So I just finished adding my financial modeling to the system. It now calculates bastardized versions of Alpha, Beta, and the Sharpe Ratio. The base for my equations are as follows:
Moving Ave of Points Scored = "Portfolio Return"
Moving Average of Points Across the League (week by week) = "Market Return"
Historical Average of Points Scored (historically, the average team scores 20 points) = "Riskless"
Using that as my basis, I made a few models. First, I made a "Super Ranker", ranking the original box picks with the new measures of volatility and return to rerank taking into account the financial modeling. The ranks end up being pretty much the same, with slight variations in the middle.
The more interesting piece has turned out to be the "suggested pick" in the matchups week to week. For instance, the picking the team with the better Sharpe Ratio (ie, team that beats the benchmark more consistently with less risk) doesn't always mean picking the team with the statistical edge. So I get lots of mixing and matching. When the suggested picks agree with the statistical picks, hypothetically I should get my best results. I haven't back tested (I figured 3am on a work night was late enough), but here are some of the layouts this week:
Sharpe Ratio Picks - Sharpe measures returns over risk (volatility)
Alpha Picks - Another way to measure returns over risk.
Which leads me to this - the Box's GENIUS SUPER PICK AVERAGING MACHINE (or G-SPAM):
Now these are some picks I can get behind. What the financials seem to do really well is pick WINNERS, not against the spread. What the stats do very well is pick SPREADS. Combining the two has given me lower risk, high return version of the spread picking machine - G-SPAM. I will be doing a combo of pick happiness this week, with some wacky moneyline parlays and
interesting spread combos. I will post that here tonight. Ain't nothing but a G-SPAM thing, baaaaby...