Creating this thread for the periodic posting of what we feel are legitimately useful Angles (some use the term "Trends").
Whazzat? you ask.
A 'legitimate' Angle is one that meets a definition with as many explainable correlations as possible.
Perhaps I could explain more clearly by defining what I think are "illegitimate" Angles.
A common I.T. is one that has just one or two corralary variables.
For example:
TEAMS playing on Sunday nights give the following RESULTS.
That's not much for a Legitimate or Angle. There's one common variable, but it's pretty much random since all teams play on Sunday night, so there's always a flat result of 15 wins and 15 losses. (And average scoring is always flat given a few weeks results)
How about TEAMS playing on Sunday nights following a Win/Loss on Saturday?
Well, now you have two variables, but can we really correlate the results without more?
How about a couple of specific teams that avg about .500 records, but who on Sunday nights for past two seasons, just 20 wins and 40 losses?
Now we have three variables...Day of week, same team(s), very odd contrast to overall record.
But the more common variables the better.
And the more we can define a legitimate correlation between these variables, the more legitimate a Trend (I prefer the term "Angle") we have to work with.
Whazzat? you ask.
A 'legitimate' Angle is one that meets a definition with as many explainable correlations as possible.
Perhaps I could explain more clearly by defining what I think are "illegitimate" Angles.
A common I.T. is one that has just one or two corralary variables.
For example:
TEAMS playing on Sunday nights give the following RESULTS.
That's not much for a Legitimate or Angle. There's one common variable, but it's pretty much random since all teams play on Sunday night, so there's always a flat result of 15 wins and 15 losses. (And average scoring is always flat given a few weeks results)
How about TEAMS playing on Sunday nights following a Win/Loss on Saturday?
Well, now you have two variables, but can we really correlate the results without more?
How about a couple of specific teams that avg about .500 records, but who on Sunday nights for past two seasons, just 20 wins and 40 losses?
Now we have three variables...Day of week, same team(s), very odd contrast to overall record.
But the more common variables the better.
And the more we can define a legitimate correlation between these variables, the more legitimate a Trend (I prefer the term "Angle") we have to work with.
