Working on developing a YPPT (yards per point) based system for selecting CFB games vs the spread. Using YPPT for a team and factoring in avg yds gained for its offense and yards allow by the upcoming opponents defense, it estimates a score for the game (think adjusted Dudley method).
Next it factors in an adjustment to account how well a team's offense has done verses its prior opponents as well as how well its defense has done vs prior opponents. For example, if team A offense is generating more YPPT than what was expected vs it's prior opponents defense over the season, team A's score is elevated accordingly for an upcoming game. The same evaluation goes into team A's defense to adjust the score they are expected to "give up". The same analysis is done for the upcoming opponent (team B). These prior performance factors for team A and team B are then used to adjust the estimated YPPT score mentioned in the above paragraph.
Next, a YPP (yards per play) factor (minor adjustment) is cooked into the mix similar to the YPPT adjustment described above. Yards per play becomes less leveraging than YPPT as seasons progresses. A high YPP and low YPPT could mean a team turns it over a lot and vice versa.
The data used in the calculations accounts for a team playing home or away. If team A is home, data for their prior home games is used in the calculations (not all their games). Similarly, if team A is away, data for their prior away games is used.
Finally, we look at if a team has strengths in areas that capitalize on an opponents weaknesses. For example if team A offense has a fantastic YPR (yards per rush) this season and their opponents defense has a poor YPR average, team A may have an advantage. Of course you have to account for who each team has played earlier in the season vs how they were estimated to play..
Obviously, early in the season there is not a lot of data available. It takes 4, 5, or 6 games to have enough data to start to see solid information and as always key player injuries must be taken into account (not done in this analysis directly - this is incorporated subjectively).
To most betting professionals (of which I am not) this is fairly obvious analysis but I'm attempting to tweak the various factors to optimize the results and possibly add in a few other twists.
Also, live by the general betting rules (not always correct but over the long haul they yield results):
- Good running teams play better at home
- Good passing teams play better on the road
- Bet dogs late in the week since the public bets the favorite as time goes on (vice versa-bet Favs early)
- Bet home dogs vs the spread. Don't bet road Favs vs the spread.
- In a shootout, take the points.
I'll post picks based on this process and we'll see if it pans out. It should, theoretically, become more accurate as the season progresses.
If it doesn't, I'll drift away into oblivion like so many other prognosticators.
Tanko
Next it factors in an adjustment to account how well a team's offense has done verses its prior opponents as well as how well its defense has done vs prior opponents. For example, if team A offense is generating more YPPT than what was expected vs it's prior opponents defense over the season, team A's score is elevated accordingly for an upcoming game. The same evaluation goes into team A's defense to adjust the score they are expected to "give up". The same analysis is done for the upcoming opponent (team B). These prior performance factors for team A and team B are then used to adjust the estimated YPPT score mentioned in the above paragraph.
Next, a YPP (yards per play) factor (minor adjustment) is cooked into the mix similar to the YPPT adjustment described above. Yards per play becomes less leveraging than YPPT as seasons progresses. A high YPP and low YPPT could mean a team turns it over a lot and vice versa.
The data used in the calculations accounts for a team playing home or away. If team A is home, data for their prior home games is used in the calculations (not all their games). Similarly, if team A is away, data for their prior away games is used.
Finally, we look at if a team has strengths in areas that capitalize on an opponents weaknesses. For example if team A offense has a fantastic YPR (yards per rush) this season and their opponents defense has a poor YPR average, team A may have an advantage. Of course you have to account for who each team has played earlier in the season vs how they were estimated to play..
Obviously, early in the season there is not a lot of data available. It takes 4, 5, or 6 games to have enough data to start to see solid information and as always key player injuries must be taken into account (not done in this analysis directly - this is incorporated subjectively).
To most betting professionals (of which I am not) this is fairly obvious analysis but I'm attempting to tweak the various factors to optimize the results and possibly add in a few other twists.
Also, live by the general betting rules (not always correct but over the long haul they yield results):
- Good running teams play better at home
- Good passing teams play better on the road
- Bet dogs late in the week since the public bets the favorite as time goes on (vice versa-bet Favs early)
- Bet home dogs vs the spread. Don't bet road Favs vs the spread.
- In a shootout, take the points.
I'll post picks based on this process and we'll see if it pans out. It should, theoretically, become more accurate as the season progresses.
If it doesn't, I'll drift away into oblivion like so many other prognosticators.
Tanko
