Part II of our investigation on making a line with the "First Down Conversion" numbers

The following research is based on the new rating system discussed in the Accurate First Down Conversion article in the main research section, and is a follow-on to the first review of the predictive value vs the spread.

In the 2002 season, the numbers were a solid 57% on all games from week five forward, and an especially powerful 61% against the line from week ten on. Given this, we deemed it a worthy stat for subsequent research and we have gone back and compiled the data for a three more seasons -- 1999, 2000, and 2001.

We used the same formula for projecting a score off of the "moving the ball" statistics, which comes down to assuming each 1% of net rating is worth one point of spread value.

Comments: well, it's above break-even at 54.3% overall, but somewhat discouraging in that the higher range overlays were strictly 50/50 terrain, and the profitability was driven by one category -- home underdogs.

Comments: the 2000 season was solid at 56.4% overall, and dominating in the high overlays. In addition, the "moving the ball" picks sparkled in the smallish favorite matchups, where neither side was giving more than 4 1/2 points.

Comments: ah, and just when things were starting to look pretty good, along comes the 2001 season to put a major dent in this new approach. A woeful 46% overall effort with terrible performance in the high overlay classes does not breed confidence. At the same time, there are those (including some of us here at TwoMinuteWarning) who believe that the 2001 season was an aberration in terms of results, brought about by the September 11th catastrophe and the ensuing war. Indeed most of our statistical measures suffered their worst season by far in 2001. The games played in the early part of the season may have been effected by the global events, dismantling the value of using 'season-to-date' data as we are doing in this test. On the other hand, who is to say whether the play of NFL games was really any different.

The other area of interest for us was the week ten on timeframe, and rather than reproduce all the individual charts again, we'll give you the summary for the four years we have examined:

Year	Overall	Week 10+	Comments
1999	54%	56%	small overlays, home teams in wk10 on
2000	56%	59%	large overlays, home teams in wk10 on
2001	46%	48%	bad all round
2002	57%	61%	small overlays
99-02	54%	56%	home teams 59% week ten on

Three out of four years look pretty good, one year looks really bad. Certainly the method has hit only 49% in games played from week five through nine, which alone suggests it isn't a tool to be relied upon until the later stages of the season. There's some more research to be done, especially in terms of coming up with a more sophisticated projection algorithm, but the 2002 rein of wondrous results appears to be an exception rather than the norm.

Another footnote to add is that we were excited when for the '02 campaign, the top two regular season teams came out to be #1 Tampa Bay and #2 Oakland. For 1999, the Super Bowl champion Rams were indeed top rated, but the AFC's representative, Tennessee, was #14. For 2000, Baltimore was #1, and the New York Giants were the top NFC team (#4 overall). Finally in 2001, it was the Rams again on top by these ratings, and New England #10 (5th best AFC team).

That means that out of eight Super Bowl teams, the 'true first down' numbers nailed six of them at the start of the playoffs -- good going by any standards!