Today, I looked into my new calculations, based on stats from previous years, and learned that I not only needed to change the colors for them, but that, with two tweaks, I could make a calculation that appears to be ridiculously good - at least for Week 1.
Don't believe me? Here are the Week 1 results:
| Calculations | | Home and Away Advantage | | Benchmarks |
Actual Winner | Official | Upset Probable | 3 Years | 2 Years | Gut | 2015 | 2014-15 | 2013-15 | | Football Locks | NFLPower Rankings |
DEN | CAR | No | DEN | DEN | CAR | CAR | DEN | DEN | | CAR | CAR |
TB | ATL | No | TB | TB | ATL | ATL | ATL | ATL | | ATL | TB |
MIN | MIN | No | MIN | MIN | MIN | MIN | MIN | MIN | | MIN | MIN |
PHI | PHI | No | PHI | PHI | PHI | PHI | PHI | PHI | | PHI | PHI |
CIN | CIN | No | CIN | CIN | NYJ | CIN | CIN | CIN | | CIN | CIN |
OAK | OAK | Yes | NO | NO | OAK | OAK | OAK | NO | | NO | OAK |
KC | KC | No | KC | KC | KC | KC | KC | KC | | KC | KC |
BAL | BUF | No | BAL | BAL | BUF | BUF | BUF | BUF | | BAL | BUF |
HOU | HOU | No | HOU | HOU | HOU | HOU | HOU | HOU | | HOU | HOU |
GB | GB | Yes | GB | GB | GB | GB | GB | GB | | GB | GB |
SEA | SEA | No | SEA | SEA | SEA | SEA | SEA | SEA | | SEA | SEA |
NYG | NYG | No | DAL | NYG | NYG | NYG | NYG | NYG | | DAL | DAL |
DET | IND | No | DET | DET | IND | IND | IND | IND | | IND | DET |
NE | ARI | No | NE | NE | ARI | ARI | ARI | ARI | | ARI | ARI |
PIT | PIT | No | PIT | PIT | PIT | PIT | PIT | PIT | | PIT | PIT |
SF | LA | Yes | SF | SF | SF | LA | LA | LA | | LA | LA |
Notice anything? Yeah, me too. The "2 Years Calculation" only got one game wrong during Week One. The Week Two numbers are not nearly as impressive, but overall, it is 79% accurate, so far. As for what is going on, the 3 Years and 2 Years calculations both factor in the general accuracy of my calculation during the last three years (2013-2015) and two years (2014-2015), respectively. I originally called the 3 Years one the "Modified Calculation".
The calculations previously called "Historical" were actually based on the win rate for home and away teams, so the names and colors were swapped to match the color themes for like predictors (orange, yellows, and browns for anything related to home or away team and blues and purples for calculations). And since the two year calculation for the Home and Away Advantage (H&AA) calculations was the most successful so far, I made the "2 Years" calculation to pull from that data. I changed a modifier for the upset percentage - using the same value that is used in the H&AA calculations - first, then sat back and waited for the spreadsheet to process.
When it was all done (I have a lot of dependent calculations and conditional formatting, and it gives my poor Chromebook a run for its money), I couldn't believe my eyes. Time will tell if this is an anomaly, or if I stumbled across something truly great.