Week 12 - Box 2.0, Where Profits Flow Like Beer in Boston

In my last post, I mentioned the idea of taking the financial scores and rankings and applying them to close spread games as a way to mitigate some of the volatility of pick 'em type games. Well, I did just that, and the results were so surprisingly strong I had to check it three times over.

What I've done here is take the statistical model from Box 1.0 and made a hybrid with the financial scores to hopefully avoid the low spread games. The games this effects are any games with spreads under 3 - basically, games that fall below a "normal" win and come down to the wire. Vegas tends to think of home field advantage as worth 3 points (or slightly less depending on the field). For instance, if DEN had been favored by 3 against TEN last night, Vegas is saying that on a neutral field, the game would basically be a pick 'em. (Note: Historically, the 3 point homefield advantage is incorrect - home teams have displayed a 1 point advantage at home in the past 3 years). It would also follow that if a team were favored by LESS than 3 and playing at home, they would actually be underdogs at their opponent's field. These are the games that the statistics have a hard time deciphering, since often 3 or less points can be decided by chance. What the financial scores do is eliminate some of the guesswork.

I've gone through and highlighted the differences in profit using financial rankings in close games in the following charts. Note that financials are not available to use prior to Week 3, since you need a minimum number of games to begin using the algorithms. The differences are staggering performance-wise:

	Box 2.0		Box 1.0
	Total Ret		Total Ret		Profit Differential
Week 1	53.50%		53.50%		0.00%
Week 2	92.73%		92.73%		0.00%
Week 3	15.59%		15.59%		0.00%
Week 4	95.29%		39.87%		55.42%
Week 5	-22.19%		-70.95%		48.76%
Week 6	-8.00%		-8.00%		0.00%
Week 7	38.86%		67.56%		-28.70%
Week 8	-11.66%		-11.66%		0.00%
Week 9	89.69%		89.69%		0.00%
Week 10	92.07%		69.04%		23.03%
Week 11	15.44%		-67.60%		83.04%
Total:	451.32%		269.77%

To put this in perspective, I needed to view it as dollars, so here is the running bankroll highlighting the above:

Box 1.0
$s Bet	$s Back	$s Not Bet
$50.00	$76.75
$50.00	$96.37	$26.75
$50.00	$57.80	$73.12
$50.00	$69.94	$80.91
$50.00	$14.53	$100.85
$50.00	$46.00	$65.37
$50.00	$83.78	$61.37
$50.00	$44.17	$95.15
$50.00	$94.85	$89.32
$50.00	$84.52	$134.17
$50.00	$16.20	$168.69
Total Bankroll:		$184.89
Starting Bankroll:		$50.00
Gain/Loss:		269.77%

Against this:

Box 2.0
$s Bet	$s Back	$s Not Bet
$50.00	$76.75
$50.00	$96.37	$26.75
$50.00	$57.80	$73.12
$50.00	$97.65	$80.91
$50.00	$38.91	$128.56
$50.00	$46.00	$117.46
$50.00	$69.43	$113.46
$50.00	$44.17	$132.89
$50.00	$94.85	$127.06
$50.00	$96.04	$171.91
$50.00	$57.72	$217.94
Total Bankroll:		$275.66
Starting Bankroll:		$50.00
Gain/Loss:		451.32%

The most important thing to note is actually not the increased winnings, but the decreased losses. There is nothing worse than losing virtually everything you bet, as was the case in Weeks 5 and 11. I don't know about you, but I hate losing. I would much rather lose 22% than 71% in Week 5. This past week, had I accounted for financials, the top five picks would have looked like this:

1. WAS +10.5 v. DAL Away	44.0
2. HOU -1 v. NO @Home x2!	40.0
3. MIA +10 v. PHI Away - Bet Soon	40.0
4. NYG -2.5 v. DET Away	41.0
5. OAK +5 v. MIN Away x2!	38.0

Suddenly, the heavy NO pick switches side to a HOU pick, and the DET dog pick becomes a bet on a strong NYG team. Both games cover, HOU v NO is still a double bet (and thus a double winner), and I end the week +15% instead of -68%.

Look out for more in depth analysis of this weeks games today after the Football Outsider DVOAs are released and I calculate this week's "What Would Wall Street Do" power rankings.