NFL Draft Quasi-Sharpe Ratio Analysis

TL;DR

I built a quasi-Sharpe ratio for NFL draft picks, a risk-adjusted measure of what each position is worth at each draft slot, accounting for rookie contract surplus, second-contract value, and supply-controlled free agency replacement costs. The model suggests that QB produces the strongest risk-adjusted returns in the top 10, while OT and EDGE peak in the late first round and WR offers the best value in Round 2. RB shows negative returns across all tiers in this framework, driven by cheap FA alternatives and a low contract ceiling. A large share of the total return from any draft pick comes from rookie contract surplus (getting starter production at a fraction of free agency cost for 4 years) rather than the second contract itself. These findings come with important caveats (see Assumptions at the end): snap share is an imperfect proxy for production, the model doesn’t capture positional scarcity or team-specific needs, and some sample sizes are small.

What About Love at 3?

Mock drafts have Jeremiah Love going as high as #3 overall. Computing the Sharpe ratio for RBs drafted at picks 2-5 (n=4), we get -0.97. Zooming out to all top-10 RBs, the Sharpe is -0.62, the worst of any offensive position. Here’s why:

• Hit rate is fine (71.4%); top-10 RBs usually become starters.
• But the rookie surplus is tiny. FA RBs cost just 6.0% of the cap, the cheapest position to replace in free agency.
• And the second-contract ceiling is low. The best RB deal is ~8% of cap. The best WR deal is ~14%. The upside is structurally capped.
• The opportunity cost is real. QB leads all positions in top-10 Sharpe (0.37), and the best risk-adjusted slot in the draft is EDGE in the late first (Sharpe: 0.59). Even a Round 2 WR (Sharpe: 0.23) shows stronger risk-adjusted returns than a top-10 RB.

What does this look like in practice? Here are all 7 RBs drafted in the top 10 since 2010:

Five of seven became starters. The hit rate (71%) is solid. But zero produced an elite return. The highest return belongs to Todd Gurley at 31.3%, followed by Christian McCaffrey at 28.9%, both short of the 33.1% elite threshold.

At pick #3, the positions that do produce elite returns are QB, EDGE, and OT, the positions where top-10 Sharpe is positive. If you need a running back, Round 2 or Round 3 gets you nearly equivalent production at a fraction of the draft capital cost.

This is the whole story in one chart. Green cells are where the draft beats free agency; red cells are where it doesn’t. Most of the green is in the late first and Round 2, not the top 10. The rest of this analysis explains why.

The Setup

Traditional draft pick valuation treats picks as interchangeable assets: a “top 10 pick” has one value regardless of whether you’re drafting a quarterback or a running back. But the return on investment varies enormously by position. A quasi-Sharpe ratio lets us quantify this: which positions justify premium draft capital, and which ones are better addressed later, or in free agency?

This analysis combines three frameworks:

1. Hit Rate (Riske/PFF): A player is a “hit” if their average snap percentage over their first 4 NFL seasons reaches at least 75% of the positional baseline. Snap % is averaged over ALL expected games (including seasons not played as zero), so injuries and busts are properly penalized.

2. Nonlinear Upside (Brill & Wyner 2024): Not all hits are equal. I capture the right-tail with a player return that combines rookie-deal surplus AND second-contract value.

3. Salary data from OverTheCap via nflreadr. All contract values normalized to % of salary cap to control for cap inflation across eras.

Why 2022? A fair evaluation requires at least 4 NFL seasons, enough to complete a rookie deal and either earn a second contract or wash out. Players drafted after 2022 haven’t had that runway yet, so including them would bias results toward high rookie surplus and $0 second contracts. By stopping at 2022, every player in the dataset has had a real chance to prove (or not prove) their value.

The Formula

The return on a draft pick comes from two sources:

1. Rookie Contract Surplus. You get starter-level production at a fraction of market cost.

\[\text{Rookie Surplus} = \text{Snap Ratio} \times (\text{FA Replacement Cost} - \text{Rookie Salary}) \times 4 \text{ years}\]

2. Second Contract. The market’s verdict on whether the player was worth it.

\[\text{Player Return} = \text{Second Contract Cap\%} + \text{Rookie Surplus}\]

Then the Sharpe ratio measures risk-adjusted return above the free-agency alternative:

\[\text{Sharpe} = \frac{P(\text{elite}) \times \text{Elite Threshold} - \text{FA Replacement Cost}}{SD(\text{Player Return})}\]

Where Each Piece Comes From

This formula combines two existing frameworks with new additions:

• P(elite) and Elite Threshold come from Brill & Wyner (2024). Their key insight: draft pick value is nonlinear. Earlier picks have fatter right tails. Using P(elite) instead of the average return captures this. A simple average would miss the boom outcomes that make top picks valuable.

• FA Replacement and the hit classification build on Timo Riske’s methodology at PFF. His snap share baselines define what a “starter” looks like at each position, which determines both the hit threshold and the snap ratio used in rookie surplus.

• Rookie contract surplus and supply-controlled FA replacement are new. Neither source quantified the cap savings from having a starter on a rookie deal, or calibrated FA cost to actual market supply by position.

Why rookie surplus matters: For most players with positive returns, the rookie contract accounts for 77-95% of the total return (interquartile range). When you draft a player who starts, you’re getting production at a fraction of what it costs in free agency, and that gap, multiplied by 4 years, adds up.

What Does a “Hit” Mean?

How often does a draft pick become a starter? I classify a player as a “hit” if they play at least 75% as much as an established starter at their position over their first 4 NFL seasons. The threshold varies by position because not every position plays every snap:

Snap % is averaged over all expected games across 4 seasons. Missed games due to injury count as zeros, so availability matters. A player who tears their ACL and misses a full season still clears the threshold at every position if they’re a starter when healthy (75% of snaps played × 3/4 seasons ≈ 56%).

Breaking this down by position and tier reveals where the hits are concentrated:

Supply-Controlled Free Agency Replacement Cost

The “risk-free rate” in the Sharpe ratio. Instead of drafting, a team could sign a free agent. But not every position has the same FA market depth.

Only 3 starter-caliber OTs and 3 QBs hit free agency per year, while 8 IOL are available.

I define “starter-caliber” by connecting FA contracts back to the Riske hit methodology: first, identify drafted players who were hits (snap share ≥ 75% positional baseline over 4 seasons). Then look at what those hits earned on their second contracts and take the 25th percentile as a floor. If a free agent signs for at least that much, the market is telling us they’re starter-caliber. I count how many such contracts are signed per position per year (the supply, N), and the FA replacement cost is the median contract among those top-N signings each year.

This matters because positions with thin FA markets (OT, QB) have inflated top-contract prices that don’t represent realistic alternatives.

For context, the chart above uses all available contract years to get stable estimates. But the FA market has shifted recently. Here’s what the same numbers look like using only 2020+ contracts:

Sharpe Ratio Curves

How does the Sharpe ratio decay as you move to later tiers? QB separates immediately. OT and EDGE peak in the late first round, not the top 10.

Is a WR3 Worth More Than an Elite RB1?

Rashid Shaheed, a WR3, just signed for $17M per year. Only 2 running backs in the entire league earn more (Saquon Barkley and Christian McCaffrey). The WR market is roughly double the RB market at the top end.

This shows up clearly in the data. The best top-10 RB return in the dataset (Todd Gurley: 0.313) barely matches the median Round 2 WR (0.236). Names like DK Metcalf (pick 64), Davante Adams (pick 53), and A.J. Brown (pick 51) all returned more than any top-10 RB since 2009.

RB Top 10 Sharpe: -0.62 vs WR Round 2 Sharpe: 0.23

Why? Two reasons:

1. Tiny rookie surplus. FA RBs cost just 6.0% of the cap, the lowest of any position. Even when a top-10 RB starts every game, the savings over a free agent are minimal.

2. Low second-contract ceiling. The best RB second contract in the dataset is ~8% of cap, while top WRs command ~14%. The upside is structurally capped.

3. The position structurally splits snaps. Even Bijan Robinson, a 1st team All-Pro, plays only ~74% of Atlanta’s offensive snaps. Tyler Allgeier, a 5th-round pick, handles another ~30%. An elite left tackle plays 97% of snaps. The RB position requires a committee, which caps the snap-based surplus any single back can generate.

In this framework, a Round 2 WR looks like a substantially better risk-adjusted investment than a top-10 RB, though this model measures contract value, not on-field impact, a distinction that matters most for running backs.

Are Teams Drafting Optimally?

6 positions show a positive Sharpe somewhere in Round 1: CB, DL, EDGE, OT, QB, WR. The remaining 5 (IOL, LB, RB, S, TE) show negative Sharpe in every first-round slot, suggesting the model doesn’t see enough surplus over free agency to justify the draft capital.

By this measure, 112 of 416 first-round picks (27%) went to positions where the model sees negative returns, though team needs and prospect quality can justify picks the model wouldn’t make.

Looking specifically at the top 10: for 98 of 130 picks (75%), a later tier showed a higher Sharpe ratio for that position. This doesn’t mean those picks were wrong; it suggests that on average, the same position has produced better risk-adjusted returns later in the draft. Individual prospect quality matters enormously and isn’t captured here.

Team-Level Draft Efficiency

Which teams have drafted most optimally? I score each team’s Round 1 picks by comparing what they actually drafted to what the Sharpe data says they should have, focusing on the last 5 eligible draft classes (2018-2022), recent enough to reflect current front office philosophy but far enough back that every player has had time for a second contract.

Note: The Rams are absent from these charts because they made zero first-round picks between 2018 and 2022, having traded them away in deals for Jalen Ramsey and Matthew Stafford.

The spread is enormous. PHI averages +0.24 per first-round pick while PIT sits at -0.64. Over 5 draft classes, that’s not randomness; it’s positional philosophy. Teams at the top are investing Round 1 capital in QB, EDGE, and OT, the positions where the Sharpe ratio is actually positive. Teams at the bottom are spending premium picks on positions where the data says free agency is a better path.

In the later rounds, the chart shifts to relative Sharpe: how well each team targets the best-available positions for that round. Since almost every position has negative absolute Sharpe after Round 1, what matters is whether you’re picking the positions that are above or below the round average. A team in blue isn’t necessarily finding elite talent; they’re making the best of the capital they have.

Best Positional Bets by Round

The team charts show who gets it right. But teams draft in every round, and they have to make the pick. The question isn’t “should you draft at all in Round 3?” It’s “given that you’re picking in Round 3, which positions give you the best relative return?”

To answer that, I measure each position’s Sharpe relative to the round average. This controls for the fact that later rounds naturally have lower absolute Sharpe, and surfaces which positions are the best and worst bets at each stage of the draft.

In Round 1, QB, EDGE, and OT are significantly above the round average; these are the positions where early draft capital pays off. Meanwhile, positions like TE and TE are the worst relative bets in the first round.

The later rounds flip the script. By Round 3, the best relative position is actually RB (+0.29 above round average). This doesn’t mean Round 3 RBs produce elite returns (the absolute Sharpe is still low), but it means if you’re going to draft an RB, Round 3 is where the risk-reward ratio is least unfavorable. Similarly, OT remains an above-average bet deep into the draft.

The takeaway for GMs: it’s not just about which positions to draft; it’s about when. The same position can be a great bet in one round and a terrible one in another. Teams systematically overdraft certain positions early and underdraft others late, relative to what the returns actually justify.

Key Takeaways

1. QB shows the strongest top-10 Sharpe (0.37), driven by a large rookie surplus and elite second contracts. A caveat: projecting QB talent from the NCAA to the NFL is notoriously unreliable, and this model doesn’t account for the difficulty of identifying the right QB prospect. The Sharpe reflects the average payoff conditional on having a top-10 pick. It says the math favors QB, not that the scouting is easy.

2. The late first round stands out. EDGE Late 1st has the highest Sharpe in the dataset (0.59), and several positions show their best returns in this range. This could reflect better team situations, more NFL-ready players, or simply cheaper draft capital; the model can’t distinguish between these explanations.

3. WR value appears concentrated in Round 2. The best WR Sharpe (0.47 in Round 2) suggests that comparable WR production is available later in the draft. Players like DK Metcalf (pick 64) and Davante Adams (pick 53) illustrate this, though Round 2 WRs also bust at higher rates.

4. RB shows negative Sharpe across all tiers in this framework. The combination of cheap FA alternatives and a low second-contract ceiling drives this. However, this model measures contract value, not on-field impact, a difference that matters most for RBs, whose market value systematically understates their contribution.

5. A large share of top-10 picks go to positions where a later tier shows a higher Sharpe. This doesn’t necessarily mean teams are wrong. Positional scarcity, team needs, and prospect-specific talent all matter and aren’t captured here.

6. Rookie surplus makes up the majority of a draft pick’s total return. For most positions, getting starter-level production at rookie-scale wages for 4 years is worth more than the second contract itself.

Player Surplus Leaderboard

Every drafted player scored by total surplus value generated over their rookie deal and second contract:

Player Return = Rookie Surplus + Second Contract (% of cap)

Rookie surplus measures how much value a player generates above their cheap rookie salary. It starts with snap ratio, the player’s average snap percentage divided by the positional hit threshold (e.g., 75% of snaps for a WR). A snap ratio above 1.0 means the player is a full-time starter. From there: Rookie Surplus = snap_ratio x (FA replacement cost - rookie salary) x 4 years, all expressed as a percentage of the salary cap.

Higher values mean the player delivered more total value relative to what a free-agent replacement would have cost. QBs dominate the overall leaderboard because their second contracts dwarf every other position, which is exactly what makes them the highest-Sharpe pick in the top 10. Below is the top 10 overall, then a separate top 10 excluding QBs.

Top 10 Draft Picks by Total Surplus Value

Top 10 Non-QB Draft Picks by Total Surplus Value

Assumptions & Potential Improvements

Every model makes simplifying assumptions. Here’s what this analysis assumes and how each could be refined in future work.

How each could be improved:

• Snap % → PFF grades or EPA. Production-weighted returns would be more precise than a pure volume metric.
• Second contract → true FA signings only. Separating extensions from open-market deals would better reflect actual market value.
• FA replacement → team-change contracts only. Filtering out extensions would give a truer picture of what’s actually available to a team shopping free agency, especially for positions where stars rarely leave (OT, QB).
• 4-year window → development curve. An extended window or position-specific development curves could capture late bloomers who break out in year 5+.
• $0 for no second contract → partial credit. Players who contributed during their rookie deal but aged out (especially RBs) could receive credit proportional to their rookie-deal production.
• Global elite threshold → position-specific. Position-specific thresholds would enable fairer within-position analysis, since an “elite” QB return is structurally different from an “elite” S return.
• FA supply → actual team-change frequency. Tracking how often starter-caliber players actually change teams per position per year would tighten the replacement cost estimate.
• Cap% → cap-growth-adjusted values. Adjusting for structural cap jumps (e.g., new TV deals) would improve cross-era comparisons.
• Uniform talent → draft-class strength controls. Year-over-year positional talent adjustments would reduce noise from unusually deep or thin classes.
• QB projection → bust-rate discount. Explicitly modeling the NCAA-to-NFL translation failure rate for QBs could produce a more honest expected value, even though the math still favors taking the shot.

Appendix: The Case for Trading Back

The Sharpe data tells a consistent story: for most positions, the best risk-adjusted returns come from Round 2 or later, not the top of Round 1. Teams that trade back accumulate more picks and more shots at positive-Sharpe slots.

7 of 11 positions show a higher Sharpe in Round 2 than Round 1 (CB, LB, OT, RB, S, TE, WR). Only DL, EDGE, IOL, QB show a meaningfully higher Sharpe in Round 1.

The implication is straightforward: a team that trades back from Round 1 into multiple Round 2 picks gets more shots at positive-Sharpe slots. A single Round 1 pick costs roughly 2-3x as much draft capital as a Round 2 pick (per the traditional trade value charts), but the Sharpe ratio doesn’t reward you 2-3x as much for being in Round 1.

This doesn’t mean Round 1 picks are bad. QB in particular still benefits from the earliest possible selection. But for the majority of positions, the data supports the strategy that teams like Philadelphia, Baltimore, and San Francisco have employed: trade back, accumulate picks, and take more swings in Rounds 2-3 where the Sharpe is often just as strong.

Appendix: Does Sharpe Predict What Teams Actually Draft?

If the Sharpe ratio captures real signal, we’d expect teams to draft high-Sharpe positions more often. Here’s how actual draft behavior lines up with the model.

Appendix: Round 1 Surplus Leaderboard

Top 25 Round 1 Picks by Total Surplus

Appendix: Top 5 by Position Group

Top 5 CB

Top 5 DL

Top 5 EDGE

Top 5 IOL

Top 5 LB

Top 5 OT

Top 5 QB

Top 5 RB

Top 5 S

Top 5 TE

Top 5 WR

Appendix: Top 5 by Position Group (Round 1 Only)

Top 5 CB Round 1

Top 5 DL Round 1

Top 5 EDGE Round 1

Top 5 IOL Round 1

Top 5 LB Round 1

Top 5 OT Round 1

Top 5 QB Round 1

Top 5 RB Round 1

Top 5 S Round 1

Top 5 TE Round 1

Top 5 WR Round 1

Appendix: Full Sharpe Ratio Table

Appendix: Team Position Strategy Since 2009

Data: nflreadr (PFR draft picks, snap counts, OverTheCap contracts). Methodology: Riske/PFF hit rates, Brill & Wyner 2024 nonlinear upside framework. Analysis covers drafted players 2009-2022 with snap count data available (2013+).