Are Safeties Risky Top Picks?

One of the Chiefs' most dire needs this off-season is a dynamic safety, but GM Scott Pioli is reluctant to take a safety with the 5th pick in the draft. Falcons GM Thomas Dimitroff is apparently on the same wavelength. There seems to be a growing conventional wisdom that safeties are high-risk picks at the top of the draft. As Peter King pointed out recently, the three best safeties of the decade--Ed Reed, Bob Sanders, and Troy Polamalu--have missed 78 games due to injuries in their combined 21 NFL seasons.

The thinking is that safety (ironically) is a fundamentally dangerous position. The nature of the position, launching head-first at high rates of closure toward oncoming ball carriers, may carry a systematically higher risk of injury than most other positions. Reed, Polamalu, and Sanders suggest this may be the case, but a sample size of three is small to say the least. Are Pioli and Dimitroff rightfully concerned?

To find out, I looked at the top draft picks at safety over the past few decades. I looked at two  statistics that indicate durability: total games played and seasons as their team's primary starter. I also looked at two other statistics that indicate total value as a player. Pro Bowl selection is a fair standard for top draft picks, and it at least indicates that a player was not a bust. I also looked at the "Career Approximate Value" statistic developed by Doug Drinen at PFR.

All data are from PFR's excellent draft query tool. I looked at 1st round picks from 1980 through 2003. Unfortunately, PFR doesn't distinguish defensive backs as either a safety or cornerback in its draft data. I classified each by hand according to their initial position classification during their NFL career. Some cornerbacks eventually transition to safety later in their careers. Ronnie Lott, Rod Woodson, and Antrel Rolle are prime examples. In those cases, I classified them as cornerbacks because that was their intended position at the time of their draft.

Although at first the cornerback/safety distinction was a complication in my analysis, both positions serve as good comparisons for each other. They each call for similar but not identical skills and talents, but they perform different roles on the field. The first comparison looks at 1st-round CBs vs. Ss in terms of durability and overall value.

1st Round Comparison

StatCornerbacksSafeties
# Players8836
Games Played110.3107.9
Yrs as Starter5.65.9
Approx Value39.741.7
Avg Pro Bowls1.21.5

It appears that 1st round Ss are at least as durable and valuable, if not more so, as their CB counterparts. With only 36 Ss taken in the first round during the period studies, we can't put much stock in the slight advantage they show in terms of value and longevity. Still, at the very least, we could say that these numbers do not show any statistical cause for concern when drafting a top S.

The second comparison looks at only the top 10 overall picks in each draft. Here we see a slightly different story.

Top 10 Picks Comparison
StatCornerbacksSafeties
# Players2711
Games Played119.7100.3
Yrs as Starter6.76.2
Approx Value51.842.9
Avg Pro Bowls2.41.7

Safeties taken in the top 10 overall picks play fewer games, have shorter careers and are less likely to be selected to pro-bowls. It's important to keep in mind we're only talking about 11 safeties in the sample, so these numbers should be taken with a heavy grain of salt. It's easy to see why some GMs and coaches would get the impression that taking a safety at the top of the draft can be a gamble.

The fact that safeties taken throughout the entire first round show no extra likelihood of injury or demise indicates that the differences observed between cornerbacks and safeties in the top 10 picks is due to sample error. Unless there is something systematically particular about the #11 through #32 picks that prevents safeties from being injured as often as their top 10 counterparts, I wouldn't conclude there is any particular risk drafting a top safety.

So far we've only looked at a comparison with cornerbacks, so let's expand the view and examine how safeties compare to some other positions. The next table lists the years as his team's primary starter for selected positions, both for all 1st round picks and for just top 10 overall picks. In the following tables, I used a slightly different data set. I limited the scope to the 1980 through 2000 drafts (instead of through 2003 in the tables above). The average career lengths will be longer because more recent picks won't have a limit of the present day.

Years as Primary Starter
Position1st RoundTop 10
RB4.15.5
QB5.66.1
WR5.77.1
DE5.67.4
LB6.07.3
CB5.67.5
S5.97.4

It appears that the longevity of safeties is in line with several other positions. In fact, they seem to have slightly longer careers than the sample of other positions listed.

Lastly, I looked at the average Approximate Value of the top picks in various positions.

Top 10 Picks Career AV
PositionCareer AV
RB53.3
QB54.1
WR53.0
DE52.5
LB55.5
CB58.9
S48.4

Here it appears safeties are less valuable than some of the other positions. This could be because AV isn't calibrated fairly across all positions (which I doubt), or it could be because players drafted highly as safeties just aren't that great. Actually, I suspect it's because a few of the very best safeties in the last generation or two have been converted cornerbacks. Hall of Famers such as Lott and Woodson, who were perennial Pro Bowlers and All Pros after moving to safety, gobbled up a lot of the AV available. That may explain both why AV for CBs is somewhat higher than average and why AV for Ss is lower than average.

In the end, I don't think there is much evidence that top safeties are any bigger injury risks than top players at other positions. The small sample size of recently injured safeties is misleading. In fact, of the 78 games missed by Reed, Polamalu, and Sanders cited above, 49 (62%) of the 78 were missed by Sanders alone. And Bob Sanders was actually a second round pick.

Perhaps there is an arbitrage opportunity. While all the other GMs and coaches are staying away from safeties, a smart GM can pick up a potentially great player later in the draft than he otherwise could.

Edit: I was pointed to PFR's own analysis of this exact same question by Jason Lisk. Sorry for the overlap, Jason. Hopefully, I've added a good deal to the discussion.

  • Spread The Love
  • Digg This Post
  • Tweet This Post
  • Stumble This Post
  • Submit This Post To Delicious
  • Submit This Post To Reddit
  • Submit This Post To Mixx

14 Responses to “Are Safeties Risky Top Picks?”

  1. Anonymous says:

    "They perform different rolls on the field" - think you meant "roles". Feel free to delete or moderate into the ether.

  2. Anonymous says:

    Its like pretending that Sanders, Polamulu, and Reed are not worth top 5 picks. If the drafts were based on their value today, each would be a top 5 pick. Debating their "value" is irrelevant, if Berry = Polamalu or Reed then you have to take him in the top 10. The only debate is whether or not the person is a top 10 talent. There are plenty more OL/DL that have been busts in the top 10 then Safeties, FACT

  3. Anonymous says:

    This is all speculation man, noone knows how good one person or the other will be... Thats why you have scouts to evaluate their talent level. And apparently Eric Berry has the talent of a "Ed Reed" from what tons of scouts, GMs and coaches have seen. "Its a dangerous position... what football position isnt? Kicker/punter?

  4. Anonymous says:

    nice work

  5. Anonymous says:

    Just out of curiosity, what formula is used to calculate approximate value?

  6. Anonymous says:

    See the link "Career Approximate Value." It's a different formula for various positions.

  7. Anonymous says:

    Hmm, I think I've seen this somewhere before...

    http://www.pro-football-reference.com/blog/?p=6464

  8. Brian Burke says:

    Oops. Missed that. Sorry Jason!

  9. Ian says:

    A thought about the top 10 safeties. Perhaps that in order to be classified as a 'top 10 safety' you have to demonstrate an ability to play the game harder than a non-top 10 safety.

    Thus, the style of play that makes you good enough to be considered a top 10 safety is the same style of play that leads to more injuries.

    Or, of course, it could just be sample error.

  10. Matt says:

    Great post. I agree with 95% of whats in here, but one thing is missing. There's another major factor to whether or not a player plays a lot of games and is the primary starter other than injury. A players ability also makes a difference.

    You would expect a top 10 pick would in aggregate play more games than the average round 1 pick. And this is the case for CB (as shown by your data). The opposite trend is seen for Safeties. Now I realize sampling size is an issue for the Safeties, but the gap widens considerable because to the two opposite trends.

    Is it possible that the value of a top CB, compared to the next level down CB (picked late in 1rst round) is greater than the dropoff in Safeties? Is there something about the Safety position that is tougher to scout, or just scouted improperly, or is this all just sampling error noise?

  11. brett says:

    Safety is not a highly paid position compared to positions that are normally picked in the top 5; franchise QB, LT, DE all make at least double what the top safeties make. Maybe teams are just apprehensive about paying top-5 salary to a player who would not normally make that much money at his position.

  12. Michael Schuttke says:

    I still wonder often how much of a selection bias goes into inflating the number of starts given to top ten picks period. After all, teams want their heavy investments to pan out and will give ample opportunities for their returns to come. In that respect, I am moderately surprised that there seems to be less safeties starting from the top ten as compared to the rest of the first round. HOWEVER...I also am starting to wonder about the ability to compare draft classes across the board regardless. For example, 2005 was considered, by and large, to be a draft that lacked "star power.' 2006 on the other hand, had a number of players who were considered franchise type players. In that sense, if a player like Berry is available in a weak draft, regardless of whether or not "good (to even great)" safeties are just as readily found in the latter half of the first round, one should still consider taking him high if that team has a legitimate need at safety. More and more, I think picking in the top ten isn't even about addressing the biggest need per say but more so about selecting a player at SOME position that needs reinforcement but, rather, picking a player that looks like the best on the board. I will not say "best player available" but, more so, best player available for that team and their respective scheme.

    So, lump sum, I just question if a.) top ten picks are just given more chances to succeed and b.) does it matter so much if you address the biggest need with a top ten pick or get the greatest return possible for the investment?

  13. Anonymous says:

    the Sean Taylor effect is one that cannot be ignored. Sean was a stud Safety before his premature murder, but he was also hurt at that point. Safe to say if he played 4 more seasons he would have been a pro-bowler each of those next 4, greatly helping the numbers out.

    R.I.P. Sean Taylor

  14. Brian Burke says:

    Very true. I went back and forth on removing Taylor from the sample, but decided to leave him in. That kind of tragedy has happened to players of several positions. Of course, none were as high profile as Sean Taylor.

Leave a Reply