|
| |
The Story...
Why the Geek Sheet?
After spending many years tireless researching teams, trends, angles, power ratings,
etc while surveying the litany of various “experts” out there and their mostly
conflicting and confusing games of the week, games of the month, and games of
the year, not to mention their conference game of the months, conference game
of the years, underdog of the month, underdog of the year, blah, blah,blah….the
one thing I realized was certain, was to continue that course of action, was to
accept consistent losses. I rationalized there had to be a better way. |
|
| |
Instead of focusing on picking winners by searching out trends and angles, I shifted my focus to accurately determining what the score should be, then focus on those games with the highest level of reliability in those redictions. Then determining of those games which had the largest variance from the current betting line. I surmised if I could devise a system where I could chart the accuracy of these forecast, I could (relatively speaking) accurately assess the odds of my predictions proving out (when combined with large variances in the lines produces winners). |
|
|
 |
|
| |
Our focus here at the Geek Sheet is to focus in on not only what these predictions are, but what is the sum of the results. Outputted results are only the product of the information inputted. We determined that if we could create a system that analyzed the results of systems calculated from independent criteria, that when similar results were derived, they stood a higher chance of being more indicative of what the expected result should be. |
|
| |
For instance if I surmised that an independent set of criteria (a,b,c, and d) created an expected result of x, while another independent set of criterion (1,2,3, and 4) created an expected result of y, it would be safe to assume that these two outputted results are the sum of nothing more than their inputted criteria, and should consistently yield high variations in their outputted results when compared to each other. Now what happens when you look at the two outputted results as compared to themselves, and relate it to football?
Assume Team A is playing Team B. System X will value the relative strengths and weakness of Team A and Team B according to its own set of criteria; it may value the Run D higher than the Pass D, or the Run O less than the Pass O. Additionally, it may (within its system) value efficiency in running the football inequitably higher than efficiency in passing defense. All theses different nuances, could easily force the outputted results to be skewed/flawed in some way. |
|
| |
Now assume System Y has it own set of independent criteria, perhaps flawed in its own way as well. Going back to our example of Team A and Team B, let's assume System X believes Team A will score 20 points against Team B and System Y assumes Team A will score 35 points against Team B. System X assumes Team B will score 14 points against Team A, and system Y assumes Team B will score 18 points against Team A. What does that mean? Perhaps it means nothing; two random sets of input data produce two random (and independent) sets of output data. Perhaps it means more. Perhaps it means that when these two independent sets of input data produce similar results, there is a higher probability the similar outputted data will prove correct when played out in real life.
Back to our example, and what we believe it means. For starters, we think it means that we can not, with any level of predictability, determine what the expected result will be when Team A's offense is on the field. |
|
| |
However, we do believe that the similarity in the outputted results of Team B Off v Team A Def, yields some consistency that should be evaluated further. Or more simply stated that Team B's offense and/or Team A's defense is consistent enough in some form of their past results, or most likely it is the combination of the two, that two separate systems, with independent sets of criteria, predict very similar results. |
|
| |
This is the foundation of our analysis. From this very basic premise we can filter the week's offerings to those games where we feel there is a higher probability of predictable outcome, and focus intently on why we believe that to be so.
Once identifying those opportunities, we then evaluate where the line values are. We don't ignore situation analysis, trends, angles, or even common sense. But we make the above assessment the foundation of our research, from which we then focus on the specifics of the game. We have found this approach to be extremely successful, and are constantly working to refine our methods, and analyze what makes up acceptable alignment of models and variances from line based on the lines themselves. |
|
|
|
|
|
|
