Monday, 2 June 2008

Counter-point

Far be it for me to display a contrarian side, but I can't resist the opportunity provided by my partner to take issues with one of those mental habits that come so easy but which, in my opinion, lead to false conclusions.

That being, the tendency to analyze a baseball team or player's performance by what calendar month it occurs in. I understand this temptation because any site which hosts sports stats offers us an easy breakdown of results by month. But there is NO logical reason to assume that the turning of the calendar page has anything at all to do with what happens on the field.

The actual truth is that players and teams go through a number of "phases" (feel free to use another word there if it suits you) in the course of a season and it is sheer coincidence if those phases ever sync up with the beginning or end of a month.


In the case of the 2008 Jays, we have seen at least 4 phases so far. The two recent losses may or may not be the beginning of a fifth - you can't really tell immediately when you are in a new phase. Observe:

April 1 - April 15:
Record 8-6;
RPG - 5.43
RAPG 3.71
Dif: +1.72

Actually, one could argue that in this phase the Jays got unlucky a couple of times along the way. that big a differential is probably more worth of 10 or 11 wins than 8.

April 16 - April 30:
3-11
2.86
4.29
-1.43

As you can see, there is an astounding difference between the first 14 games and the second 14 - what logical sense does it make to put all those games in the same pile and try to analyze the whole simply because they both took place in April?

May 1 - May 12 (game one):
6-5
2.64
3.73
-1.09

The reason we won more here than the differential would indicate is because 12 of the 41 runs in this phase occurred in one game. in the other 10 we allowed 2.90 runs per game which is close enough that the record isn't that far off.


May 12 - May 30:
14-4
4.89
2.78
+2.11

If you include the last two games in those figures (since it's unclear if we are in a new phase yet) you still get a record of 14-6, 4-65-2.85 (+1.8). And in both cases the runs against is skewed by the disastrous start of David Purcey but I don't feel like arguing for excluding that from the sample set because it's so impressive anyway.
Again I move the previous point - who should we dilute the outstanding results of 18 games by marrying it to the mediocre results of the previous 11 simply because they occured in the same month?

One could argue, perhaps, that phase 2 and phase 3 are all the same period because of the similarly pathetic offensive output but the half run better pitching (and the better results) seem to me to indicate a potential divide. It seems to me patently obvious that Phase 2 was a period when almost everything went freakishly wrong. I prefer to assume Phase 3 was different but I would not argue against the other position.

That internal debate aside, I strongly maintain that the coincidental timing of Phase 3 and Phase 4 within the same calendar month tells us exactly NOTHING about the Jays offense or overall performance.

If you take the Jays offensive output over the past 20 games and extrapolate it to the 59 games they have played in all, they would have 274 runs right now - good enough for third in the AL. We have far fewer reasons to assume this is a flukishly high output for this group of players than we do to assume the 25 games with pathetic offense was an abberation.

I, for one, would be driven to madness by the improbability that the Jays hitters would return in June to the level that they achieved during the late offensive drought.

I cannot of course argue that the jays will continue to win 3 out of 4 games, but I will argue that this offense is perfectly capable of producing between 4.5 and 5 runs a game the rest of the season and should be expected to be reasonable observers.

~WillRain

3 comments:

Unknown said...
This comment has been removed by the author.
Unknown said...

As someone who has read dozens upon dozens of history books seeking an answer to the question of whether interwar European diplomacy should be treated as a whole or broken into pre and post-Hitlerian halves, I can appreciate the logic against looking at the season in one month blocks.

But then again the problem with setting arbitrary dates for informal phases of the season is one of sample sample size. In a 6-month season, I think 6 1-month blocks is about the least bad means of evaluating progress. It's a good chunk of games, enough to really say something about the way the team's going, without forcing us to wait too long to reach some preliminary conclusions.

The way I looked at it, the offense was consistently bad through two months and the pitching went from really good to outstandingly awesome. Offensively, I think there might be some room for minor improvement with the guy's we've got, but I don't think we get into, say, 4.5 runs per game territory sustainably for the duration of the season without clearing out some of the obvious dead weight and bringing in a Big Scary Bat. Much less intelligently, I think that's kind of what Griffin was getting at, too.

If our offensive expectations for June are reasonably predictable, the pitching we got in May is clearly unsustainable, it was just too incredible.

I'm not going to go all Wilner and insist that I'm right and you're wrong until I'm blue in the face. I just think we NEED that bat as soon as it becomes available. Actually, probably a week earlier.

The Southpaw said...

1. I DO agree this level of pitching is unsustainable. If we regressed to .35 runs allowed per game it would still be an amazing staff.

2. I agree that the current lineup is flawed - most particularly with the presence of Wilkerson and that to be followed by a presumably weakened Wells. The addition of Jason Bay or some reasonable facsimile IS imperative.

3. That said, I do think the offense of the last 20 games (and the first 14) is MUCH CLOSER to being an authentic representation of this lineup than that of the 25 intervening games. Furthermore, I would argue that when your given month in baseball ranges from 26-30 games, a sample size of 20 or 25 games is not dismissable as too small.

4. I would agree that the delination of "phases" is arbitrary to the opinion of the analyst, but I would argue that if an analyst fails to make his case by using the chosen break point, then that flawed claim still doesn't justify calendar divisions.

5. Another way to do it is, of course, by "Event"...how many wins since Wells was hurt? How many since Rolen took the field? etc. But in my experience there is no firm pattern that such events mark a turning point.

~Will