I have a question for those of you who put together projections and rankings for IDP players: Do you "normalize" historic data when you use that data to make this year's projections?
I know "normalize" is a pretty vague term, and I think what I am asking is whether you try to distinguish skill from luck particularly when it comes to sacks and interceptions.
Taking sacks as an example, and using Merriman's stats from last season, he had 17 sacks in 12 games. The league average team gave up 36 sacks last year, while the teams Merriman played gave up an average of 41 sacks. Thus it appears that he was a "little" lucky - around 2 sacks as a result of playing against poor offensive lines (or bad QBs, etc.). I would think you could use the adjusted 15 in 12 games as a baseline to compare against this season's schedule - I have not looked, but I know the DD has a strength of schedule for each position, so I assume someone has done the legwork for this part of the analysis.
Interceptions would be a little more difficult to discern - and hence the hardest to predict with any accuracy. But, I think you can break down the causes of an interception into 4 categories: 1) Bad throw by the QB; 2) Bad route/hands by the WR, 3) tipped ball at the line of scrimmage, or 4) skill of DB. Identifying how many fit into category 4 (by default), should result in identifying the "safest" DBs - if Int's are a big score opportunity. Once a player has a skill, he is likely to duplicate it, you then have to figure in the variances for the other factors - no small task. But again, you can look to the schedule to determine the frequency of 1 and 2 for historic data, and factor those projections into the DB's projections.
I realize it is probably more work than the potential value of identifying big play DBs, but I wondered if any of the prognosticators went to such lengths on the non-tackle stats. (It seems that I have read a few threads that suggest that the best predictors of basic tackles are scheme and position, but do you also factor in the strength of schedule?)
Sorry for such a convoluted question - just curious more than anything.
Thanks!!
I know "normalize" is a pretty vague term, and I think what I am asking is whether you try to distinguish skill from luck particularly when it comes to sacks and interceptions.
Taking sacks as an example, and using Merriman's stats from last season, he had 17 sacks in 12 games. The league average team gave up 36 sacks last year, while the teams Merriman played gave up an average of 41 sacks. Thus it appears that he was a "little" lucky - around 2 sacks as a result of playing against poor offensive lines (or bad QBs, etc.). I would think you could use the adjusted 15 in 12 games as a baseline to compare against this season's schedule - I have not looked, but I know the DD has a strength of schedule for each position, so I assume someone has done the legwork for this part of the analysis.
Interceptions would be a little more difficult to discern - and hence the hardest to predict with any accuracy. But, I think you can break down the causes of an interception into 4 categories: 1) Bad throw by the QB; 2) Bad route/hands by the WR, 3) tipped ball at the line of scrimmage, or 4) skill of DB. Identifying how many fit into category 4 (by default), should result in identifying the "safest" DBs - if Int's are a big score opportunity. Once a player has a skill, he is likely to duplicate it, you then have to figure in the variances for the other factors - no small task. But again, you can look to the schedule to determine the frequency of 1 and 2 for historic data, and factor those projections into the DB's projections.
I realize it is probably more work than the potential value of identifying big play DBs, but I wondered if any of the prognosticators went to such lengths on the non-tackle stats. (It seems that I have read a few threads that suggest that the best predictors of basic tackles are scheme and position, but do you also factor in the strength of schedule?)
Sorry for such a convoluted question - just curious more than anything.
Thanks!!
