Mark's Blog

I'm still in Toronto and it's getting stickier and more humid by the day, it seems. The traffic is somewhat insane here, it seems stepping into the car entails a bare minimum of a half-hour journey, between finding parking and dodging the other cars and pedestrians. If I lived here, I think having a car would be more or less unnecessary.

I went to the Yankees - Blue Jays game last night. The home nine (ten, I suppose, given the DH (which I like)) lost a close game, as closer-if-not-by-name Jason Frasor gave up three runs. The Yankees brought on the witch and a quick three up, three down ended the game. I have to confess, I was actually kind of hoping to see the Blue Jays fall behind at the end, so I would get the chance to see Mariano Rivera pitch.

I went to the Ontario Science Center today and the ROM yesterday. I was a little underwhelmed by both, perhaps partly because we are spoiled by such good museums in Ottawa, and I am no longer the optimal height for a lot of the exhibits. Also both buildings were under some fairly extensive renovations.

Well, I'm in Toronto. Went to the Blue Jays - Red Sox game last night. I didn't even know Pedro Martinez was pitching until I got to SkyDome. I had seen Pedro once before, in Montreal, I think in 1996 or 1997, I believe against San Diego. If memory serves, he retired 18 batters in a row in the middle of the game, but I'll have to check Retrosheet to be sure.

It turned out that the real star of the show was Ted Lilly, who pitched one of the best games of his career, a shutout, giving up 3 hits, and striking out 13 batters. Lilly used his excellent curve ball to fool the Boston hitters time and time again. There were a surprising number of Red Sox fans at the game, apparently most of them in my section (117). I had some great seats, in the 14th row. Lilly's game score was 92, which ties Johan Santana of the Twins for the top game score in the AL this year. A most impressive performance.

Today I got up around 10, watched some live Olympics, including the awesome gold medal win by Lori-Ann Muenzer in the match sprint. Of all the events in the Olympics, this has to be my favorite, with its unique combination of tactics and brute strength.

Also checked out the Hockey Hall of Fame, which was interesting. Seeing all the trophies was interesting -- they look a lot bigger in real life somehow. Also seeing the old-time equipment was amazing. I can't even imagine staring down a slapshot with one of those masks the goalies used to use.

A couple of unrelated things: first I'm going to shill for Bloglines.com, a web-based news aggregator that is amazing and a regular part of my web surfing. It lets you easily view new stories in RSS-enabled web sites and they just did a very slick re-design. Highly, highly recommended. It also has a feature to manage your blogroll, so I just hooked that up to the side menu, so you can see the feeds I'm currently watching.

I joined a lob-ball team with work and have been playing that once a week. It's been fun, but it seems like whenever I hit the ball hard, it gets caught, and when I hit the ball weakly, it falls in. I assume that will turn around with time though. We lost 15-13 tonight, but it could have been a lot worse since we started off with only 7 players. I think we're 2-6, but it's still a lot of fun.

A few movies seen recently (and my IMDB rating for them)
* Cidade de Deus (City of God) (10: amazing cinematography, moving story of a boy who escapes the violence of the Rio de Janeiro favellas)
* Paycheck (7: interesting concept but mundane execution)
* Shrek 2 (8: fun summer movie, similar to the first one)
* Spider-Man 2 (8: very good, maybe a bit better than the first but I enjoyed the first more since it was new)
* Super Size Me (8: moronic experiment and dubious science but punctuated with a number of good points about our fast food culture and obesity in general)

Time to delete the porn referrer spam...

As a fan of the Ottawa Senators, I was very happy today to learn that they acquired Capitals' star Peter Bondra. Bondra gives the highest-scoring team in the league and even greater power play presence and is a fast, skilled player that will fit right in with Havlat, Hossa, Alfredsson, Spezza and company. Bring on the playoffs!

On a hockey statistics note, I've been dissatisfied with Goals Against Average, GAA (goals allowed per 60 minutes played), as a measure of goalie success, since goals against are a function of shots faced, and this is not normalized between teams. Save percentage, or shots saved divided by shots faced, is a better way of looking at a goalie's skill level.

The problem with save percentages, similar (although in general a bit lower) to fielding percentages in baseball, is that they are presented as numbers less than but quite close to 1, which makes quick comparisons very difficult, and obscures significant differences. For example, if Player A makes 2 errors in 40 plays, he would have a fielding average of .950, and if Player B makes 1 error in 40 plays he would have a fielding average of .975. Player A makes twice as many errors, but the untrained eye only sees a .025 difference.

I decided to calculate the league average of shots allowed per game, in order to normalize save percentage to a more understandable number, in the form of a GAA. This was a bit more tricky to do than I initially thought, since the number of minutes a team has played isn't readily available (at least I couldn't find it online). I added 2.5 minutes for a team for every OTL, since there is more or less an equal chance of a goal being scored throughout the 5 minute OT session. I also removed empty net goals from this, since (I think by definition) all shots on an empty net are goals. This gave an estimated league average shots per 60 minutes of about 27.4.

For each goaltender, this number is multipled by 1 minus the save percentage (i.e. 27.4 * (1 - Save%)). I'll call this AGAA ("Adjusted" GAA). The list isn't very interesting in and of itself, since it's in the same order as the Save percentage list, but if we look at the difference between AGAA and GAA then some names pop out. It would also be a much more useful way of talking about save percentage in general.

The five goalies (among qualifiers) that are most overrated by their GAA are (ranked by AGAA - GAA) as follows (using numbers from games as of February 17, 2004). A positive number means that if the goaltender faced a league average number of shots they would have allowed that many more goals per game, based on their save percentage:

1. Marty Turco, Dallas, +0.41
2. John Grahame, Tampa Bay, +0.40
3. Patrick Lalime, Ottawa, +0.31
4. Robert Esche, Philadelphia, +0.29
5. Tommy Salo, Edmonton, +0.29

And the five goalies most underrated by GAA are:

1. Marc-Andre Fleury, Pittsburgh, -0.80
2. Sebastien Caron, Pittsburgh, -0.62
3. Roberto Luongo, Florida, -0.51
4. Olaf Kolzig, Washington, -0.35
5. Sean Burke, Phoenix/Philadelphia, -0.29

Note that these lists do not represent good or bad goaltenders, just those who have misleading GAAs, because they played with teams that allowed much more or much less shots a game than the league norm. AGAA is a normalized, easier to follow version of save percentage. As an example, the current save percentage leader, Vesa Toskala has a GAA of 2.01 and a save percentage of .933. Multiplying 27.4 by (1 - 0.933) we get an AGAA of 1.84. This would mean that Toskala, facing a league average amount of shots per game, would be expected to have a GAA of 1.84.

Being an Ottawa fan, I note that Martin Prusek's save percentage is .928 (GAA 1.80), which corresponds to an AGAA of 1.97. Lalime's .906 (GAA 2.27) corresponds to an AGAA of 2.58 (see how the true difference of just a few points of save percentage, is actually 0.6 goals a game! And while it might at first seem that Prusek, being the backup, might play against weaker opposition, a quick check of his record shows 10 games against playoff teams, 8 against non-playoff teams (and a few minutes of mop-up duty against the Rangers in a game in November), which is an average level of opposition.

One improvement that I could see for this would be to also adjust for the team's penalty minutes taken. It's not fair to penalize goalies for their teammates putting them in a situation where a lot more goals are scored (teams score at about 1.5 to 2.0 times the rate on the power play than they do at even strength).

I watched The Hours this week, the only 2002 Best Picture nominee that I didn't see in the theatre. I didn't have super-high expectations but ended up really enjoying the movie. I usually like films with interlocking stories and those that play around with time, with different personalities or events that may or may not come together at some point. There were a number of interesting philosophical points in the movie. I wouldn't recommend it to everyone, since it wasn't exactly action-packed, but if you like slower, more thoughtful movies you should give this one a shot. I gave it a 9 on IMDB.

I've been addicted to Civ 3 Conquests lately. I'm in a multiplayer game right now, against a friend and (right now) four computer civilizations. I have the lead in score and tech, but we are on different continents, and just like in real life, inter-continental wars are extremely costly, so the end is not in sight yet. I'm playing as the Romans, which is not one of my favorite civs to play, but it has worked out well. The Legionary was useful in the Ancient age and the Militaristic trait has helped to produce a couple of Military Leaders, which are always useful. I'm not a huge fan of the Commercial trait but I'm doing well in gold pieces in this game so I shouldn't be complaining. I usually play (and win, most of the time) on Regent level, and tried a game on Monarch recently, but got crushed quite quickly. I suppose I'll have to refine my strategies in order to make progress. I still have no idea how people can win on the Emperor and Deity levels.

The Super Bowl is tomorrow. I'm predicting a closer game than the bookmakers, but what do I know. Seven points seems to be an awful lot, given that the Patriots' offense isn't exactly tops in the league and Carolina has a very good defense. Plus, New England doesn't have home field advantage. I'll take the under of 38 points, and Carolina +7. I might as well do a final score prediction -- I'll go out on a limb and say Carolina 16, New England 13.

Everyone in Ottawa seemed to be down on the Ottawa Senators this year, after they got off to an average start. I was not one of the doubters, but it's still nice to see them play to their ability, with their defeat of the league-best Philadelphia Flyers on Monday and Florida Panthers tonight, 4-0, for Martin Prusek's first-ever shutout.

One of the lessons that I've learned from following baseball and sabermetrics in general is the lesson of small sample sizes. Simply put, a win-loss record over 20-odd games just doesn't prove much of anything. The team won the President's Cup last year, and have made very few personnel changes. Also, the Sens have an unusually bad record in one-goal games, 0-6-0-3. It *could* be some massive defect of character, but given their .571 winning percentage in one-goal games last year, it's probably just a spell of bad luck.

Another lesson to be learned from sabermetrics is that if a particular statistic isn't repeated year after year, then it isn't a skill, but a matter of chance. A lot of times you'll hear people talking about "clutch performance", but that's usually in the eye of the beholder, and doesn't stand up to analysis. If something is really a skill, then it can be repeated year in and year out. For teams, scoring and preventing goals is most definitely a matter of skill, even more so than winning.

Let's look at the Toronto Maple Leafs for a minute. They have a most impressive 13-6-5-2 record (a 104-point pace), but their goal differential is even, 68 GF and 68 GA. The Senators are 11-8-2-3 (a 92-point pace), but have 76 GF and 55 GA. The baseball statistician Bill James noticed a relationship in baseball between runs scored (RS) and runs allowed (RA), and winning percentage, which he calls the "Pythagorean Winning Percentage" (after the vague similarity to the Pythagoream Theorem). The equation is RS²/(RS² + RA²), and turns out to be a really good estimator for winning percentage, both in baseball and in hockey.

Just out of interest, I ran a statistical correlation on the number of points obtained last year and the goal differentials. There was a correlation of 0.983, which is ridiculously high (1 is the maximum possible correlation and 0 indicates no correlation). The slope was 1.04 and the intercept was -8.21, which is largely due to the fact that the NHL gives out a point for an overtime loss (which I wish they didn't, but that's fodder for another rant). Subtracting out the points for OTL, we get a correlation of 0.978, a slope of 0.969, and an intercept of only 2.59. In the aggregate, goals score and goals allowed allow us to predict won-loss record with great accuracy.

If we estimate the rest of the season based on this formula, and using the league average of about 1 OTL per 16 games played, we get a projection of 93 points for Toronto (33 already + 56 (.500 PWP * 56 games left + 4 (estimated OTL)) and 103 points for Ottawa (27 already + 72 (.656 PWP * 55 games left) + 4 (estimated OTL)). Quite a different picture than our original estimate of 104 points for Toronto and 92 for Ottawa.

I haven't given a justification of exactly why this formula is better than the previous record, but it's mostly due to the fact that goals relate quite directly to wins, and there are relatively a lot of goals scored to wins, so it's a more significant piece of evidence.

Of course, they still have to play the games, so we shall just have to see what happens. And whatever the numbers say, I'll still be worried if Toronto ends up facing Ottawa in the first round.

As everyone who would care knows by now, the Marlins won the World Series, defeating the New York Yankees 4 games to 2. I incorrectly predicted Yankees over Marlins in 6 games. It's pretty clear that the Yankees are a better team, especially given the 10-game difference in regular season records, but they ran into some strong pitching from the likes of Josh Beckett and Brad Penny, and their rotation was out of synch after going the distance with the Red Sox in the ALCS. For all the talk of the Yankees payroll, they had too many unproductive hitters, and shaky-at-best fielding. For the offseason, they need to resign Andy Pettite, get a solid free-agent starter to replace the supposedly-retiring Roger Clemens, and a right fielder (Vladimir Guerrero?!), at the minimum. Slumps from Giambi and Soriano, in particular, and a few injuries didn't help their chances either in the series. I liked Torre's move of batting Soriano lower down in the order (maybe not 9th), since his on-base average and free-swinging ways are not suited for a regular leadoff spot. The spot is surely meant for Jeter, with his usual .380-.400 OBA and decent speed.

As for the Marlins, they now hold the strange distinction of never having won a division title, but never having lost a postseason series (6-0). I haven't followed them much, and I don't think they'll be a favorite to repeat next year, but they had a great run in the playoffs. It remains to be seen whether the team will be gutted again, or whether they will try to build on this season and make a serious effort to win again.

As I did for the last round, I'll review my previous picks:

• Yankees over Red Sox in 6 (actually Yankees over Red Sox in 7): Well, my excuse here is that I did actually think in terms of ability level it would go 7 games. But I had a feeling the Yankees were going to win, by "destiNY" or whatever, and I didn't want to pick against Pedro Martinez, who was going to start game 7 for the Red Sox.
• Cubs over Marlins in 6 (actually Marlins over Cubs in 7): In about 99 out of 100 parallel universes, this actually happened. Unfortunately for a nation of Cubs fans, ours is the universe that saw the Cubs lose in bizarre, dramatic fashion.

For the World Series: Yankees over Marlins in 6. This is kind of unoriginal, but I think the Yankees' experience and pitching will win out in the end. Their rotation is a little out of whack after Clemens, Wells and Mussina all pitched yesterday, so they might get off to a slow start. I think the home-field advantage will definitely help New York here. All this being said, I've already picked against the Marlins twice, like a lot of other people, and I wouldn't be shocked if they pulled off an upset. There's always the possibility of a letdown on the part of the Yankees after defeating their biggest rivals in an incredible LCS.

I will definitely be watching...

Time to review my last round's playoff predictions:

• Yankees over Twins in 5 (actually Yankees over Twins in 4). Not a terrible prediction, Minnesota just didn't generate any offense.
• Red Sox over Athletics in 5 (got that one right). Pedro was not totally dominating last night but he was good enough to win. I watched the last five innings of the game and it was just great, dramatic baseball.
• Giants over Marlins in 3 (actually Marlins over Giants in 4). I didn't see that one coming at all. A couple of the games were really close, and that series could easily have gone the other way. Florida avoided Bonds and got some stellar play from Ivan Rodriguez.
• Cubs over Braves in 5 (got that one right). As I thought, Wood and Prior were just too tough for the Braves

For the championship series round

• Yankees over Red Sox in 6. I know I predicted the Red Sox would win the World Series, and along with almost everyone else, I'd like to see a Red Sox / Cubs series (the "someone's got to win" series), but I think Boston is going to be too worn out to beat the rested Bronx Bombers. The Red Sox rotation is out of whack, and the relief corps is shaky. New York has four aces and the witch in the bullpen, who seems perfectly comfortable working two innings. The defense on both teams is pretty weak, so that should be a tossup. Boston's offense is a bit better but New York is always capable of scoring runs.
• Cubs over Marlins in 6: I have to respect the Fish after they took it to the Giants. I don't know what they're doing, but they're doing it well. I think Prior and Wood will continue to shine and Florida will run out of last minute heroics.

And my new World Series prediction is Yankees over Cubs in 6.

It is really late, but I should do my baseball playoff predictions:

• Yankees over Twins in 5: Yankees have the experience and usually dominate the Twins. Thing is the Twins are always tough in the Metrodome and Johan Santana and Brad Radke have been lights-out in the second half.
• Red Sox over Athletics in 5: I think this is Boston's year to win it all, but it might well come down to game 5 with Pedro Martinez on the mound, a matchup that's always in Boston's favor. I expect the Athletics' awesome pitchers to somewhat neutralize the potent Boston offense, but I think Oakland's average hitting will catch up with them.
• Giants over Marlins in 3: The Giants have been taking it easy for a couple of weeks now but showed they are still sharp, taking of 3 of 4 from Los Angeles in the last weekend. Florida had a nice run to make the playoffs but I think the Giants, led by their All-World leftfielder are just too strong for them.
• Cubs over Braves in 5: This might be a bit of a surprise pick but I'm going on a hunch here. Atlanta underachieves in the postseason and Kerry Wood and Mark Prior are great.

Predicting past the first round is usually a waste of time, but my crystal ball says Red Sox over Giants in 6 for the World Series.