# Digging deeper into draft probabilities

In last week’s article titled “The Real Seven Rounds of the NFL Draft”, DRAFTMETRICS presented the case for viewing the draft as having seven segments (called Value Groups) based on the similar probabilities of success for all selections within each Value Group. In this article DRAFTMETRICS takes it a step further, first reaffirming the conclusions in last week’s article for more restrictive time periods and then presenting data by playing position.

In this article all data except five-yearinformation is based on the results of the 1993-2008 drafts. Five year information is based on the results of the 1993-2006 drafts. These ranges were selected in order to allow the opportunity for each playerin the selected time period to achieve the milestones. For example, a player selected in the 2006 draft has had a reasonable opportunity to play seven NFL seasons and establish himself as a five-year starter.

Let’s clarify a few protocols before proceeding with further discussion:
• A player earns a “starter year” for each year in which he started eight or more games
-A five-yearstarter has started at least eight games in each of at least five seasons
-A rookie starter started at least eight games in his rookie season
• A player is credited with a season played only if he appears in at least one game
• A player receives credit for a Pro Bowl appearance only if he was an original selection for the game, whether he played or not
• A player receives credit for an All Pro selection if he was named to either the Associated Press or Pro Football Writers of America All-Pro teams.

Following is a table that reflects historical results by Value Group. This is essentially the same information shown in “The Real Seven Rounds of the NFL Draft” but for the time periods described above and with a few moremilestones. While good players come from all parts of the draft,the table shows that, among other things, a player drafted in the first 13 picks is 10 times more likely to become a five-year starter than a player selected in the last 60 picks (74.7% versus 6.8%)

Again, this table does not show that you can only get good players early in the draft. Rather, it makes the somewhat obvious point that the earlier a player is drafted the more likely he is to be successful and addresses the question of “how much more successful”.

As one might expect, results vary by playing position within each Value Group. Tables showing the probability by playing position are included at the end of this article. If you like numbers, you will love those tables.

For those who are more interested in a brief overview,the following table focuses on only one milestone (five-year starters), which could possibly be the most important measure, and shows the probability of drafting a five-yearstarter by playing position within each Value Group. Data is shown only for those positions with 10 or more draftees in that Value Group.

It is interesting to note the significant fluctuations across playing positions afterthe first 13 picks of the draft. Is this due to randomness or a real difference in risk? Here are the main conclusionsreached by DRAFTMETRICS:
• Offensive linemen tend to be the least risky positions to draft
• Offensive tackles stood out as by far the safest bet in the first 13 picks of the draft
– 20 of the 21 offensive tackles started as rookies and the only exception started in his second season
– 94% became five-year starters, 81% were selected for at least one Pro Bowl and 62% were honored as All Pro at least once
– All OT who were three-time Pro Bowl selections were selected in the first 13 picks• RB andWR tend to be the riskiest positionsto draft
DRAFTMETRICS has no evidence to support this, but for RBs this is probably largely a function of the injury risk at the position
– Significant drop-off in the probability of three-year versus five-year starters

– Over one-third of the RBs who became five-year starters were selected in the first 13 selections
– There were 70 WRs who became five-year starters and 57 of those were selected in the first 86 selections in the draft
– Later round WRs are lottery tickets

– Rookie WRs are less likely to start than any other playing position
– No WRs selected between picks 150 and 189 started as rookies and less than 1% started from selection 190 through the end of the draft
• It’s no surprise that QBs tend to get pushed up earlier in the first round
– As many QBs were taken with the 1-13 picks as are taken with selections 14-66
– QBs had the lowest probability of starting as rookies with selections 1-13 and were second only to WRs over the entire draft
– Success rates are low with the 14-66 selections and none of the 12 QBs taken in the 67-86 selection range became three-year starters
• Late round LBs (picks 190 and after) rank among the lowest probabilities of success
• Corners drafted with selections in the 67-86 range and then 150 and after tend to be risky
• Offensive linemen and safeties have the highest probability of starting for picks 190 and after
• DEs selected between the 67th and 149th picks tend to be more successful than other positions
– Pretty close to average in the first 40 picks and 150th pick and later
• On an overall basis about three-quarters of three-year starters go on to become five-year starters with RBs the lowest at 58% and QBs the highest at 90%