The Hosers

Patrick Kaleta played one of his best games of the season Wednesday, his continued

strong play could go a long way to a turn around for the team.

Dallas will travel to Buffalo to take on the Sabres tonight in search of their first regular season win in Buffalo since Lindy Ruff's first game behind the bench in Buffalo (10/7/97). This time around Ruff is recovering from three broken ribs and it is unclear whether he will be behind the bench tonight. Dallas is coming off of a 4-2 victory over Columbus last night and they've now won four of their last six games. Buffalo will also look to continue their own good fortune as they are 4-0-1 in their last five games. The Sabres will also be seeking their first three game win streak since early November (when they won four in a row).

Some thoughts:

Buffalo's last regulation loss at home was exactly two months ago, a 4-1 loss to the New York Rangers. They have gone 5-0-4 at home since then.
Kari Lehtonen played last night against Columbus, it has not been confirmed but given the back to back situation, it's expected that Richard Bachman will get the start tonight. Bachman is 6-3-1 with a 2.72 GAA and .912 SV% this season.
It's worth noting that the Stars are 0-8-1 in the second game of back to back situations this season. Additionally, in situations where the Sabres were off and the opponentsplayed the night before, Buffalo is 4-1 this season and 30-12-3 since the lockout.
Patrick Kaleta played his best game of the season on Wednesday night against the Bruins. Kaleta took on public enemy #1 in Milan Lucic and was involved physically all game, both in hitting and engaging himself in scrums and battles all night long. He had the Bruins running around all night looking for him, that is exactly how you know when Kaleta is on his game. His role is invaluable to this team when it's executed correctly. To cap it all off, Kaleta scored a goal late in the 2nd period to make it 4-0 and followed it with his standard dice-roll celebration, surely not earning him any further points with the visiting Bruins.
Both the Sabres and Stars excel in one-goal games, the Stars to the tune of a 14-6-2 record and the Sabres to the tune of a 13-6-6 record. With both teams in the bottom half of the league in scoring and both goalies playing well, I would expect a very close, hard fought, low scoring game tonight.
Speaking of close games, 16 of the Sabres last 20 games have been decided by two or fewer goals. Additionally, nine of their last 12 home games have been decided by one goal.
Both Buffalo and Dallas seem to come out strong as both have a positive goal differential in the first period but get blown away in the latter two periods. Dallas has been outscored 90-108 in the second and third frames this year while the Sabres have been outscored 79-102 after the first period.
Dallas' offensive attack will be centered around All-Star Jamie Benn who, coincidentally, will be playing in his 200th NHL game tonight. Benn is one of the best kept secrets in the NHL; big and strong with a lethal shot, the Sabres defense would be remiss if they didn't game plan around the talented center. The 22 year old former 5th round draft pick checks in as the team leaders in points with 48 and second in goals with 17 through 48 games. Among players who started their careers with Dallas and played their first 200 games there, Jamie Benn has the most points in history through the first 200 games.
It's also worth noting that Jamie Benn sports the Melrose mullet. He is the only player in the NHL that can challenge Melrose for best hair in the biz (it's also creepy how eerily similar they look).

Ryan Miller has been on top of his game of late. By far his best stretch of hockey this season, Miller has allowed just five goals in his last five games, including two shutouts. He sports a 0.95 GAA and .969 SV% over that five game period. He looks like his old self and is quickly making up for lost time. If Miller stops the next 40 shots he faces he will, amazingly, be back to the league average of .914 which seemed so unattainable as recently as the All-Star break.
Mike Harrington of the Buffalo News confirms that Thomas Vanek will return to the lineup, but it is unclear where exactly he will fit in. Originally it was expected he would return to play with Roy and Pominville but since that line has been playing so well with Leino taking Vanek's place, they may look elsewhere to place Vanek. In responding to questions about whether he'll break back into the lineup with Roy and Pominville, Vanek was quoted by the Buffalo News:
"Geez, life goes on," he said slowly with a smile. "The line's been playing well. The three games I missed the team played great. For me, it's just to try to find a spot in the lineup then work myself up again."
Ville Leino is playing his best hockey of the season as well. Leino, Pominville, and Roy seem to have found some chemistry and are really clicking offensively. Leino seems to finally be playing with some confidence as he appears to be much looser out on the ice and just letting things come to them rather than forcing it.
I'm not one to jump on the injury excuse bandwagon by any means but there are some pretty compelling numbers to support it on the defensive side of the puck. The team started the season with seven defensemen on the roster; all but Marc-Andre Gragnani have missed some time with an injury (total of 57 man games lost). Of the Sabres 53 games, only 24 have been played with six out of those seven defensemen in the line-up. In those 24 games, the Sabres are 15-8-1 while allowing only 2.29 goals per game. In the remaining 29 with a banged up blue line, they are 8-16-5 while allowing 3.41 goals per game.
In addition to Leino and Miller, Ennis has also been playing his best hockey of the season of late. He did have two points on Tuesday but in general his play has not been reflected on the scoreboard as those are the only two points he's posted his four games since returning from an ankle injury. Upon Ennis' return to the lineup against Montreal four games ago, the coaching staff decided to return him to the center position where he played all through juniors. He made the switch to wing in the AHL because they felt his game was a better fit there. That may be the case but thus far he has been a revelation at center as he's been the dynamic offensive player we haven't seen out of him at all this season. The Sabres are 11-5-3 with Ennis in the lineup.
Luke Adam is also confirmed as a scratch. It is baffling why a 21 year old struggling rookie is being scratched on and off while being fed limited 4th line minutes when he is in the lineup rather than being sent down to the AHL. Their are plenty of guys in Rochester that would be a better fit on the 4th line in Buffalo while allowing Adam to regain his confidence and play 18-20 minutes per game in Rochester.

Follow Derek on Twitter - @TheHosers_DSJ

Photo via Jonathan Daniel/Getty Images

The numbers indicate that Marc-Andre Gragnani may of been the

beneficiary of some significant luck thus far this season.

PDO (named after its creator) is likely the most pertinent advanced hockey statistic for people to learn. PDO is really not a complex or advanced statistic at all; rather, it's simply the measure of a team's save percentage while an individual player is on the ice plus a team's shooting percentage while that specific player is on the ice. Obviously, the median number with this statistic is 1; each goal scored will eventually mean a goal against over the course of a season or a couple seasons.

The reason PDO is so important is because it is likely the most predictive of all advanced statistics in hockey. PDO is basically hockey's rebuttal to baseball's "batting average on balls in play" (BABIP). BABIP basically measures the amount of luck a certain player has experienced over the course of a season. If the player is getting hits on a very high percentage of balls he puts in play, it's likely that luck plays a role and his batting average will likely regress as BABIP begins to correct itself. The same is true with PDO, if a player's PDO is extremely high, it will correct itself over time. PDO basically measures the amount of luck an individual player has experienced and is very predictive in the fact that this number will always migrate back to the mean of 1. Sometimes it takes a month to correct itself, other times it takes a full season, but over time it will correct itself.

Why is this statistic so reliable?

Shooting percentage while a player is on the ice involves a high degree of luck. It's no surprise that individual players may be more adept to goal scoring and thus, they will have a higher shooting percentage. However, this statistic measures a team's shooting percentage while a player is on the ice, so it will regress back to the mean much quicker than an individual player's shooting percentage will. For example, this season the gap from the top team in shooting percentage to the lowest team in shooting percentage is just over 3%. There is still roughly one-third of a season left to play and that gap will likely close even more as the season goes on. On a team, that gap in shooting percentage could range as much as 15% between individual players. Given varying linemates and such, this typically will even out in the end. When you take a large enough sample size, the numbers will migrate back to the mean. A great example of this is Corey Perry; he scored on 17.6% of his shots last season en route to winning the Rocket Richard trophy with 50 goals. That is an unsustainable shooting percentage so it would be expected that Perry's shooting percentage would regress back to his career average. This season he's scoring on 12.5% of his shots, which is just a hair below his career average.

Many players will sustain a higher shooting percentage than the league average, but many times this will be counter-acted by sub-par defensive play and that advantage in shooting percentage will be countered by a lower save percentage which is then reflected in PDO. This isn't the case for all players, there are some who excel at both ends but those players are few and far between. It may not all even out in one month, or even one season, but given a period of time the luck will even out.

Save percentage while a player is on the ice is also a largely luck driven stat. Granted, a stellar defensive player will likely have a higher save percentage, but PDO takes that into account as a defensive minded player will likely have a much lower shooting percentage while they're on the ice. So again, that will all even out in the end. It does often take more than one season for PDO to even out, if a player is well above 1 in any given season, it should be expected that they would regress the next season.

PDO Numbers (Proof)

A study was conducted which took the top 40 individual players in PDO and the bottom 40 individual players in PDO from the 2007-2008 season and they compared that season to the following two seasons to see how a player's PDO numbers responded after a good season or a bad season. The results were quite telling. The top 40 players in 07-08 averaged 1.033 PDO. Over the next two seasons, they averaged just a 1.002 PDO, resulting in a difference of -0.031 (3.1%). The bottom 40 players in 07-08 averaged 0.971 PDO. The following two seasons they averaged out to 1.002 PDO, again resulting in a difference of 0.031 (3.1%). This is a good illustration of how these numbers correct themselves over time.

Now some more Sabre specific numbers, listed below is the top five Sabres in PDO as well as the bottom five Sabres in PDO as well as their PDO's for the following season. The idea of PDO is that if it's well over 1.000 (or 100% or 1000 as it's displayed here) then it's likely that that player has experienced a little bit of luck. Conversely, if a player has a PDO well under 1.000, then it's likely that that player has experienced some bad luck. PDO is highly predictive because these trends often correct themselves from one season to the next. A high PDO one season will likely result in a significant decline in the next season, and vice versa. Here are the numbers:

As you can see, the numbers over the past four seasons do seem to back-up the theory that PDO will correct itself from one season to the next. Out of the 20 players listed here from the top five of each season, only one did not experience a decline the following season (Patrick Kaleta from 07-08 to 08-09).

One specific recent example of PDO correcting itself is in the case of Drew Stafford. Stafford posted a PDO of 1.035 last season. To simplify that, PDO is team shooting percentage while a player is on the ice plus team save percentage while they're on the ice. Last season, Drew Stafford clocked in at 103.5% which means we could have predicted a regression from Stafford this season. Sure enough, Stafford only has 8 goals and 24 points, not even on pace for 40 points this season, a far cry from the 41 goal/69 point pace he scored at last season. Part of the high PDO resulted from a career high (and unsustainable) 17.1% shooting percentage last season. That was bound to come crashing down to earth; his 6.1% shooting percentage this season is what we call a market correction.

To make it more clear and concise, below is a table of the averages from each chart. It covers the average PDO for the top five and bottom five players each season, along with their average PDO the following season and the average difference.

These charts show the more general trend, rather than looking at the numbers of each individual player each season. To make it even more general, the table below takes the average of all those listed above and summarizes it into one clear table.

As is evident here, you can see a very clear difference between the top five in PDO from one season to the next and the bottom five in PDO from one season to the next. The difference is so large that the top five not only experienced a regression, but on average they finish with a lower PDO the following season than the bottom five from the season prior. That is pretty compelling evidence of just how predictive PDO is.

To the left is a chart of the team's current PDO numbers (note: these numbers were from before last nights game). One of the most shocking things to me is the astonishingly high PDO for Gragnani. On one hand, this should not come as a huge shock due to Gragnani's surprising +10 rating, a team high. We have delved into the +/- debate before, this is further proof just how luck driven that statistic can be as high PDO and high +/- fall hand in hand. On the other side, this should not come as a surprise as his play on the ice has been downright brutal at times, yet his statistics do not justify just how bad he's been. For example, Gragnani is leading the team with a 1.68 GAA per 60 minutes when he's on the ice. This is driven by a ridiculously high (second on the team) .940 team save

The Hosers

Pages

09 February 2012

Pre-Game Thoughts: Stars at Sabres 2.10.12

Advanced Statistics: Why you should know PDO