QB Stats and Future Performance: Part I
You are viewing free content provided by FantasyGuru.com. Why not consider subscribing today?
by Mike Horn, Staff Writer
Published, 3/26/14
Most statistics are descriptive. They tell us a lot about what happened. They may even explain why it happened. They are good about the past but maybe not about the future; in other words, they may not do a good job of telling us what will happen. But if we’re trying to pick a fantasy team for 2014, stats that tell us a lot about 2013 are not necessarily useful. So I spend a good deal of time trying to find stats that help me put together a 2014 team – statistics that are predictive.
A couple of QB stats that I’ve found to have some predictive value in the past are Completion Percentage (Comp%) and Yards Per Attempt (YPA), when taken together. I’m going to take a fresh look at those stats in this article to try to predict what QBs could be in store for a big 2014.
I’ll use the period from 19882012 to examine how those stats help predict the nextyear performance (that brings us up to 2013) of Top10 fantasy QBs. I start at 1988 so that I’m only looking at consecutive 16game seasons, and avoid the strike era of 19821987.
For fantasy scoring, I’m using 25 pass yds = 1 Fantasy Point (FP); 10 rush yards = 1 FP; passing TDs = 4 FP, and rushing TDs = 6 FP. No turnovers are included and the odd receiving yards/TDs that are accumulated by QBs in this period are scored the same as rushing yards/TDs (no PPR).
I limited the QBs to those who had at least 250 passing attempts in one year (I’ll call this the base year or “Y”) and then played the next year (“Y+1”). I’m also going to use fantasy points per game (FPG) rather than total FP to reduce the impact of injuries in Y+1 – while injuries probably affect FPG, they have less impact on that number than on total FP.
I want to start by acknowledging that base year FPG do a decent job of predicting next year FPG. We all know that things will change from yeartoyear, but there are a lot of things that don’t change, as well. Look at this graph:
This shows the FPG of all the QBs in the study, with “base year (FPG)” along the xaxis and “next year FPG” on the Yaxis. The gray lines extend out from 15 FPG on both axes – 15.0 is the average base year FPG of those QBs.
We can see that there is a general pattern – the more FPG in Y, the more FPG in Y+1. Of course there are outliers, but the dotted black line shows the linear trend. We can also see that while about half the diamonds are on each side of the 15 FPG line for the base year, more than half of them are below the next year 15 FPG line. That is because, on average, these QBs only managed about 13 FPG in Y+1. The typical QB loses about 2 FPG from yeartoyear. The decline is some combination of aging (although this is partly offset by young players improving), injury, and regression to the mean.
By “regression to the mean,” I’m saying that some of the QBs in this study played above their natural ability in the base year, got a lot more attempts than their talent justified, and then came back to earth the next year. This is a fancy way of saying “he had a career year” or “fluke performance” – think Derek Anderson in 2007.
What I want to do is to use Comp% and YPA to help us predict which fantasy starting QBs will remain startable in Y+1. For this I’m going to focus in on QBs who ranked in the Top 10 in base year FPG. Top 10 guys are typically clear starters; after that, you may be more willing to play matchups with your QBs.
First, let’s look at the connection between Comp% and YPA:
On this graph, the gray lines represent the averages for both stats: 59% Comp% on the xaxis and 7.0 YPA on the yaxis. Again, the dotted black line is the trend, showing the average relationship between the two stats. Since YPA depends in part on Comp%, not surprisingly the stats are (positively) correlated – better Comp% means better YPA.
Here’s the same graph but now, the Top 10 QBs in FPG in the base year are shown as green or red diamonds. The green indicates that they also ranked in the Top 10 in Y+1 while the red means they dropped out. The roman numerals indicate the four quadrants of the graph. Notice that the green diamonds are heavily clustered in the “I quadrant”. The next graph may show it more clearly by eliminating the other diamonds:
Generally, good QBs have above average Comp% and YPA, and most starting fantasy QBs are good reallife QBs (I’m deliberately not discussing about the running component of FPG here).
Looking at the red diamonds, they tend to be a little lower and to the left (worse Comp% and lower YPA) than the green diamonds. If we look at the overall data in a table:
Category

Comp %

YPA

Top 10 Both Years

62.2%

7.6

Didn't Repeat Top 10 Rank

60.4%

7.4

Not Top 10

58.3%

6.7

QBs who were in the Top 10 in FPG in both Y and Y+1 had the best completion percentage and Y/A of these the categories. Those in the Top 10 in the base year but not the next year were in the middle. And QBs who didn’t make the FPG Top 10 in the base year had the worst stats, which shouldn’t be a surprise.
Now let’s look at the table by quadrant:
Quadrant

Top 10 Both Years

Didn't Repeat Top 10 Rank

Top 10 Both Years (%)

Didn't Repeat Top 10 Rank (%)

2013 QBs

I

89

50

64%

36%

See graph

II

16

20

44%

56%

Stafford

III

9

12

43%

57%

Luck

IV

14

20

41%

59%

Grand Total

128

102

56%

44%

The first column lists the four quadrants indicated on the graphs, Quadrant I is the High Comp%/High YPA quadrant. Then there are the numerical counts of the QBs in the sample that fall into the Top 10 in FPG in both Y and Y+1, and those that failed to repeat their Top 10 performance. The next two columns calculate the percentages of the repeaters/nonrepeaters. The last column shows what quadrants the Top 10 QBs in FPG in 2013 fall into.
Overall, there are 230 QBs in the sample, with 128 (56%) of them returning to the Top 10 in Y+1. Since I’m looking at a 25year period, you might wonder where the other 20 QBs went – they either fell below the 250attempt minimum or didn’t play in Y+1 after a Top10 finish in the base year.
The key point here is that 64% of the Quadrant I QBs who managed a Top 10 FPG rank in the base year repeated their fantasy performance, while less than half of the base year “Startable” QBs in the other quadrants did so. That looks to me like having both above average Comp% and YPA are strong indicators of future fantasy success.
What does this mean for the Top 10 QBs in 2013? Well, eight of the 10 were in Quadrant I, as you can see from the next chart (Manning of course is Peyton, not Eli). Those eight have a higher probability of ranking in the Top 10 again in 2014.
The outliers are Andrew Luck and Matthew Stafford. This analysis indicates they have, at least historically, a less than 5050 chance of ranking in the Top 10 again. The best thing in their favor is that Quadrants II and III have small sample sizes.
What if we tighten the criteria for being in Quadrant I?
Now I’ve placed the gray lines one standard deviation above the averages for both stats: at 63.7% for Comp% and 7.7 YPA. A lot fewer QBs fall into Quadrant I. Here are the numbers:
Quadrant
(Based on
+1 Standard Deviation)

Top 10 Both Years

Didn't Repeat Top 10 Rank

Top 10 Both Years

Didn't Repeat Top 10 Rank

2013 QBs

I

33

12

73%

27%

Brees, PManning, Rivers, Foles, Rodgers

II

21

19

53%

48%

III

13

9

59%

41%

Romo

IV

61

62

50%

50%

Dalton, Newton, Luck, Stafford

Grand Total

128

102

56%

44%

Instead of having over half of the QBs in Quadrant I (149, see the first table), now about half are in Quadrant IV (123). The future success rate of Top 10 QBs in Quadrant I is very good. Quadrant III is a little above average, although that is a bit of a small sample. Here’s the chart with the 2013 QBs in the adjusted quadrants:
What I make of all of this:
· If I want security from my QB choice, a QB who is likely to finish in the Top 10 again, I’d go with one of these five: Manning, Brees, Rodgers, Rivers, or Foles, in no particular order.
· This analysis really supports Foles as a fantasy choice for 2014.
· Meanwhile, it downgrades Luck, who is probably going to be one of the first five QBs off the board. These numbers suggest Stafford also is probably going to be overdrafted.
· The stats don’t think too highly of Dalton’s chances to repeat in the Top 10, which I think is consistent with where his very early ADP has him being drafted.
· The real bargain identified by these numbers may be Rivers, who has an ADP around Dalton’s, but with what appears a much higher chance of being back in the Top 10.
Two key things this analysis ignores: the running component of fantasy scoring and age. Luck got 20% of his fantasy scoring from rushing yards and TDs last year. Newton got 29% and Foles 15%; the rest were under 10%. One of the things I need to do is examine if the percentages are different for running QBs vs. pocket passers.
Also, this ignores the effects of aging: possibly positive for young QB (Foles, Luck, and Newton all were only 24 last year, Stafford was 25) and possibly negative for older players (Peyton was 37, Brees was 34).
Finally, there are lots of other changes to the teams – players and coaches – going on around these QBs. Two stats based on last year’s performance cannot hope to capture that. They are just one indicator of future performance. I wouldn’t make a decision just on this analysis of what QB to draft for 2014 – but I also wouldn’t ignore the red flags these stats are waving at us either.
Instead, using these stats to build a target list and analyzing the situation around the players seems to be a very strong way to ensure a big performance from your fantasy QB in 2014.
Bad draft picks are down 2,906%. Must be the Guru!
Back to the top