Marginal Utility

by Ken Houghton

CNBC Headline: Fixing the Credit and Mortgage Markets.

*Regular readers know my answer is "Not later than February of 2007."

Labels: efficient markets, financetheory

by Ken Houghton

If you haven't figured it out yet, I'm distracting myself from the vagaries of Health Economics and the trenchant, profound political "analysis" of the ProgLib Blogsphere—not to mention tomorrow's memorial—by going through the data from Coate (2008), available in PDF form here.*

I've found a minor mistake in the Coate data; at least two games allocated to the Chicago White Sox (AL) are actually Cubs games (NL). So the data presented below may have minor differences from the previous three posts.

One of the things about making a bet is that there are two possible motivations: one might be betting for a team, or one might be betting against the other team. So it might be worth looking (again, excluding bets involving the Reds) at which teams have large differentials (in both absolute and result basis). So, as a first run:

There is a difference between "0" and a blank in the above data. Using the example of the Atlanta Braves, Rose bet for the team 12 times: he won six and he lost six, leaving a net of zero (0).

By contrast, and following up to the issue raised in a previous post regarding his previous teams, there is NO indication in the data that he bet FOR the Montreal Expos, so there is a blank or null in that field.**

Also, in some cases, one might believe that a team has a significant advantage due to playing at home due, for instance, to the shape of the ballpark and the team developed to play in it.*** So, especially in the absence of other information, there might be a bet placed on a home team with such a park. Following is the bets on a general basis, broken down by league and home/road team being favored. (Again, the Reds are excluded.)

More ongoing.

*CREDIT UPDATE: Some data in this portion available from Baseball-Reference.com and/or Baseball Almanac.

**Similarly, he never bet against the Toronto Blue Jays, betting on them ten times during the period, winning seven and losing three for a net of +4. Other than the Canadian teams, all of the zeroes on the summary table indicate the net result of an even number of bets.

***There are teams that have had a significant advantage playing on the road, of course, but such advantages generally appear later in the season.

Labels: baseball, efficient markets, Rutgers

by Ken Houghton

In order to keep from discussing the Obama campaign's methods here—the Hilzoy/De Long axis keeps trying to tell us they are pure, and the commenters here appear to agree; why let matters of fact* get in the way?

So let's look a little more at the gambling activities of one Peter Edward Rose.**

The simplest way to answer the question most recently posted is to look at his cumulative returns. While we know the ultimate answer must be significantly negative, the accumulation is frightening in its consistency***:

After April 11th, Rose doesn't even approach break-even. There is a brief "winning streak" from the 15th to the 18th, and a fairly solid positive trend in American League bets from 15 April to 4 May (14 nonnegative days in 17 days of bets), but the time from May 5th to the 12th produces $30,000 in losses (three days—May 6, 8, and 9—lose $26,000, almost all of which is due to bad AL bets that more than negate the net $21,000 gain of the AL streak).

On the surface, on balance, and with few exceptions, the only obvious trend in the data Coate presents on Rose's betting is that he fails to consider the primary rule of betting: if you can't find the sucker at the table, the odds are good it is you.

*Link added here, as it was deleted at the original discussion, though a peculiar manifestation of The Obama Rules remains in response to it.

**Again, all data from Coate (2008), available in PDF form here. Dollar amounts do not foot for the National League; presumably this is an error of mine.

***All of the following data excludes the Cincinnati games, since we are working on the presumption, supported by the evidence presented, that Rose would not have bet against the Reds. The overall effect upon the data of adding those games does not appear as if it would be significant.

Labels: baseball, efficient markets, Rutgers

by Ken Houghton

Following up to the previous post, some more details.*

1. On an absolute basis, any presumption that Rose's working within the National League should give him special insight is not borne out by the data. Indeed, he performs notably better when betting on the American League:
   

   
   
   

2. His winning and losing splits almost perfectly. If he bets $2,000 or less on a game, he is likely to win. Bets of over $2,000 appear more likely to lose.**
   

   
   
   

3. The combination is disastrous.
   

   
   

Overall, the high-level data appears consist with a gambler whose instincts and information are near-perfect contraindicators.***

More later.

*All data from Coate (2008), available in PDF form here. Dollar amounts do not foot for the National League; presumably this is an error of mine.

**It would be prudent to assume that bets of $2,200 consist of a $2,000 and the 10% "vig" charged to the loser. However, the data indicates five (5) bets of $2,200, two (2) of which are winning amounts.

***Another example of this may be his consistent betting in favor of the Phillies, one of his former teams, which had a major P&L impact, as shown before:

but there is not a similar effect for his other former team, the Montreal Expos.

Labels: baseball, efficient markets, sports

by Ken Houghton

I walked past Douglas Coate's office a couple of times yesterday. Had I seen this paper (PDF) (h/t Tyler Cowen via Portfolio), I might have stopped in.

Curiously, especially if you exclude CIN games,* it appears that Rose was a better gambler when he didn't have access to information (i.e., better overall betting on the AL than the NL).

UPDATE: The distribution is interesting. Assuming the data is correct, The Philadelphia Phillies are responsible for 29% of his losses over the period. Indeed, Philadelphia, Detroit, and Houston account for over 60% of the dollar losses.

And to no one's great surprise, the same teams, along with San Diego, are the teams where Rose bets frequently and (presumably) the wrong way, with the Phillies again setting the standard:

Maybe more tomorrow.

*Which makes sense, since the consensus is that Rose always bet those, and only to win. Though the April 25th entry of "no bet" may be indicative something else.

Labels: baseball, efficient markets, Rutgers

by Ken Houghton

Headline on CNBC scroll: "IS $100/BARREL OIL JUSTIFIED?"

Labels: Economics, efficient markets, Market Failures