Tracking the DCTO performance by year

The following research is based on the new predicted scores system discussed in the combining drive chart predictions with turnover difference theory article in the main research section.

To review, the new formula for projecting scores against the spread showed a very strong 57.8% success rate from 1998 to 2002 looking at all games from week five and on.

That's nice, but one good way to look at the results (other than testing against fresh data which we essentially did with the '98-'02 run since the fomula was created in advance of the test) is to break out the performance by year and see if there is much consistency on a yearly basis or if the overall positive results were spiked by one or two extreme seasons.

Comments: right in line with the overall numbers, and favorites were solid even with the year being so strongly skewed towards the dogs. The turnover difference had a solid year (37-26 in weeks five to twelve, although 24-24 the rest of the way), while the drive chart projections were 52.3% on the year. The DCTO numbers are 93-63 with a minimum 2 point overlay.

Comments: very similar overall results although the favorites struggled. This was a season when the turnover difference was phenomenal, clocking a 61% record from week five through the end of the regular season. The drive chart projections were a modest 52.4% on the year overall, thus showing the benefits of combining the two schemes.

Comments: in 2000 the drive chart projections had a 56% season overall whereas the turnover difference suffered its worst season in over a decade, with a lousy 46% record during the prime week 5 to week 12 timespan. Combining the two methods into one had a pleasing 56% record across all games, albeit this was driven by the low overlay categories.

Comments: another healthy season for the DCTO picks, which isn't shocking given that the turnover difference was 54% with the T/O selections, and the drive charts were also a profitable 54% overall. What's nice about this set is that favorites and underdogs were both worthy predicted spread-beaters.

Comments: much like the other years, the DCTO results are excellent. This season even saw the ideal progression of the higher the overlay the better the results! The standard drive chart predictions hit 56% on the year all told, while the turnover difference was 55% in the weeks 5 to 12 zone.

Summary: it would be hard to ask for a better year to year consistency, as a string of 59% - 59% - 56% - 58% - 57% (from oldest to most recent) is admirable. The overlay ranges didn't unfortunately hold up as well, since excluding the low overlay picks would have seriously hurt your seasonal results in 2000 and 1999, although improving it in other years. As a trend, home teams are better plays than away teams, and underdogs are stronger than favorites, but when you have something that seems to work fine overall, looking for additional ways to spike the predictiveness may be tantamount to being greedy.

Needless to say, the DCTO predictions will be a very visible part of our 2003 coverage.