With all the hoopla surrounding Josh Freeman and the back and forth of his season, as well as the attempts to place him historically against other quarterbacks, I figured a look at other quarterbacks in similar situations may help us get a better idea of where Freeman is as a player and where he may be headed. (To view all graphs, you can set your viewing to "Wide" as opposed to "Narrow" or you can click on links at bottom)
In our previous looks at Freeman, we've covered completion percentage and yards per attempt and how each class of quarterback fared as well as the additional groupings we researched. In this edition of our article Part II, we'll begin to explore the fantastic world of quarterback ratings. The main question we are trying to answer is with Freeman having no data beyond his rookie year, where can we project him in future years? By utilizing the 16 quarterbacks we selected and the historical seasons of quarterbacks from 1970-2009, we will try and place Freeman into a range by using quarterback rating.
If everything pans out, we will be able to come up with a statistical idea as to where Josh Freeman will end up in Years 2 and 3.
Just a reminder on a few points. I broke the quarterbacks into two classes based on career arc. These quarterbacks will be singled out and compared to the NFL mean by utilizing data from the past 40 years of NFL seasons. This also will not be an all-inclusive look, but merely an attempt to project Freeman. Sample size issues and segregating data are the two biggest challenges we face.
Lets first look at all of the data at hand. This will give us an individualistic view of each quarterback and their quarterback rating through their first three years of starting. For reference sake, I don't know what the mean NFL quarterback rating is, but I'd venture it's in the 75-80 range based on the number of seasons I've looked at with all quarterback types and seasons.
Name
Category
QBR (1)
QBR (2)
QBR (3)
Year 1 to 2
Year 2 to 3
Year 1 to 3
Ben Roethlisberger
Starter
98.10
98.60
75.40
0.50
-23.20
-22.70
Kerry Collins
Starter
61.90
79.40
55.70
17.50
-23.70
-6.20
Joey Harrington
Starter
59.90
63.90
77.50
4.00
13.60
17.60
Eli Manning
Starter
55.40
75.90
77.00
20.50
1.10
21.60
Donovan Mcnabb
Starter
60.10
77.80
84.30
17.70
6.50
24.20
Tom Brady
Starter
86.50
85.70
85.90
-0.80
0.20
-0.60
Drew Bledsoe
Starter
65.00
73.60
63.70
8.60
-9.90
-1.30
Peyton Manning
Starter
71.20
90.70
94.70
19.50
4.00
23.50
Kyle Boller
Backup
62.40
70.90
71.80
8.50
0.90
9.40
David Carr
Backup
62.80
69.50
83.50
6.70
14.00
20.70
Alex Smith
Backup
40.80
74.80
57.20
34.00
-17.60
16.40
Charlie Batch
Backup
83.50
84.10
67.30
0.60
-16.80
-16.20
Daunte Culpepper
Backup
98.00
83.30
75.30
-14.70
-8.00
-22.70
Byron Leftwich
Backup
73.00
82.20
89.30
9.20
7.10
16.30
Gus Frerotte
Backup
61.20
70.20
79.30
9.00
9.10
18.10
Brian Griese
Backup
75.60
102.90
78.50
27.30
-24.40
2.90
NFL Average
-
69.85
74.99
74.88
5.14
-0.11
5.03
Sample Set Mean
-
69.71
80.22
76.03
10.51
-4.19
6.31
Starter Mean
-
69.76
80.70
76.78
10.94
-3.93
7.01
Back up Mean
-
69.66
79.74
75.28
10.08
-4.46
5.61
St Dev All
15.52
10.61
10.90
11.96
13.52
16.33
Starter SD
14.96
10.77
12.41
8.90
13.73
17.22
Backup SD
17.09
11.16
9.95
15.06
14.25
16.53
Now that we've dumped our data into plain view, lets start to assess what it all means. The most constant fact and situation I've seen is the average increase in every stat from years 1 to 2, which is followed by an a drop or flattening of that data in years 2 to 3. We have the same scenario when looking at quarterback rating. The numbers I'll give you are in order from NFL mean, starter mean to back up mean. The mean movement from years 1 to 2 was 5.14, 10.94, and 10.08. Again, we see a nice movement up, particularly in our sample classes, and a somewhat smaller move by the NFL mean. This is all fine and well, and what we presumably expected to see.
Years 2 to 3 is the same as previous stats, but still somewhat confusing. The average move was -.11, -3.93 and -4.46. A drop from year 2 to 3 seems to be standard and almost expected. This leaves the change from rookie year to third year as 5.03, 7.01 and 5.61, a positive movement, but somewhat tempered by the movement from year 2 to 3.
To look at the graphical representation between NFL quarterback sample, our starter sample and backup sample, lets view the all too familiar scatterplot.
This gives us the individual plotting of each quarterback in our sample. To look at a comparison of the means, lets break out another bar chart.
You can see from this that our groupings outperform the NFL mean, but stay relatively close in years 1 and 3. The variance in year 2 can probably be attributed to the selection of quarterbacks I took and the small sample size. It should be noted that of the criteria I set forward, only 47 quarterbacks fell within their parameters. This is over a 40 year time period, so in terms of sample size, while 47 isn't a gaggle of individual data points, without widening the parameters, there isn't a much bigger sample we could take.
In previous editions we've looked at the ebb and flow of each of our quarterbacks in years 1 through 3, but for this article, we'll take a break from that exercise. I'm not sure it provides us with any additional insight into the performance, it's merely a visual representation of the data table posted above.
By reviewing our data, we can see that the starters outperformed both other groups by small margins, beating the backup mean by approx 1.50 points. To project Freeman, we again turn to our friends known as mean and standard deviation. The mean for the NFL in years 1 through 3 looks like so; 69.85, 74.99, and 74.88. Then, by taking the standard deviation, we are able to project Freeman into a range based on expected turnout from normal distributions.
Lets first hit the year 2 data. The 68% certainty range (one standard deviation) gives us 60.33 to 89.67 as the range. Freeman, and most other quarterbacks in year 2 could be reasonably expected to fall within this range. The higher certainty range (95%) produces a wider spectrum of 45.68 to 104.32. It is almost assured that Freeman would fall within that range given our criteria and assumptions.
Year 3's ranges are a bit closer together. 62.55 to 87.21 give us our one standard deviation measure while 50.22 to 99.54 encompasses the two standard deviation range.
With Freeman checking in at a quarterback rating of 59.8 (towards the low end of year 1, one standard deviation) we know that while he is due for some sort of uptick in year 2, it is not guaranteed for him to follow the means we have laid out. All things being equal, he should continue to regress towards the mean, which would see an approximate jump of 5.14 points in year 2 and a total of 5.11 from years 1 to 3. This is not encouraging news. While he could certainly outperform those expectations, that puts him at a quarterback rating of 64.9 in year 3. This is not what we want to see. While the improvement would be welcome, we want to see bigger jumps than what we are projecting. He would have to outperform the mean over the next few years to get in the 70's in regards to quarterback rating, which still isn't very good.
This is the first piece of disheartening news on Freeman. While year 2 hasn't started and the knowledge that he could very well outperform the mean or underperform it, based on what we see, he shouldnt be counted on for a huge jump forward. The best we could hope for, if he holds true to these numbers is a Drew Bledsoe like career, minus the injury. Relatively poor in the first few years, and then leaps forward. Of course, he could also go all Donovan McNabb on us and just rocket towards the top of the ranges set forth. One thing is certain, we don't know exactly where he is headed, but have reason to believe that this measurement of his game will not be one we want to acknowledge as we tout him as a franchise quarterback.