bookie, I tend to believe the book when it comes to the playoffs and think that the playoffs are basically a crapshoot. They take a game that plays 162 games a season and squeeze it down to a 5-game series followed by 2 7-game series'. When you consider that Hockey, with an 82-game season, uses 5, 7, 7, and 7-game series' and Basketball, also with an 82-game season, uses 4 7-game series', it's kind of ludicrous that Baseball plays the same number of games.

In other words, Baseball's season is designed to weed out luck and produce the truly best teams in the league, but their playoffs are the equivalent of their season ending at the ASB. Because of this, baseball's playoffs introduce an unnatural amount of luck compared to the regular season.

But, let's look at the Oakland A's in the playoffs last year game by game:

Game 1: OAK 5 - 4 BOS
OAK actually had 2 SB and, for some unknown reason, Hatteberg was caught stealing. This is the same Scott Hatteberg that has 1 SB and 4 CS for his career. Why he was running in the playoffs is anybody's guess. There are some interesting numbers for the stats that have been determined that a pitcher can directly control (BB, K, HR).

BOS: 10 BB, 10 K, 0 HR
OAK: 7 BB, 11 K, 3 HR

Now the 3 OAK HR's weren't huge because there was a total of 1 man on base for all 3. The biggest thing that stands out to me is the BB/K ratio, especially considering Pedro (4 BB/3 K) was on the mound for BOS.

Interestingly enough, OAK took the game to extra innings by playing their version of "small ball". The inning started with a R.Hernandez flyout, followed by a McMillon BB (Byrnes runned). Then, Singleton was HBP and Ellis K'd. Durazo then singled in McMillon and Chavez grounded out to end the inning. That's the way OAK plays and it worked for them.

What was even further interesting was that OAK won the game on a bunt single by R.Hernandez in the 12th with 2 outs: "R Hernandez reached on bunt single to third, E Chavez scored, S Hatteberg to third, T Long to second."

So who says they don't "manufacture" runs? However, because they did it in extra innings, a factor of luck should be considered. Why did this set of plays work in this inning and not any other? Typically, no team bunts with 2 outs anyhow, so why did it work this time?

Game 2: OAK 5 - 1 BOS
OAK got all the runs they scored in the bottom of the 2nd. The inning went as follows:
-S Hatteberg grounded out to first.
-J Guillen walked.
-J Guillen to second on passed ball by D Mirabelli.
-R Hernandez singled to right, J Guillen scored.
-J Dye hit by pitch, R Hernandez to second.
-E Byrnes doubled to left, R Hernandez and J Dye scored.
-M Ellis walked.
-E Durazo grounded out to first, E Byrnes to third, M Ellis to second.
-E Chavez safe at second on throwing error by second baseman T Walker, E Byrnes and M Ellis scored.
-M Tejada flied out to center.

Not once in that inning did they play small ball, yet they managed 5 runs, in the playoffs no less.

Game 3: OAK 1 - 3 BOS
There are 3 reasons OAK lost this game...
Hits: 6
BB: 2
HR allowed: 1

OAK was only on base a total of 8 times. That's a failure of OAK's system. Their system tries to maximize men on base through OBP. It's not surprising that the 2 "Mr. Swing at Anything's" (Tejada and Chavez) were a combined 0-for-9 with 1 BB. The only hits OAK got were from the bottom 4 players in their lineup. That's not going to work for ANY team except maybe the Yankees.

Now, realistically BOS didn't do much better (7 hits, 4 BB), but the game winner was an 11th inning HR off of Rich Harden. That's something Harden could control and he didn't. Still, OAK lost in extra innings, which again introduces the luck factor.

Game 4: OAK 5 - 4 BOS
Again, Chavez and Tejada are nowhere to be found. A combined 1-for-8, with 0 BB, and 5 men LOB. Again, the majority of hits come from the bottom 4 players in the lineup. Again, the game comes down to a late-inning HR (2-run by Ortiz in the 8th) that is in the control of the pitcher (Rincon). Again, you have to consider the luck of it all. Rincon is no slouch, as I'm sure you read in the book. In 55.1 IP last season, he gave up just 4 HR. That he gave up 2 in 4 IP in the playoffs simply points to the luck of it all.

Game 5: OAK 4 - 3 BOS
Again, where in the heck are Tejada and Chavez. A combined 1-for-4 with 0 BB. At least Tejada managed an RBI. But, realistically, the game came down to BOS's 6th inning. Here it is:
-J Varitek homered to left.
-J Damon walked.
-N Garciaparra fouled out to first.
-T Walker hit by pitch, J Damon to second.
-M Ramirez homered to left, J Damon and T Walker scored.
-D Ortiz struck out swinging.
-K Millar popped out to second.

The one thing you'll notice is that all the players who scored were a result of actions that were directly controlled by the pitcher: BB, HBP, and HR. Again, OAK's pitching failed them. Those were the only HR's that Zito gave up in 2 games, but they were costly. He gave up 19 in 231.2 regular season IP, but then gave up 2 in 13 IP in the playoffs, almost double his regular season rate.

So games 3, 4, and 5 all came down to pitching mistakes. Could those mistakes have been offset by more offense? Sure, but Tejada and Chavez did exactly what they weren't supposed to do: make outs. Combined, for games 3, 4, and 5, Tejada and Chavez were 2-for-25, with 1 BB. That's not OAK baseball and I doubt that's going to work for any team in baseball.

When questioning OAK's abilities, I also look back to their 2002 playoffs when they scored 26 runs in 5 games. Remember Ray Durham saying, "I don't see a lot of playoff games where the score is 8-5. It's always 1-0 and 2-1," when he was trying to talk about how important it is to steal bases? The average score of the MIN/OAK series was 7.6-3.0. That' doesn't look very close to 1-0 or even 2-1 to me and that was the point. People think the playoffs are about small ball, but it's actually a microcosm of the game with more luck than usual added in because of the short series'. Do you think that if the playoffs were twice as long as they current are that Tejada and Chavez would continue to hit a combined .080? I wouldn't think it. The more games played, the more the averages return to the averages.

And while people say OAK can't win in the playoffs, they seemingly forget that they were up 2-0 on the Yankess and then had luck strike against them in the final 3 games.

Game 3 had 1 run: A Posada HR off of Zito. OAK had a combined 6 hits and 1 BB. Again, not OAK's way of playing.

OAK simply got destroyed in Game 4 and lost Game 5 mostly because of 3 errors. Both teams had 3 ER allowed. Without the errors, it could have been a completely different game, series, and playoffs. Again, the games coming down to luck.

-------------------------------

Okay, enough of baseball, this is a hockey forum! ;)

I'm not sure that you can take the next step and throw the goaltenders GAA into the mix to extrapolate points in a given stretch of time.

Honestly, I don't know if you can do this either. I kind of made this one up as I went along and I think I was wrong. It seems that GAA will vary vastly from year to year for goalies.

Looking at GAA and Sv% compared to career averages, I found something interesting. In order to ensure relavent stats, I only used years where goalies started more than 40 games, except for this year, where I allowed stats as long as the goalie has started 30 games.

Martin Broduer has seen jumps as much as 113% (1994) and 80% (2003) off his career GAA. But, what remains the same, or at least consistent enough to bare further looking into, is his Sv%. The biggest deviation from his career Sv% that Brodeur has had is 102% (1996,2003) and 99% (1994,1998,2000,2001).

But let's look at a goalie who maybe isn't viewed as being quite as stable as Brodeur. Sean Burke has seen jumps as much as 140% (1992) and 77% (2000,2001) off his career GAA. Meanwhile, the biggest deviation from his career Sv% is 102% (2000,2001) and 97% (1988,1992).

Let's look at someone like Ron Tugnutt. Because of his career, I'll include all years where Tugnutt started at least 30 games. Again, we find wide variances in GAA, from 151%, 132%, and 132% off his career average in 1989-1991 to a GAA as low as 59% of his career average in 1998. But, again, his Sv% only varies as high as 103% in 1998 and as low as 96% in 1989. The difference between '98 and '89? Tugnutt went from 32.76 SOG/60min in '89 to 24.04 SOG/60min in '98.

Now this may seem obvious, that the more SOG a goalie faces, the more goals he'll give up. But I don't recall ever giving Sv% as much of a look as GAA when considering how good goalies are or which goalies to draft. Through the limited data I've been able to process this morning and into the afternoon, it looks like Sv% could be a much bigger determination of who is a good goalie and who is not than GAA could ever attempt to come close to. Again, this is interesting to me because all we ever hear about and all that is usually ever focused on is GAA. It's like the Oakland A's focusing on OBP instead of AVG.

Now, let's revisit McLennan. It should now not be surprising that his Sv% this year is 102% of his career average while his GAA is 74%. Apparently, he has quite a defense in front of him that is preventing shots and this would also help explain Kipper (103% of his career Sv%, 61% of his career GAA). So, when substituting McLennan for Kipper, really we should just have to compare Sv%. If there is a large difference, we can break down the numbers further. McLennan has a career Sv% of .898 and Kipper has a career Sv% of .909. That's a 99% difference for McLennan, which suggests there should be little to no drop off with McLennan in net compared to Kipper. This means that CGY should come closer to the 24 points in 21 games that would be expected while Kipper was in net, barring any major injuries in defensemen or other major players for CGY.

1. GF/GA predicts team success. Further, (GF/GA)*GP can predict a team's Points.
2. A goalie's Sv% does NOT vary greatly from year to year and can remain constant even when changing teams.

With those assumptions in mind, I want to see how CBJ, who just fired their coach, could be different if they had a different goalie. Let's swap Marc Denis (.904 career Sv%) for Roberto Luongo (.918 career Sv%). In the 27 games that Denis has started, CBJ has 19 points. Their GF/GA is 51/71, or 0.718 (exact same as their total GF/GA), with 791 SOG. The 0.718 times 27 games yields 19 points, right on par with what has actually happened.

If we put Luongo and his .918 career Sv% in place of Denis, it yields a GA of 65 and a GF/GA of 51/65, or .785. Multiplying by 27 games, that gives us 21 points, or 1 extra win.

How is this useful? It could tell CBJ that they don't need to worry about their goaltending position. Even inserting one of the best career Sv% goalies in the league won't help but for about 1 win over the course of half a season. They need to either build more offense or build a defense that stifles SOG. Again, it sounds straight forward, but doesn't seem to be put on the minds of those making the decisions.

And then the question becomes, which is easier to build? A high-powered offense or a low-SOG defense? Further, which is cheaper? Can a team get by with minimal offense while allowing fewer SOG than normal? My GF/GA research suggests there's a limit, but if a team like CBJ could manage their same 74 GF, while allowing an average SOG more like NJD's 23.9 SOG/Gm instead of 30.4 SOG/Gm, they might be able to end up on the positive side.

If Denis had started all 37 games (an assumption just for these purposes) and only faced 23.9 SOG/Gm, CBJ would have a translated GA of 83 using Denis' career Sv%. Using this year's Sv%, they would have a GA of 75. Those numbers translate to a GF/GA of .892 and .987, or team Points of 33 and 37.

As the standings are today, 37 points would be just 3 points out of the playoffs behind DAL at 40 with 39 GP. At a rate of .987 GF/GA, CBJ would have 38 points after 39 GP, leaving them just 2 points shy of a playoff spot.

I don't recall ever seeing a stat for something like SOG Allowed While on Ice. There's always a focus on +/-, but if we can show that GA are a factor of SOG, does +/- really matter, or is it more of a function of the goalies Sv% combined with allowing SOG?

In other words, wouldn't a stat that shows how many SOG a team faces compared to how many SOG a team takes while a player is on the ice be a better sign of how that player helps/hurts his team?

For instance, let's say that while Player A is on the ice, his team allows an average of 10 SOG and takes an average of 8 SOG. Let's call that a new -2. Let's say that while Player B is on the ice, his team allows an average of 8 SOG and takes 12 SOG. Let's call that a new +4. Given those situations, it's possible that both players could be +, -, or even opposite +/- of their SOG +/- depending on the goalie behind them and the goalie in front of them. However, (again, as long as goalie success is indeed tied to SOG) SOG would be more descriptive of scoring opportunities. The rest would simply be the luck of when the goalie allows the goals he would be expected to allow.

--------------------------------

I've managed to look at 291 goalie seasons going back to the 1995 season and have found some interesting stats regarding GAA, SV%, and SA/60min. I started my research with goalies who, in their career, had at least 2 40-start seasons. That gave me 49 goalies (ranging from Biron to Roy to Giguere to Fuhr and so on) and 622 seasons of stats start with.

Once I gathered all of those goalies stats, I narrowed my focus and only looked at the seasons where the goalie actually started 40 or more games. In some cases, Yahoo did not have the number of starts a goalie made in earlier seasons (like the 80's), so I used a threshold of 40 or more games played for those particular seasons. This gave me the 292 seasons that I worked with.

Comparing seasonal GAA to career GAA, there was a minimum percentage difference of 63.9% (Tom Barrasso, '97, PIT) and a maximum percentage difference of 144.9% (Byron Dafoe, '95, LOS). The average percentage difference was 99.1%, which is to be expected since these seasons represent the "major" seasons (over 40 games started) of the goalies' careers and will have more impact on their career averages.

Comparing seasonal SV% to career SV%, there was a minimum percentage difference of 96.2% (Patrick Roy, '85, MON) and a maximum percentage difference of 103.4% (Tom Barrasso, '97, PIT). The average percentage difference was 100.1%, again as was to be expected.

It is no small wonder that Barrasso's '97 season shows up in both areas. It is, statistically, a complete anomoly. Outside of '97, his SV% was never more than 101.1% and never less than 98.0% of his career average (0.892). Yet, in '97, he managed a SV% of .922, or 103.4% of his career average.

As I mentioned at the beginning of this post, I also looked at SA/60min to see how that affected goalie performance. If it is true that SV% varies little and can be used to predict/determine goalie success, then a rise in SA/60 compared to a goalies career SA/60 should translate to a higher GAA, and vice versa.

Out of 291 seasons of data, there were 79 seasons (27.1%) where a goalies' GAA was 90%, or less, than their career GAA. If the assumption above is correct, we would expect a decline in SA/60 for those 79 seasons. Out of those 79 seasons, 70 were accompanied by a decline in SA/60.

Now, let's look at the reverse. Out of the 291 seasons of data, there were 62 seasons (21.3%) where a goalies' GAA was 110%, or more, than their career GAA. Again, if the assumption is correct, we would expect a rise in SA/60 for those 62 seasons. Out of the 62 seasons, 52 were accompanied by a rise in SA/60.

It looks like this might be well on the way to showing that SV% is far more important of a stat than GAA. It further begins to show that a significant rise/decline in GAA is almost always the result of a decline/rise in SA/60.

I haven't worked backwards from SA/60 to see what significant changes result in, but I have a feeling that it's going to be directly tied to GAA.

And back to the team aspect of this, it is becoming more and more clear that teams that can stop SOG will be more successful than those who can't. Again, this may seem like a "duh" concept, but if it is, it's certainly not one that many teams seem to grasp.

As for the numbers, what I'm really wishing I had was some numbers to work out that SOG +/- I was mentioning earlier. I really think it might be worth looking into. In the meantime, I think ukula and I might be on to something regarding predicting team success. I'm also going to look deeper into goalie success to see if any of these ideas hold true when you go WAY back into the history of the NHL. Further, I want to see if there's a translation from the minors to the NHL so that you could accurately predict goalie success based on minor league (or college or European) success.

I've expanded the goalie research as much as Yahoo will allow. The new raw data now includes 58 goalies and 695 seasons. As I mentioned in my last post, Yahoo does not include GS for earlier years. Because of that, I have changed my filter for seasons to research to 41 GP and am now ignoring GS. With this new filter, I now have 325 seasons of goalie info to work with. Not much more than before, but still more.

FINDINGS

Season/Career Comparison Findings
As was mentioned before, GAA varies greatly from a goalies' career average. There were 67 seasons (20.6%) when a goalie's GAA was 110% or more over his career GAA. There were 87 seasons (26.8%) when a goalie's GAA was 90% or less under his career GAA. The low percentage difference was 58.7% (Tugnutt, '98, OTT) and the high percentage difference was 144.9% (Dafoe, '95, LOS).

Again, as was mentioned before, there is far less variance in a goalies' SV% compared to career SV%. There were 14 seasons (4.3%) when a goalies' SV% was 102% or more over his career SV%. There were 11 seasons (3.4%) when a goalies' SV% was 98% or less under his career SV%. The low percentage difference was 96.2% (Roy, '85, MON) and the high percentage difference was 103.4% (Barrasso, '97, PIT).

Again looking at SA/60 for the seasons where a goalies' GAA rose more than 110% or declined more than 90% off their career GAA, I found the numbers from above continued to hold true. In the 67 seasons that a goalies' GAA rose by 110% or more, it was accompanied by a rise in SA/60 over their career average 57 times, or 85.1%. In the 87 seasons that a goalies' GAA declined by 90% or less, it was accompanied by a decline in SA/60 under their career average 74 times, or 85.1%.

Now for the new info. I decided to look at a year-by-year comparison of GAA and SV% along with SA/60. Because I'm still only using 41-GP seasons, it actually means a comparison between a goalies' 41-GP season and their previous 41-GP season. For most goalies, there is not more than a 1-year gap between any 2 41-GP seasons, so this should not be a concern.

Looking at year-by-year GAA, the average change is 99.3%. The high percentage change was 152.5% (Kidd, '97, CAR to '00, FLA). The low percentage change was 55.7% (Tugnutt, '90, COL to '97, CBJ). The high percentage change for 2 back-to-back years for the same team was 149.2% (Dafoe, '98, BOS to '99, BOS). The low percentage change for 2 back-to-back years for the same team was 68.7% (McLean, '90, VAN to '91, VAN). Not surprisingly, 3 of the 4 seasons were accompanied by a significant corresponding rise/decline in SA/60 between the years compared:
Kidd ('97 to '00): 112.1%
Tugnutt ('90 to '97): 67.0%
Dafoe ('98 to '99): 99.2%
McLean ('90 to '91): 92.6%

Dafoe essentially saw little change in his SA/60, yet still managed to lower both his GAA and his SV%. This goes against the other data and shows that it is possible for a goalie to break outside of his career SV%. It's just not very likely.

Looking at year-by-year SV%, the average change is 100.1%. The high percentage change was 103.9% (McLean, '90, VAN to '91, VAN). The low percentage change was 96.0% (Dafoe, '98, BOS to '99, BOS). These both also take care of back-to-back seasons for the same team.

Looking at significant change, there were 62 seasons when a goalies' GAA was 110% or more over their previous season's GAA. Of these 62 seasons, 51 (82.3%) were accompanied by an increase in SA/60 from the previous season. There were 76 seasons when a goalies' GAA was 90% or less under their previous season's GAA. Of thse 76 seasons, 57 (75.0%) were accompanied by a decrease in SA/60 from the previous season. This simply verifies what the previous data showed.

Looking at significant change for SV%, there were 26 seasons where a goalie's SV% was 102% or more over their previous season's SV%. Not surprisingly, 22 of those seasons were accompanied by a decrease in GAA of 90% or less under the previous season. There were 13 seasons where a goalie's SV% was 98% or less under their previous season's SV%. Again, not surprisingly, 12 of those seasons were accompanied by an increase in GAA of 110% or more over the previous season.

So now we're back to, "what does it all mean?" I think I might be getting closer to drawing some conclusions. The initial ones are as follows:

• A goalie's SV% will tend to remain close to their career SV% and their previous season's SV%.
• A goalie's GAA will vary widely, both compared to career GAA and a previous season's GAA.
• Career and previous season's SA/60 and SV% may be the biggest determining factors for what a goalie's GAA will be in a season.

So again, why is this important? Because we've had GAA shoved down our throat since we started following hockey. Every time I watch a game, SV% is usually either never talked about or rarely talked about. According to these stats, it should be the ONLY thing, maybe along with SA/60, that is talked about.

And let's look at the Vezina Trophy. According to NHL.com, "the Vezina Trophy is an annual award given to the goalkeeper adjudged to be the best at this position as voted by the general managers of all NHL clubs."

Last year, the award went to Martin Brodeur. 2nd place was Turco, followed by Belfour. Given that, and what has been outlined above, who would you say is the best goalie of the following:

• A: 2.26 GAA, .922 SV%, 7 SO, 29.1 SA/60
• B: 1.72 GAA, .932 SV%, 7 SO, 25.5 SA/60
• C: 2.02 GAA, .914 SV%, 9 SO, 23.4 SA/60
• D: 1.83 GAA, .925 SV%, 6 SO, 24.5 SA/60
• E: 2.30 GAA, .920 SV%, 8 SO, 28.9 SA/60

Personally, I would give it to either goalie B, A, or D. Those guys are Turco, Belfour, and Cechmanek. C is Brodeur and E is Giguere. I just don't see Brodeur as the best goalie on that list. All he really did was play more games and win more games, something that is not completely in his control like the stats above (with the exception of SA/60).