This is a "sequel" to this thread:
I gathered some more data on this, focusing specifically on the percentage of "very good" starts rookie QBs have made since 2004, in comparison to their percentages of "poor" starts.
I defined a "very good" start as a start with a QB rating of 90 or higher, and a "poor" start as one with a QB rating of 69 or lower. Starts with QB ratings between 70 and 89 were not used in any of these analyses.
Here are the data:
QB %age of Very Good Starts %age of Poor Starts Franchise QB? Career QB Rating Tannehill 47 47 Luck 31 25 Wilson 69 19 RGIII 67 7 Weeden 27 40 Newton 44 25 Yes 86.2 Dalton 44 38 Yes 87.4 Bradford 25 38 Yes 82.6 Palmer 38 46 Yes 86.2 Roethlisberger 62 23 Yes 92.7 Ryan 56 31 Yes 90.9 Flacco 44 31 Yes 86.3 Ponder 30 60 Yes 81.2 Stafford 20 60 Yes 82.8 Gabbert 14 14 No 70.2 Sanchez 27 33 No 71.7 Young 15 46 No 74.4 Leinart 36 36 No 70.2 Freeman 22 56 Yes 79.8 Gradkowski 27 64 No 65.8 Orton 7 67 No 79.7 Edwards 22 55 No 75.5 Clausen 20 60 No 58.4 Walter 0 75 No 52.6 Smith 14 86 No 79.1 EManning 29 71 Yes 82.7
The thinking here is that a QB's future "ceiling" may be revealed in how often he's able to perform at a very high level during his rookie season, whereas his future quality of play overall may be revealed in how he minimizes poor play as a rookie while maximizing very good play.
As it turns out, the correlation between the percentage of very good starts as a rookie and career QB rating is 0.70, which is strong.
The correlation between the percentage of poor starts as a rookie and career QB rating is -0.37 and is comparatively weak, and that variable will therefore not be used as a basis for any analyses in this thread.
What the data also reveal is that the average percentage of very good starts among the future "franchise QBs" is 37.6, with a standard deviation of 13.78.
The average percentage of very good starts among the future "non-franchise QBs" is 19.7, with a standard deviation of 9.63.
Ryan Tannehill's percentage of very good starts (47%) places him within a standard deviation above the average of the future "franchise QBs," while also placing him nearly three (!) standard deviations above the mean of the "non-franchise QBs."
So, the take-home message is that Ryan Tannehill, in terms of his percentage of very good starts (QB rating of 90 or higher), which was 47%, played much more like a future franchise QB as a rookie than a future non-franchise QB.
I think this lends support to people's perceptions that Ryan Tannehill appears to have the makings of a franchise QB based on how often he was able to play at a high level this year.
For the sake of further comparison, Chad Henne in 2009, in his second season in the NFL (first as a starter), following far more college starts at QB, had a 90 QB rating or higher in ony 23% of his starts, which puts him less than a standard deviation above the mean of the non-franchise QBs, and more than a standard deviation below the mean of the "franchise" QBs.
In other words, with regard to this particular stat, Tannehill played like a future franchise QB this year, whereas Chad Henne in 2009 played like a future non-franchise QB. I think you could also argue that Ryan Tannehill played even better as a rookie than Chad Henne did this year for Jacksonville, as a fifth-year player.