What fun would draft week be without looking at some data? In that spirit we decided to look how draft position affects the willingness of a team to play a given player. In other words, does the sunk cost of the draft pick encourage a team to give a player more playing time than the player’s skill might otherwise call for? We looked at data for running backs and quarterbacks. Also, for quarterbacks we looked at when the playing time was occurring: was it during mop up time in a blowout or was it when the outcome was still in question.
Are running backs drafted higher more likely to “get a shot” than players drafted lower?
Median Rushes in Season
|Round 1, First Half||215.5||96.51|
|Round 1, Second Half||114.5||100.04|
|Round 2, First Half||125.5||98.49|
|Round 2, Second Half||71||79.24|
|Round 3, First Half||122||72.11|
|Round 3, Second Half||62||57.55|
|Round 4 and Above||14||52.17|
The running backs selected in the first half of the first round definitely get a better chance at playing time. There is an oscillation when comparing the first half of a round to the second half of a round. This is probably a result of the fact that teams picking in the second half of the draft probably need to give a rookie running back playing time.
The distributions of rushes demonstrate this even better. Running backs selected in the first half of the first round are very likely to get more than a 100 carries in their rookie season (as demonstrated by the peak above 100 in the graph below).
The question remains: do the running backs selected in the first half of the first round perform better than those selected in the second half of the first round or in the beginning of the second round?
|Round 1, First Half||3.86||0.65|
|Round 1, Second Half||4.16||1.24|
|Round 2, First Half||3.96||0.56|
|Round 2, Second Half||3.99||1.19|
|Round 3, First Half||4.85||0.97|
|Round 3, Second Half||4.26||1.02|
|Round 4 and Above||3.64||0.90|
In a word, the answer is no. The performance is no better for a rookie selected early in the first round than later rounds (in fact it might even be worse). The median is lower for this group and is in line with the performance of all non-rookies (labeled veterans). Also, the second half of the first and second rounds has running backs with significantly bigger variance in yards per carry. This could be the effect of better offensive lines in the teams that had better records, and thereby picked in the latter half of a round. In this case variance is your friend if you are picking in the draft: if you want a median running back, you can find a veteran with that performance for likely less than the cost of a first round pick. The density plot of veterans’ yards per carry is peaked at the same level as RBs selected early in the first round, but with thinner tails.
An interesting analysis would be look at yards after contact per rush as a way to control for offensive line abilities.
Data from first year appearing in an NFL game
|Round 1, First Half||377||176.34||6.38||1.99|
|Round 1, Second Half||143.5||181.55||6.48||1.20|
|Round 2, First Half||109||230.61||4.78||1.52|
|Round 2, Second Half||28||65.63||5.11||2.73|
|Round 3, First Half||218||82.02||6.10||0.71|
|Round 3, Second Half||79||143.19||5.55||2.54|
|Round 4 and Above||27.5||103.53||5.19||3.32|
Looking at the table above, it appears that quarterbacks selected early in the first round are given a better shot at playing time than quarterbacks selected later in the draft. The median number of passes for QBs selected early in the first round is more than 2.5 times the number of passes for QBs selected later in the first round. This is very clear on the figure below: the density plot for the first half, first round QBs is peaked at a much higher number of passes than any other rounds and indeed the distribution of that of non-first year players (veterans).
Note: Only two quarterbacks were selected in the first half of the 3rd round in the time period under study. That is why there is no density plot for that part of the draft.
As the rounds progress, the number of attempts per player goes down significantly. Are these quarterbacks just as good, or are they not “getting their shot”? Based on the yards per attempt (YPA) they do not appear to be as good, at least in their first year. This can be seen from the table above, or from the density plot below of Yds per attempt. For example, the early first-round QBs not only have a higher median YPA, but they also have a better chance of outperforming the median as the tails in their distribution are skewed to the right (towards better YPA) more than any other graph besides veterans.
The question remains, however, are these first-year quarterbacks playing in meaningful situations in meaningful games? To see this, we have plotted the distribution of Leverage when the QB threw a pass. Leverage is defined as the % change if likelihood of winning a game if the team were to score a touchdown. That is, the higher the leverage, the more important the game situation. For instance, in a tied game with less than 10 seconds remaining, a touchdown can changes the likelihood of winning percentage from about 50% to near 100%. When the leverage is near zero, the game is a blowout one way or another. In most games, the average leverage for a given pass play is 20% (this can be seen in the density plot for veteran QBs). The plots above are for pass plays in the first 8 weeks of the season. This part of the season was looked at because in the first half of the season, all games are still meaningful (no one has been eliminated from playoff contention). Note that there is a blip near zero leverage on all of these graphs. This is a result of teams switching to pass-heavy play calling when they are far behind. Similarly, the blip in high leverage situations is likely the result of teams running a pass-laden hurry up offense in a close game with little time remaining.
The distributions for the first round quarterbacks, regardless of position in round, seem to follow that of veteran QBs. This suggests that these quarterbacks are playing in typical game situations. In later rounds, starting with the late 2nd round picks, we see a spike in passes with leverage near 0%. These are passes thrown during mop-up time: i.e., when one team has a large lead. This could explain some of the findings for YPA in the later-round QBs: these players are throwing passes when their team is far behind and the passing plays being called are high risk/high reward type plays.
The data used from this analysis covers the last 13 years the NFL and was sourced from ArmchairAnalysis.com.