This is a follow-on to our earlier Turnover Articles. If you did not read the original analysis we suggest you first see Turnover Difference Revisited.

We hope all our subscribers are well aware at this point of the importance of turnover difference in handicapping games. As our research has shown, by backing teams with a significantly worse turnover record coming into a game you would have fared very well against the spread.

An idea that occurred to us in the midst of our motivational research was to try and combine the G-A-P (Good-Average-Poor) theory of rating teams with the turnover difference play situations. For our research we elected to use the following standards:

• Good teams have won 60% or more of games coming in
• Average teams have won 40% to 59% of games coming in
• Poor teams have won less than 40% of games played

ANALYSIS: As would be expected with a series of plays that have hit roughly 57% over twenty years, there are a lot of positive areas. Indeed two of the nine pairings produce a 60%+ record (Average-Poor and Poor-Poor), and there are five home/away specific cells that hit the magic mark or better:

Good home vs Poor away: 75%
Poor home vs Poor away: 69%
Poor away vs Average home: 66%
Good away vs Average home: 62%
Average home vs Average away: 61%

Another breakout would be to compare the teams based on their GAP category and class the turnover play pick as better - same - or worse:

This is quite striking in the instance where you have teams with the same rough record/class. In those cases, when the turnover pick is a home team it's been a spectacular 63% cover over the years. Otherwise it's not especially revealing save for the better team as home turnover pick which perhaps not surprisingly shows a below 50% mark as the teams are likely heavy favorites in most cases.

Some final comparisons would be to simply group teams by the home class or away class:

Here you could make a case that matchups where at least one of the two participants comes in with a poor record are the most promising from a turnover difference standpoint. However, when you have something that hits roughly 57% without any additional rules or refinements, it would take a really strong sample to make us feel you should move away from simply favoring all the turnover keys in preference for favoring those with a slightly higher historical return.

What's impressive is that the turnover difference theory works across almost all cross-sections and sub-sections of the data. Still, since there were a few areas of promise, what we will do is look at the results from the 2001 season (which wasn't included in the above tables) and see if there is any sign that the particular niches held up better than the norm. We'll be back shortly with the results.