FOOTBALL RPRs & PROJECTIONS
What are Relative Performance Ratios?
One of the principal problems with evaluating statistics for collegiate sports is that the teams play such varying schedules. As a result, one can never be sure whether a stat represents a team's performance, or simply reflects the quality of its schedule. "Relative Performance Ratios", or "RPRs", are an attempt to solve this problem by comparing what a given team has done to what its opponents have collectively allowed in other games.
For example, if a team's offensive "Yards per Rush" RPR is 1.15, it means that team has rushed for 15% more yardage on each attempt than its opponents have allowed in their other games. Similarly, if a team's defensive "Yards per Pass Attempt" RPR is .85, it means the team has held its opponents' passes to 15% less yardage per attempt than those opponents have gained in their other games.
This system has clear limitations. It assumes that each team's opponents' opponents are of roughly the same quality, which is often not true. Two teams' opponents' opponents do, however, tend to be much more similar than the teams' opponents, simply because there are more of them. After week 5 of the season, for example, most teams will have played 4 opponents; those opponents will have played 12 other teams. Overall, I think RPRs can be a valuable tool in assessing the relative strengths of college teams. They provide an objective basis for subjective analysis of teams' performance.
The best way to view these tables is to open them directly in a spreadsheet program such as Lotus 123 (which was used to create them), Microsoft Excel or Quattro Pro. That allows the tables to be sorted using the spreadsheet's functions.
- RPR - Offense
- RPR - Defense
- RPR - Returns, Fumbles, Penalties
- RPR - Calculated Statistics (Pass Efficiency, Turnovers, 14/3 Estimates)
Projections
Using RPRs and the basic stats published by the NCAA, I prepare projections of each week's Pac 12 games. These are NOT predictions of what will happen, but are instead straight algebraic extrapolations of what would happen if (IF!!!) both teams were to play as they have, on average, in earlier games. That "IF" is crucial, because teams never play exactly as they have before. These projections make no attempt to evaluate morale, injuries or the teams' game to game improvement. If these aren't predictions one can take to one's bookie, what good are they? I use them as an objective basis for further subjective analysis of a game, and in that role, they serve quite well.