Anyway, a few years back, I was puzzling over some of this stuff and I emailed him for some assistance. He cranked out a spreadsheet for me that had a macro in it which I’ve added to the Files section of the site. If you’re interested download it and I’m going to talk through the macro as I understand it. You may get a warning about it having a macro when you open it; don’t worry about it, it’s a safe macro.

Once you open the thing up in Excel, you’ll see two buttons. Ignore ‘em for the time being. To take a look at the script, click on “Tools”, go down to “Macro” and then open up “Virtual Basic Editor”. Open up Module1 for NHLWorksheet in the Project window on the left hand side. You’ll see a script. I’ll explain the components of it here as best I can.

The basic idea of the thing is that it pulls the data off of the NHL.com site into an Excel file that you can then play with to your heart’s delight. Key bits of code:

For i = 1 To 1230

The sheet is set up to automatically grab all of the games. I think it’s set up to repeat by inserting a different game number over and over. If you want to just grab a couple of games to see if you’re getting the right thing or if it’s set up properly, change 1230 to 3 or something; that way you’re not stuck with some script hammering away and pulling the wrong thing.

findCleanRow = Range(”A65536″).End(xlUp).Offset(1, 0).Row

For reasons beyond my understanding, sometimes the data from a specific thing on the NHL site will spray sideways across my Excel sheet. I usually find that if I change the first number after “Offset(” it solves the problem. I assume it’s because some of the tables are different lengths, depending on the number of events of a particular type. Again, trial and error is key here.

URL = “http://www.nhl.com/scores/htmlreports/20052006/GS02″

Let me explain the NHL’s organization system of their game sheets. First of all, there’s everything before the “/GS02″ part. This changes from season to season. If you want data from a year other than 2005-06, go open a game report from NHL.com for that year and copy everything in the URL. Cut off everything after GS02 and replace this line. For instance, for 2003-04, it’s “http://www.nhl.com/scores/htmlreports/20032004/GS02″. Don’t assume that different years work the same - the NHL does baffling stuff with their system of organization.

The “GS” part means that you want a gamesheet. If you want something else, there’s a different code. ES is the event sheet, which tracks TOI, shifts, shots, hits etc. FC is faceoff data. PL is play by play (where I get shot distances).

The “02″ apparently indicates which season you’re referring to. “03″ is playoffs and I would guess that “01″ is pre-season. The script then plugs in the appropriate game number. I think, but cannot confirm, that that part happens magically.

tableNumber = 2

Last key piece. Pages appear to be made up of tables and different tables have different numbers. If you take a look at this gamesheet it becomes obvious. Depending on what you want to pull out of it, you need to play with this. I’ve got no list as to what table corresponds to what - she’s a trial and error process.

There are other hurdles too. The times are a pain in the ass - they come in in some weird format. In order to change them to time with a decimal, I usually copy them into a different spreadsheet, save it as a .csv file and then open it in Notebook, copy the text and past it into Word where I do a “Convert text to table” which gives me the minutes in one column and seconds in another. I’m strongly suspicious that there’s a better way to do this.

Anyway, that’s how I suck out the data. Ask questions in the commments; no guarantee I can answer them but I’m sure someone here can.

Yep. This was Andy Grabia’s solution for the PP. Big thanks to Loxy, both for alerting me to the existence of the MySpace and having the class herself to refrain from throwing public light on the matter. I’m sure that Marty Reasoner would never do this.

Deadspin is on the Garth Snow story. This is good.

The look on Snow’s face in the second picture absolutely kills me.

Well, so much for the Islanders brief flirtation with credibility. Neil Smith is gone after just six weeks, replaced by…Garth Snow. Yes, the former goalie. Charles Wang ought to be adopted by the left wing as the poster boy for the argument that the wealthy don’t earn their money on merit and have no moral case against severe taxation of wealth and excess income. There is just no way that a rational person hires Garth Snow as a GM with no experience in an NHL front office or business. The only way that this makes any sense whatsoever is if Wang has some sort of Trading Places style bet with Milbury that Snow can achieve Smith’s level of competence and achievement as a GM given the right circumstances. Presumably they’ll announce shortly that they’ve signed Smith as a backup goalie. It’s unclear to me whether or not a topless Jaimie Lee Curtis would factor into this scenario. Outside of the possibility I’ve just outlined, this is completely insane. It narrowly tops my previous favourite story of the summer, the Carl Monday/Mike Cooper affair.

Five or six years from now, the NHL and NHLPA will be closing in another labour fight. When the league is bitching and moaning about how they just can’t seem to make a profit, keep in mind that these are the same people who place people with no business qualifications or experience in charge of divisions of the company (in this case, the Islanders hockey division) that have budgets of $40MM or so. It’s even worse because the success of the Islanders hockey division will have profound implications for the success/failure of the other divisions of the company. You know why you don’t hear about companies pulling guys off the line and putting them in charge of vitally important divisions without any training or experience? Because it’s an appallingly stupid thing to do.

I’d usually be willing to cut Snow some slack on the fact that he has no training or experience but I think his poor judgment is demonstrated by the very fact that he thought taking this job was a good idea. Wang is nuts. Milbury is notable not only for the fact that he’s been the GM most consistently whipped on trades in the past 10 years but also for the fact that he’s no whiz when it comes to contracts. He also appears to have Wang’s ear. Whatever you think of Ted Nolan’s coaching ability, the man is a walking controversy. Pat LaFontaine is there, apparently just hoping that someone will tell him stories about the Islanders back when he was on the team as he no longer remembers any of them. I mean…what possible factors could lead Snow to consider this to be a good situation?

Absolutely insane. These people deserve to lose buckets of money.

Update: According to TSN, LaFontaine resigned after all this went down. Wang should take the Solomonic step of hiring him as GM and firing everyone else now that he’s shown the only good judgment exhibited today.

Further Update: Upon further contemplation, I’m shocked and disappointed that the Isles press release had the boring title “Garth Snow Named GM” as opposed to “Party on Wang! Party on Garth!” If you’re going to make this move, you should at least have the sense of humour to acknowledge the absurdity of the whole thing.

I’m taking a look at the fantastic hockeygoalies.org for a follow up to my post on goaltending replacement levels. He’s got goaltending save percentages dating back further than I’ve seen. Check out Fuhr and Moog during their time with the Oilers. They played behind the same defence and yet Moog put up numbers that matched Fuhr. Older fans might know better than me if Fuhr got the tough teams or something but I’ve looked at the back half of their careers pretty closely and Moog has an edge over Fuhr, IMO. Funny that the one guy is considered an all-time great and the other is best remembered as his backup.

$10.8MM/3 years for a guy who was probably a top 20 NHL centreman last year? That’s a good contract, I’d say. Good work by the Oilers front office. Evidence that Horcoff likes it here as well - if he repeated this year next season, he was heading for a much larger payday as a UFA.

I’ve been looking at my old Baseball Prospectus annuals and decided to take a kick at something that Keith Woolner wrote about in the 2002 edition: establishing a replacement level. This is a real area of interest to me and one that should be of interest to anyone who’s enough of a hockey fan to follow the business side of the game. For now, I’m focused on goalies but it’s a concept that applies equally to skaters.

I’ve mentioned the concept here before; in his 2002 article in Baseball Prospectus on the subject, Woolner came up with the following definition of replacement level: “Replacement level is the expected level of performance that a major-league team will receive from one or more of the best available players who substitute for a suddenly unavailable starting player at the same position, and who can be obtained with minimal expenditure of team resources.” Those of you who don’t troll EBay looking for old baseball saber books may find an article on Woolner’s personal site about the topic to be of use.

It’s important to know for a number of reasons. First, it gives you an idea of where to spend the money. If there were accurate measurements like this for all players in the NHL, I strongly suspect that we’d see that money isn’t being spent very efficiently by NHL GM’s. Big bucks on marginally better defencemen, paying for goalies who aren’t that great but won Stanley Cups, paying players who aren’t particularly skilled on the PP but who get a lot of time there and have good counting stats as a result…my belief is that there’s a lot of inefficiency. Second and related to the first reason, it would allow for a clearer examination of some of hockey’s truths, like the value of a #1 defenceman etc. It’s important stuff if you’re seriously interested in following the sport and even more so if you’re charged with putting a decent team on the ice.

Prior to his discussion of where he thought replacement level was appropriately set, Woolner examined a number of factors that would influence replacement level. He ran a computer simulation to look at the impact of imperfect information. With perfect information, you won’t have any starters worse than replacement level; with imperfect information, you will as teams make decisions that turn out to be wrong.

Woolner also cited reasons other than imperfect information that might lead to teams using players who were below replacement level, including poor decision making (Tommy Salo is probably an example of this for his last two seasons in Edmonton where he had to be getting starts because of his contract), strategic reasons beyond winning now, injury (this may well be a serious problem in Edmonton with skaters), alternate beliefs about replacement level (Woolner suggested this would occur when management believed available replacements were significantly worse than they actually are; this too might have been an explanation for Tommy Salo), defensive performance (an issue in baseball where the replacement level analysis to this point is focused on offensive performance; hockey analogues would be the goalie’s puckhandling, impact on shot quality, impact on shot volume and offensive contribution) and sample size (possibly an explanation for Jussi; I have to think that Conklin had more serious problems).

In setting his replacement level for various positions in baseball, Woolner looked at the ways in which the performance of starters differed from the performance of bench players. I’ve produced a similar chart for goalies based on save percentage. My definition of starter isn’t perfect: if the NHL had X teams, I took the X goalies who saw the most shots over the course of the season. For 2003-04 for instance, it was the 30 goalies who saw the most shots; for 1988-89, the 21 who saw the most shots. The rest of the goalies were labeled backups.

The chart at left graphs the performance of starters and backups in terms of save percentage over that time span. Starters are represented by the pink line, the league average by the blue line and backups by the yellow. The chart immediately below it graphs starter and backup save percentages as a percentage of the league average.

Based on this, I’d argue that the replacement level for save percentage falls somewhere between 99% and 99.5% of the league average save percentage in a given year. By looking at the backups, I’m effectively looking at the players who are a step away from playing and who will be looked to to step in if the starter falters or is injured. There’s probably a fair argument to be made that the lower end of the scale is the more appropriate choice as teams are loathe to make a backup who can achieve at anywhere close to league average available under normal conditions.

From 1998-99 on, I’ve split in into ES/PP/PK information as well. The blue and yellow lines represent the starters; pink and teal the backups. This is important because setting different replacement levels for ES/PP/PK play will prevent certain goalies from relying on the fact that their team doesn’t take a lot of penalties to look more valuable than they actually are. Martin Brodeur has had an inflated save percentage for years because the Devils just don’t take very many penalties. He sees fewer shots against the opposition’s PP than your typical goalie, making it easier for him to compile a good save percentage.

The graphs just display ES and PP; there’s so few shorthanded shots taken that I don’t know that you can actually glean that much from the information. The six year average for both ES and the PP shows the average backup posting a save percentage of about 99.3% of league average. It’s lower for shorthanded shots over that period - 99%.

In light of this, using a replacement level figure of 99.3% of league average seems to make sense to me. Some reading this might object that this is going to be skewed because different teams allow different difficulty levels of shots. While this is true, it can be made context specific. To illustrate, consider two goalies, Goalie A and Goalie B. I’ll limit this example to ES save percentage but it applies to all situations. Assume that they both face 1500 shots at ES. Goalie A puts up a .925 ES save percentage against shots that the average goalie would post a .915 ES save percentage against. Goalie B posts a .920 ES save percentage against shots that the average goalie would post a .900 ES save percentage against. In order to credit the value properly, this needs to be accounted for.

It seems straightforward enough to me. Say that the league average at ES was .915. Using 99.3% of league average as replacement level, that would mean a replacement level of .909. For Goalie A, who played on a team that allowed shots of league average difficulty, the replacement level against which he’s measured would be .909. For Goalie B, who played on a team that allowed shots that the average goalie would post a .900 save percentage against, or .015 below the league average, his replacement level should be similarly lowered.

When you then turn to consider which goalie provided his team with more value at ES, you’d calculate it as follows:

Goalie A: ((1-context specific replacement level save percentage of .909)*1500 shots) - 113 goals allowed = 24.1 goals above replacement level
Goalie B: ((1-context specific replacement level save percentage of .894)*1500 shots) - 120 goals allowed = 39.5 goals above replacement level

Simple, eh?

Repeat the process for PP and PK shots and you’ve got what I believe to be a very reasonable estimate of the value of the goalie’s shot stopping to the team above replacement level. There’s leaves only the questions of how to quantify the value that a goalie adds to a team in terms of his puckhandling, shot prevention, shot quality reduction, penalty taking/drawing and offence. I figure that I’ve made my contribution and that’s someone else’s problem.

If forced to comment on those areas, my guess would be that those contributions are surprisingly small. I’d guess (an educated guess, but still) that shot quality reduction is the only one of any real signficance, due to rebounds and rebound control and that even that is a less important skill than people believe. Puck handling is a popular one but I have a hard time thinking that it amounts to much, for a number of reasons. First, there’s the issue of a replacement level. Virtually every goalie in the NHL can make the simple plays, going behind the net and stopping the puck (enjoy Columbus Ty!). It’s rare that you’ll actually see a goalie go out and make a difficult play on a puck, just because of the risk if something bad happens. Most of the time, I’d imagine that goalies are just taking plays away from their defencemen. I’ve kicked around numbers on this before, how many plays a goalie would have to make to have it be a significant skill and the number was pretty high. I’m open to being convinced the other way but it’s not a skill on which I place a great degree of value.

A couple of random comments on the whole thing:

I’ve got a reasonable system of estimating the difficulty of ES shots that a goalie sees now. It’s not perfect, it misses information but it’s a reasonable starting point. I don’t know of a method that I’m really impressed with for shots against the PP - they simply need to be split out for 5 on 3’s or when shorthanded. It’s possible that simply using the ES numbers would work for shorthanded shots - that’s something that I intend to look at. My point is, for the 2003-04 and 2005-06 numbers, I can come up with an ES shot quality number to make a context adjustment; I don’t have a reasonable system right now for other shots. Absent that, I’d just go with the league average.

I’ve considered the possibility that I’ve assigned too low a replacement level, as the method that I’ve used leaves me open to the possibility of being fooled by the numbers, so to speak. Consider a very simple three team league. It’s got the Oilers, Flames and Canucks. Edmonton has Grant Fuhr and Andy Moog, Vancouver has Richard Brodeur and Kirk McLean and Calgary has Mikka Kiprusoff and Mike Vernon. For the purposes of this example, assume that the goalies face the workloads and put up the results in the chart beside this paragraph.

That would give me a league average save percentage of .899. My starters would have a save percentage of .900. My backups would have a save percentage of .894, which happens to be (well, in this case I built the example that way but you get my point) 99.3% of league average. If I say, as I did here, that I’ll set 99.3% of league average as my replacement level, I get the goals above replacement (GAR) levels indicated in the chart above.

The problem here lies in the possibility that Edmonton allows shots at a .904 level, Vancouver at an .894 level and Calgary at an .884. In that case, the real goals above replacement (RGAR) would be correct and it would turn out that all of the starters and backups made the exact same contribution. More importantly to me in looking for a definition, I would have been misled by the vast disparities in shot quality. Effectively, I’d be giving goalies way too much credit for their role in preventing goals against. This is obviously important if I want to make arguments about what percentage of a team’s budget should be allocated to the goalies. Conventional wisdom holds that goalies have a huge impact; it seems to me that it’s not proven that this is true and people should be open to the possibility that it isn’t. I obviously come down on the other side of the fence but I’m open to the possibility that I’m wrong about that.

I don’t think that this happened to me, or, in the alternative, if it did happen, I believe the error to be far less significant than that in my example. I’ve got two reasons for thinking this: first, the fact that I had 16 years worth of teams would seem to reduce the possibility of it lining up in this way. Second, I’ve got two years worth of numbers at ES to work with. There doesn’t seem to be any strong trend there towards goalies ES save percentages being tightly tied to their expected ES save percentage numbers. Mathy suggestions for methods of investigation are welcomed.

Alan Ryder’s Player Contribution method has been discussed a bit in the comments to recent threads. In his paper explaining it (available at Hockey Analytics, he set the “save percentage threshold” (SPT) for the 2002-03 season (which was the one that he examined) to .893. He’s appears to be using the term in the same sense that I’m using “replacement level”. I think it’s fair for me to say that he was doing so because that save percentage threshold fit his model:

When do goaltenders tend to flunk out of the NHL? This is a tricky question to answer since there are so few goaltenders and some sub-marginal players play more than a few games below this level. The SPT clearly is a function of the openness of the game.

I have set the 2002-03 Save Percentage Threshold (SPT) to 89.3%. This is the level at which MGG [Marginal Goals Goaltending] equals 16.67% or 1/6th of league Marginal Goal totals. The following formula forces SPT to be 1/6th:

SPT = .893 =1-(GA x 7/6 - ENG)/(SOG - ENG)

The result of using 89.3% for SPT is that there are some goaltenders playing at a sub-marginal level. BUt it is clear from their playing time that the coach would prefer someone else in the net.

Goalies with .893 save percentages or worse saw about 9% of the league’s shots in 2002-03. It seems an inappropriately low replacement level to me - it’s 98.2% of league average in that year. I’m sure that JavaGeek (Chris? I’m going to start referring to him as Chris because I feel like a dork typing JavaGeek. Not be confused with Chris! of course) knows the workings of Ryder’s system better than I do, so maybe he can offer some comments on whether this is more than an artificial construct and whether it blows up the model if you start to shift to more empirical (in the broadest, most expansive sense of the word possible) values.

An advantage that I think my method of going about this has over Ryder’s method is that by relying on the averages for backup and starter instead of coming up with a model and then seeing what percentage I need for it to work, I’m hopefully counteracting the effect of really good defensive teams carrying goalies and really bad ones burying them. I’d hope that the effect was washed away with the volume. Again, math style suggestions as to approaching this would be appreciated.

I don’t mean to bag on Ryder’s method too much here - I seriously disagree with some of his assumptions as well as his methods of dispersing credit for events but I’m sure that some people will have the same thoughts on this. I’m in agreement with him in terms of the idea of allocating value and credit based on some concept of marginal goals - his method with crediting goalies is the exact same as mine once you get beyond setting the levels. I just think that he’s too low in terms of the point at which teams will/should look elsewhere.

The effect of what Ryder has done would be to provide goalies with too much of the credit for what their teams have accomplished defensively. While I’d agree with his expressed belief that goalies receive an insufficient amount of credit for their contribution, I think that he gives them too much, given the observed results here.

In terms of presenting this data, I think that the best way to do it is to use a variation of Woolner’s Value Over Replacement Player (VORP) stat. His stat asks how many runs an average team would score with a given player getting a certain percentage of team at-bats as opposed to how many they’d score with a replacement level player getting those at-bats. The relationship of OBP and SLG to run scoring is well enough known in baseball that this is possible. For instance, as of the All-Star break, Travis Hafner has 10.6% of Cleveland’s plate appearances and has put up a .322/.461/.650 line. You plug that in in comparison to a replacement player and conclude that Hafner would make the average team 55.8 runs better than they’d be with a replacement level player in that position.

This information should be presented in the same way, I think. Ask what percentage of his team’s shots a goalie saw and then plug it in compared to a replacement level goalie seeing the same percentage on a team that allows an average number of shots. This would address the problem of guys like Roberto Luongo seeming more valuable than he actually is because he played on a team that allowed shots at an alarming pace while at the same time systematically underrepresenting the value of someone like Mikka Kiprusoff who plays in a system where he sees few shots. While the team specific information is obviously important as well - Luongo’s is more important to the Panthers than Kiprusoff would be to the Flames, if you want to look at the question of which goalie had the better season while stripping the effect of team from the equation, this is something that you need to know.

As far as a framework for evaluating goaltender contribution and which goalie made the largest contribution, this makes more sense than anything else I’ve seen. It’s a decent enough framework that some organization with access to really good information on shot quality could have a real solid footing on what a given goalie has actually contributed in the past to his team. I’ve got a second post in the works on this that looks at the results, particularly from this past season and 2003-04, for which I’ve got some sort of shot quality numbers. I’m interested to read any comments or suggestions on improving this though.