What Actually Matters in NWSL Defense

Tackles Don’t Matter

We correlated every measurable team defensive metric against expected goals against per game (xGA/G). The results are not subtle.

Shots against per game dominates everything(r = +0.703). Pressing volume is the second-strongest signal (r = −0.629). After that, the numbers fall off a cliff. Block rate, interception rate, tackle win percentage: all noise.

The Chicago Red Stars had a 60.8% tackle win rate, third-best in the league. They also had the worst defense in the league at 1.74 xGA/G. Winning tackles doesn't prevent goals if you're constantly in situations where you need to make tackles. Clearances tell a similar story: they correlatepositivelywith xGA (r = +0.347). Teams that clear a lot are teams under pressure, not teams that are defending well.

Pressing Is the Mechanism

If shot volume is what matters, the next question is straightforward: what prevents shots? The answer is pressing. Teams that pressure the ball more frequently allow fewer possessions to develop into shots.

Gotham FC and Chicago sit at opposite ends of the spectrum.

Gotham allowed 9.1 shots per game. Chicago allowed 15.6. Both had essentially the same tackle win rate (~61%). The difference was upstream: Gotham's pressing disrupted possession before it could reach shooting range. Chicago's pressing was less effective, and by the time attackers were in position to shoot, winning individual duels couldn't compensate.

Across all possessions in our data, roughly 68% of non-shot possessions end because of attacker errors (misplaced passes, bad touches, losing the dribble) rather than direct defender actions. This doesn't mean defenders aren't contributing. Pressing creates those errors. The presser forces the bad pass; the interception that follows gets credited to a different player. Pressing is the invisible mechanism behind a huge share of turnovers.

Measuring Everything in Expected Goals

Traditional box-score stats count actions (tackles, interceptions, blocks) but don't tell you whether those actions mattered. A tackle in your own box is worth far more than a tackle at midfield. An interception that kills a counter-attack is worth more than one that stops a sideways pass.

To solve this, we built a possession value grid: the pitch divided into a 12×8 grid (96 zones), where each zone holds the average xG a possession is worth when it passes through that location. The gradient runs from ~0.003 xG in the defensive third to ~0.084 xG in the penalty box. A zone in the attacking third is roughly 3–4x more dangerous than a zone at midfield.

With this grid, every action on the pitch gets a value. We split player contributions into two sides.

Defensive xG Impact

Four components, summed into a single number:

• xG Blocked — expected goal value of shots physically blocked
• xG Denied — possession value at the location where the player won the ball (tackles, interceptions, recoveries)
• xG Cone Saved — marginal xG reduction from being positioned in the shooting cone during opponents' shots
• xG Turnover Cost — possession value lost via the player's own dispossessions and errors (subtracted)

Offensive xG Impact

Four parallel components:

• xG (shots) — expected goal value of the player's own shots
• xA (assists) — expected goal value of shots assisted
• xG Progression — possession value added by moving the ball forward
• xG Off Turnover Cost — possession value lost via offensive turnovers (subtracted)

The combined metric (offensive + defensive) tells you a player's total contribution in xG currency. A center back who prevents 6 xG defensively is as valuable as a forward who creates 6 xG offensively. The currency is the same.

The Defensive Rankings

When we sum up all four defensive components across the 2023 season (minimum 600 minutes), center backs dominate the top of the table. This is mechanical: CBs are positioned in the most dangerous zones, they appear in shooting cones, and they make blocks and interceptions where the possession value is highest.

Abby Erceg leads at +6.84 xG prevented. Her profile is balanced across all three categories: she blocked 2.43 xG worth of shots, denied 2.54 xG through interceptions and ball recoveries, and reduced 1.91 xG through cone positioning. Samantha Staab is close behind at +6.32, with the highest shot-blocking value in the top 10 (2.57 xG).

Eight of the top ten are center backs. The two fullbacks (Dydasco and Milliet) rank high because of xG Denied: they win the ball in dangerous areas at high rates. But fullbacks generally accumulate less cone value because they're positioned wider, outside the shooting corridor.

The Complete Picture

When we combine offensive and defensive value into a single number, the rankings tell a different story than either side alone. The most valuable players in the league are not necessarily the top scorers or the top defenders. They're the ones who contribute the most across both sides of the ball.

Samantha Staab is the most interesting name on this list. A center back ranked second overall at +9.27. Her defensive value (+6.32) is obvious, but she also generated +2.94 xG offensively through assists (3.10 xA) and ball progression (3.55 xG Progression). She moved the ball forward more effectively than most midfielders in the league while also being its second-best defender. Washington had two of the top three combined players (Hatch and Staab), which helps explain their 2023 season.

Ashley Hatch led the league at +9.62, almost entirely through offense (8.42 xG from shots and chance creation). Abby Erceg was the mirror image: +7.35 combined, with 93% of her value coming from the defensive side. These are the two archetypes: the pure attacker and the pure defender, both among the most valuable players in the league through completely different mechanisms.

Top 5 by Position

Defenders

Midfielders

Forwards

What This Doesn’t Capture

This model has real limitations, and they're worth being explicit about.

• One season of data (2023).StatsBomb's open NWSL data covers a single season. Player rankings would be more stable with multiple years, and year-over-year validation isn't possible.
• No lineup or shift data.Soccer doesn't have shifts the way hockey does, which means we can't build RAPM-style models that separate individual impact from teammate and opponent effects. A center back's value is tangled with her partner's.
• Positioning credit is limited. We can measure positioning during shots (freeze frame data), but not positioning during the other 95% of the game. Off-ball movement, organizing the back line, and communication are invisible to event data.
• Pressing credit goes to the presser. The model credits the player who makes the pressure, not the system that created the pressing trap. A well-organized press might funnel the ball carrier into a cul-de-sac where the nearest defender gets credit for a ball recovery, but the real value was the structure around her.
• Cumulative, not rate-based. These rankings reward playing time. A player who accumulates +6 xG over 2,100 minutes is valued over one who accumulates +4 xG over 1,200 minutes, even though the latter's rate is higher. For a season-level view, cumulative totals better reflect who actually contributed the most value to their team.