I was bored, so I decided to try to come up with a formula that would create a power ranking system based simply on certain statistics. Such an approach would be an unbiased statistic for determining overall team strength.
Team judging formula
In my opinion, there are four important things, when taken relative to the rest of the league, that should be used for judging how talented a team is: record, offense, defense, and strength of schedule. I chose to prioritize them as such:
ranking = 45(record) + 15(offense) + 15 (defense) + 25 (strength of schedule)
So now I had to make some definitions about what exactly "record", "offense", "defense", and "strength of schedule" meant. Here is what I came up with:
Record was pretty easy to define. Simply taking a win percentage would work:
record = wins / games played
For offense and defense I wanted to get how teams fared overall, not whether or not they were over-preforming or under-preforming (since that usually is accurate displayed in record). So I decided to use offense yardage and defensive yardage as the judging stat. Also, one must realize in power rankings, you are ranking teams relative to eachother. Therefore, using their team ranking seemed like the best course (and I wanted the 1st ranked team to equate to "1" with 32nd lowest ranking team equate to "0"):
offense = (32 - off.rank) / 31
defense = (32 - def.rank) / 31
Finally, strength of schedule is pretty easy to define, but hard to implement. This will take a look at wins/losses of opponents, while trying to exclude ones own games. For example, when the Ravens (4-1) beat the Steelers (3-1) it is not fair to punish the Ravens for the 1 Steeler loss. Likewise, teams that lose should not benefit from the resulting win. So I had to use trimmed strength of schedule:
opp.wins = sum of all opponent's total wins
opp.games = sum of all opponent's games
trimmed opp.wins = opp.wins - losses
trimmed opp. games = opp.games - games
strength of schedule = trimmed opp.wins / trimmed opp.games
Because all record, offense, defense, and strength of schedule are relative of each-other, there is a limited amount of points available to distribute and the rankings should average out to 50. It will actually occasionally be higher than 50 because of ties in offensive and defensive ranking, but the idea is that there is a limited number of points in the formula the teams compete for. I took the liberty of doing formulaic power rankings for this season (which at the moment completed Week 5), and power rankings for the 2009 season, for my readers' enjoyment to see how the formula works out. I also will likely include them from now on in my weekly power rankings. I actually think they came out to be pretty accurate:
Week 5 Formulaic Power Rankings
- Pittsburgh (71.0)
- Baltimore (69.7)
- Atlanta (69.0)
- New York Giants (68.7)
- New York Jets (66.3)
- San Diego (62.2)
- Indianapolis (61.3)
- Kansas City (60.0)
- Chicago (58.1)
- Green Bay (56.4)
- Philadelphia (55.8)
- New England (55.6)
- New Orleans (55.2)
- Dallas (54.9)
- Denver (54.8)
- Miami (54.4)
- Houston (52.9)
- Tampa Bay (52.1)
- Cincinnati (49.2)
- Washington (47.5)
- Tennessee (46.7)
- Minnesota (43.8)
- Oakland (43.6)
- Arizona (43.4)
- Jacksonville (41.8)
- St. Louis (39.6)
- Cleveland (36.7)
- Seattle (32.3)
- Detroit (32.0)
- San Francisco (28.3)
- Carolina (25.5)
- Buffalo (19.6)
2009 Formulaic Power Rankings
- Minnesota (68.9)
- Dallas (67.8)
- Green Bay (67.8)
- Indianapolis (67.1)
- New England (64.7)
- San Diego (64.5)
- New Orleans (63.7)
- Pittsburgh (62.3)
- Philadelphia (61.9)
- Baltimore (61.4)
- Houston (60.4)
- New York Jets (58.8)
- Cincinnati (57.2)
- New York Giants (56.8)
- Denver (56.1)
- Carolina (54.1)
- Arizona (52.9)
- Atlanta (50.7)
- Miami (46.4)
- San Francisco (45.0)
- Tennessee (44.8)
- Chicago (44.0)
- Jacksonville (43.5)
- Washington (40.2)
- Buffalo (37.7)
- Seattle (36.0)
- Oakland (31.7)
- Kansas City (29.8)
- Tampa Bay (28.5)
- Cleveland (28.3)
- Detroit (23.6)
- St. Louis (21.0)
No comments:
Post a Comment