How NFL Teams Value Positions

Using the conventional draft value point chart, we can evaluate the relative weight that teams place on each position. By summing the draft points "spent" on players at certain positions we can rank the relative importance of each position in the draft. Here is a list of each position and how many total draft points have been spent on them based on the '99-'06 drafts for all 7 rounds.

It's an interesting list because high-profile positions such as QB or RB are not at the top of the list as we'd expect. Many other analysts make a chart like this each year, but there is a flaw with ranking positions this way.

A far better measure of the relative value placed on each position in the draft would be one that takes into account the number of players needed at each position. We can allow for the difference in the number of players needed in two ways. First, we can divide total points by the number of players of each position typically on a roster. Alternatively, we can divide by the number of players of each position typically on the field.

Here is the table of position draft pick value per player on the field:

Here is the table of position draft pick value per player on a typical roster:


Expectedly, "singular" positions such as QB and RB move up the lists because there are fewer of them required and "multiple" positions such as WR or CB move down the list. A lot of conclusions can be drawn from these tables, but what surprises me most is that RBs are valued as highly as QBs. Conventional wisdom says that solid QBs are so much harder to find than solid RBs, but it appears teams choose to spend roughly the same amount of draft points on those positions.

Perhaps the "typical" roster has more than 3 RBs or less than 3 QBs, but I'd suspect the true average number of roster spots isn't that far off. Plus, we get the same result when accounting for players on the field, and it's been a long time since we've seen 2 RBs on the field simultaneously.

"Day 1" Massey-Thaler Comparisons

Teams could theoretically rack up "Massey-Thaler" surplus value with lots of bottom-round picks, but teams can only carry so many players. Rare exceptions aside, players from the top rounds are the ones that make a difference in the fortunes of a team. For every Tom Brady (6th round) there are several Mannings, Rivers, Palmers, Roethlisbergers, etc.

Let's compare total conventional draft values, M-T values, and their differences for each team for the first 3 rounds only, the "Day 1" picks. The rankings are listed below:

There are some exceptions, but we see generally successful teams at the top of the list and generally unsuccessful teams at the bottom. It's a rough analysis, but it leads me to believe there is some level of correlation between maneuvering for M-T surplus value and team success on the field. Keep in mind, bottom-feeder teams with the top picks have more M-T surplus value than winning teams with the later picks. Therefore, if there truly is a significant correlation between wins and M-T value, the direction of causation would be from M-T value to wins, not vice-versa.

Playing the Massey-Thaler Game

Are any teams actually aware of the phenomenon proposed in the Massey-Thaler Study? Perhaps one or more NFL teams has been conscious of the true costs of various draft picks and have maneuvered to maximize the value of their picks with respect to both performance AND cost.

To investigate I ranked each team's M-T surplus values and each team's conventional draft values for the last 8 drafts. To see if any team or teams stand out as "M-T aware," I subtracted the M-T ranking from the conventional ranking. The resulting number indicates how well a team has converted conventional value into M-T surplus value. For example, if a team was #32 in conventional value but was #1 in M-T value, the resulting difference of 31 would indicate that team is successfully trading picks to convert low conventional vlue into high M-T value. Low (negative) numbers would indicate the opposite, that a team is "wasting" its conventional draft position without earning much M-T surplus value. Here is the list:

One team does stand out at the top. Tennessee is certainly an outlier at the top of the list with a difference of 20. Admittedly, this is not a sophisticated statistical test. Ordinal rankings are rarely as powerful as proportional continuous values, but they are still very useful.

Taking a look at the bottom of the list we find teams that have generally enjoyed the luxury of top draft picks over the last several years, but, with rare exceptions, have not been able to convert those picks into wins. At the very least, we can say they have underperformed their expectations, such as in the case of Washington. Perhaps Houston can be excused due to its expansion status.

At the top of the list are teams such as Tennessee, New England, St. Louis, and Indianapolis. These are teams that we can generally agree have achieved consistent regular season success over the past 8 years, even considering recent rebuilding efforts in Tennessee and St. Louis. But one corollary of the M-T concept is that "rebuilding" should not be required if a team's salary cap is managed well.

Conventional Pick Value Rankings

In comparison to the Massey-Thaler values discussed in the previous post, NFL teams generally use a much simpler method to value draft picks. Every pick in the draft is awarded a point value relative to the 1st overall pick, which is given an arbitrary value of 3000 points. I'll refer to these as "conventional" draft values. Some teams' draft charts might vary slightly but not significantly, as shown by the M-T study. They plotted every pick for pick trade and they universally conformed to the chart cited 2 posts ago. While the chart may not reflect the true relative worth of each pick, it does reflect how the NFL teams value them.

Using the same database of draft picks from 1999 through 2006, we can tally the total "conventional" points associated with each team. Teams with high point totals are teams which had higher value picks, more picks, or both than most other teams. Here is the list:


The usual suspects top the chart. Arizona, Cincinnati, and Detroit have had the highest convential-value picks over the past 8 drafts. Cleveland could forgiven because it was an expansion team beginning in '99. Houston might have the highest per year value, but it didn't come into existence until '02.

At the bottom of the list I'd expect New England, Tennessee, Philadelphia, or St. Louis--teams that have been somewhat dominant over the past 8 yrs. But instead we find Tampa Bay and Miami--solid teams, but not quite as dominant. Just as in the M-T value rankings, we see that trades can significantly alter the values associated with each team. In the next post, we'll compare each team's M-T values and conventional values to see if any of them might be deliberately maneuvering to maximize M-T value.

Massey-Thaler Draft Values

As discussed in the two previous posts, the Massey-Thaler study quantified the surplus value of draft picks by estimating the excess performance of a pick above his contract cost. The excess performance is measured in dollars by calculating the price of equivalent performance by a free agent.

One of the first things I wanted to investigate was how each team compared in terms of "Massey-Thaler" values (M-T values) over the past several years. Does a team (or teams) stand out as one that capitalizes on this phenomenon? Does it correlate with performance on the field?

Using data from the published study and a database of draft picks from 1999-2006, I assigned each pick its M-T value. I then summed each team's total M-T values and ranked them. Here is the list:


What stands out at first is which teams are at the top of the list and which are at the bottom. Tennessee is first, which makes sense because it's had a particularly strong regular season record during the 99-06 period which puts them near the bottom of the 1st round where M-T values are highest. St. Louis, which is third, has also had strong records. San Francisco, however, has not, and neither have several of the other teams at the top of the list. This indicates there is something at work beyond where teams' first round picks land based on regular season records.

The Redskins place last, consistent with their reputation of poor draft and salary cap management. In fact, they are significantly worse than the next worse team, Miami. The difference between the top team, Tennessee, and the bottom, Washington, is about $12 million in "surplus value" over the past 8 years.

How significant is this spread? If teams never traded picks, the team with the top pick in the draft would 'naturally' have a $3.7 million surplus value and the team with the 32nd pick each round would have a $3.4 million surplus value. Over an 8-year period, a team that had the highest 'natural' surplus every single year would accumulate only a $3.2 million advantage over the team with the lowest 'natural' surplus. A $12 million spread means something is at work well beyond simply where teams find themselves in the draft order each year.

Are some teams actively trading picks to maximize M-T surplus value? Without ever reading the M-T study or even conciously trying, teams might be forced into doing so because of various salary cap constraints. In following posts, I'll examine this further.

Draft Pick Values

Below is a conventional draft pick value chart for the first round. Not all teams use the identical chart, but this is an example of a very typical one. In fact, the Massey-Thaler study plots actual draft day trades (pick for pick, no actual players involved in the trade), and the trades match the curve of this chart nearly exactly.

# PTS
----------
1 3,000
2 2,600
3 2,200
4 1,800
5 1,700
6 1,600
7 1,500
8 1,400
9 1,350
10 1,300
11 1,250
12 1,200
13 1,150
14 1,100
15 1,050
16 1,000
17 950
18 900
19 875
20 850
21 800
22 780
23 760
24 740
25 720
26 700
27 680
28 660
29 640
30 620
31 600
32 590

According to the chart, the #1 pick is worth three times the #16 pick, or the #15, #16, and #17 picks combined. At least that's how the NFL teams value the picks. As an approximate rule of thumb, the following year's value is worth one round less than an equivalent pick this year. In other words, a second round pick this year is worth a first round pick next year.

This chart gives us some insight into the thinking of GMs. Whether or not the chart is accurate, i.e. a #1 is actually worth 3 times a #16, is almost irrelevent. Teams believe it and act accordingly.

By tallying the conventional draft points by position, we can see how the teams value the importance and scarcity of skilled players at those positions. Although conventional wisdom is that good RBs can be found anywhere in the draft, are teams' actual picks consistent with this view? Just how prized are QBs relative to other positions? We can't say how truly important one position is relative to the others, but we can determine how the GMs value them.

Draft Analysis

With the recent completion of the NFL Draft, I thought it would be interesting to look into the stats available for the draft and see what kinds of trends or surprises can be found. A recent study on the draft by Massey and Thaler was very interesting. They found evidence that the salary cap has changed the value of draft picks.

The study covered several points. First, they found that top draft choices are overvalued based on the value teams put on draft picks according the the trade chart widely used in the NFL. Player performance falls off much slower than the value teams place on the picks. They also considered the steep contract amounts top picks receive and found that teams can maximize their draft values by trading top picks for multiple picks at the bottom of the 1st round and throughout the 2nd round where the cost and risk of the players are much lower.

I intend to address their findings, but first I want to take a step back and look at some simpler things we can learn from the draft.

I created a database of draft picks from 1999-2006. According to the conventional NFL draft value chart, each pick is assigned a relative value from 1 down to "Mr. Irrelevant." In addition, the database includes the surplus value calculated by the Massey-Thaler study defined by the average marginal performance of each pick above their contract price measured in dollars. Performance value is estimated by starts, pro-bowl appearances, games played, roster appearance, and lastly comparison to each player's salary after 5 yrs, when the player would by then have earned free-agent status. The database can be downloaded here.

Menu backgrounds