More research into the baseball win distributions:

I looked at the difference distribution again, and it appears that the difference distribution follows the normal curve. I wasn't looking at the standard deviations yesterday, which I should have.

Furthermore, I compared teams in the top 25% of expected wins, bottom 25% of expected wins, and the 25th-75th percentiles. Basically, I wanted to determine if the difference distribution was dependent on a team's expected win total.

The standard deviations for these groups were as follows (the overall SD was 9.7):

E(W)	SD
more than 88	9.3
74-88	10.4
less than 74	8.6

These are statistically significant differences. And contrary to my earlier beliefs, there is more variance in the middle of the pack.

So here is an updated list of the teams that generate a 20% profit rate when using the Baseball Prospectus expected win totals, a normal difference distribution with the above SD in each of the 3 scenarios, and the Vegas odds listed here.

Team	BP E(W)	Vegas E(W)	Diff	z = Diff/SD	Probability	Odds	Expected Profit %
LAA	83	92	-9	-0.87	81%	1.83	48%
NYY	95	102	-7	-0.75	77%	1.80	39%
ATL	82	89	-7	-0.67	75%	1.83	37%
FLO	80	85	-5	-0.48	68%	2.00	37%
COL	72	68	4	0.47	68%	1.90	29%
OAK	88	78	10	0.96	83%	1.53	27%
HOU	78	83	-5	-0.48	68%	1.80	23%
KC	69	65	4	0.47	68%	1.80	22%

Some notes:
-If the difference is negative, then we'd bet on the under, and vice versa.
-The probabilities of winning our bet are based on the z value and the normal distribution table.
-The expected profit % is found by multiplying the probability with the odds, and subtracting 1.

But no, I will not be betting on the under for the Angels.