Why Moneyball Isn't Finished: Behavioural Market Inefficiencies in Football

Core Question

If Moneyball transformed baseball by identifying market inefficiencies, what are football's equivalent inefficiencies — and why have current recruitment models struggled to find them?

The Context

Moneyball succeeded because baseball is a discrete-event sport. Individual pitches, swings, walks and strikeouts are largely separable — they can be counted, compared, and priced independently.

Football is different. It certainly contains events — passes, tackles, shots, carries — but those events emerge from continuous interaction between twenty-two players, where every action changes the environment the next action occurs in. A successful through ball depends on movement before the pass. A goal often depends on defensive occupation that never appears in the statistics. A counterattack may be prevented by positioning that receives no credit because nothing visibly happened.

Modern analytics already measures many football events extremely well. The question this piece explores is narrower: perhaps football's remaining market inefficiencies no longer lie in overlooked events, but in overlooked behaviours.

Exploring the Question Through FSL

FSL proposes a distinction between behaviours that produce events and behaviours that produce conditions.

Progressive carries, successful dribbles, key passes, shots, and goals are naturally rewarded because they're obvious, measurable, and easy to compare. Other behaviours are quieter: a midfielder delaying a counterattack by two seconds so teammates can recover their shape; a full-back narrowing inside early enough that the opposition abandons a passing lane altogether; a centre-forward occupying both centre-backs so a teammate attacks the space behind them; a holding midfielder calming possession after three chaotic transitions instead of forcing another attack. These moments generate no highlight and almost no statistical footprint, yet they may fundamentally change what becomes possible next. Rather than creating events, they create favourable conditions — in FSL's terms, they restore Settle before accelerating Drive, maintain Connect under pressure, and prevent Scramble tipping into Overheat.

If football markets naturally price visible events more efficiently than invisible conditions, then behavioural market inefficiencies should emerge. That's the central hypothesis explored here: the transfer market may systematically undervalue players whose primary contribution is preventative and stabilising, while efficiently pricing players whose value appears through highly visible attacking moments. Markets may be better at valuing players who change outcomes than players who preserve the conditions that make those outcomes possible. This piece doesn't claim behavioural inefficiencies are the only remaining inefficiencies in football recruitment — only that they may be among the more structurally persistent, precisely because they're the hardest to see.

Floor-raisers and ceiling-raisers

This is really Agency × Clarity applied to recruitment rather than to a single passage of play, and it's worth being explicit about that rather than treating it as separate vocabulary. Football culture tends to reward what might be called ceiling-raisers — goalscorers, creators, elite dribblers, progressive carriers — players operating in FSL's high-agency territory (Drive, Ignite, Slice, Command), producing moments that raise a team's maximum performance. What might be called floor-raisers occupy the other territory: high-clarity, controlled-agency players (Settle, Contain, Connect) whose job is maintaining stability, restoring rhythm, absorbing pressure, protecting structure. They don't raise the ceiling; they raise the floor beneath it. FSL's proposal is that many floor-raisers are behavioural stabilisers whose preventative contribution — precisely because it's about stopping deterioration rather than producing a moment — is harder for markets to see and price than the visible output ceiling-raisers generate.

The pattern is easier to recognise with names attached, even if any single instance is illustrative rather than proof. Sergio Busquets is widely credited with a specific, recurring contribution: receiving under pressure immediately after a turnover and restoring Settle before an opponent can convert instability into a chance — a repeated act that rarely shows up as a notable statistic but is frequently cited by teammates and coaches as central to how his sides retained control. Similarly, a holding midfielder who repeatedly prevents a team's Scramble from tipping into Overheat — intercepting the panicked, rushed pass rather than making a spectacular one — is doing exactly the kind of floor-raising work this piece describes, without generating the underlying data that transfer valuations are usually built from.

Illustration: Brentford

Brentford provide an interesting illustration worth investigating. Across multiple recruitment cycles they have repeatedly acquired relatively inexpensive players — Ollie Watkins, Ivan Toney, Neal Maupay, Bryan Mbeumo, Yoane Wissa among them — before selling them for substantially larger fees. Many explanations exist: excellent coaching, strong data, good timing, superior decision-making. FSL suggests another possibility worth testing rather than asserting: that Brentford may consistently identify players whose behavioural contribution is initially undervalued, place them in a system that converts those behaviours into visible attacking output, and sell once the wider market prices those visible outcomes. The mechanism, if it holds, would be simple — buy behavioural stability, sell behavioural visibility — but whether this explanation survives testing against the alternatives below remains genuinely open.

Illustration: Thomas Müller

Müller illustrates the other side of the same hypothesis. He's never been easy to classify — not an elite dribbler, not a dominant ball carrier, not defined by progressive passing, not comfortably described by a single position — and yet he's consistently influenced elite football for more than fifteen years. Modern analytics frequently rates him highly, so the claim here isn't that existing models fail to value him. It's narrower: by common scouting and coaching consensus, many of his most important contributions — manipulating defensive attention, creating passing lanes without touching the ball, occupying spaces others ignore, altering defensive timing through movement rather than possession — occur before any measurable event appears. If that consensus is right, FSL would describe Müller as changing behavioural conditions rather than merely producing isolated events, though that's a claim about mechanism that the reputation alone doesn't settle.

Alternative Explanations

Coaching quality may explain far more than recruitment. Clubs like Brentford may primarily develop behavioural value after signing players, through coaching, rather than identifying it beforehand through recruitment insight. This is the strongest rival explanation, because it's fully compatible with everything described above — the same transfer outcomes, the same underpriced-on-arrival pattern — without requiring that any market inefficiency in behavioural evaluation exists at all. If Brentford's edge is development rather than identification, this piece's central claim collapses into a coaching-quality story wearing recruitment language.

Behavioural market inefficiencies may simply be temporary. As tracking technology improves, analytics may eventually capture spacing manipulation, defensive attention, and tempo regulation directly, closing the gap this piece describes without needing a distinct behavioural theory to explain it.

The transfer market may already be correcting itself. If clubs increasingly compete for behavioural stabilisers as well as visible creators, whatever inefficiency exists here may be narrowing already, making this a description of a closing window rather than a stable, exploitable pattern.

Of these, the coaching-quality account is the one that most needs ruling out, since Brentford — the piece's central illustration — is compatible with it in full.

What Evidence Would This Need?

The decisive test: among players with closely matched conventional event-output profiles, those independently coded as stronger behavioural stabilisers should show measurably better subsequent team performance and greater long-term transfer value than those who aren't — even after controlling for the coaching environment they moved into. If no such difference appears once event output and coaching context are held constant, behavioural contribution isn't adding anything current metrics don't already capture, and the hypothesis fails.

Supporting evidence would need:

Behavioural coding of stabilising and destabilising actions, done before a transfer, not reconstructed afterward from the deal's eventual success.

Longitudinal tracking of players whose behavioural profile remains stable across different coaches and tactical systems, to separate the player's own contribution from whichever system they're placed in.

Comparison of recruitment models across clubs with strong transfer returns, to test whether the same behavioural-coding approach predicts success elsewhere, not just at Brentford specifically.

Direct testing of whether behavioural stabilisers improve collective team performance independently of their own individual statistical output.

Open Question

Is football's next Moneyball genuinely behavioural, or are these contributions simply waiting to become measurable within increasingly sophisticated event and tracking models? If future tracking technology eventually captures spacing manipulation, tempo regulation, defensive attention, and other behaviours now described as "behavioural," the market inefficiency described here may simply disappear. In that case, FSL's contribution would not have been discovering a permanently different layer of football, but identifying one that conventional analytics had not yet learned to measure. Whether football ultimately requires a distinct behavioural language, or simply richer measurement of phenomena that were always there, remains an open empirical question — and this piece can't settle it either way.

Potential Implications

For recruitment: evaluate players less as collections of technical outputs and more as contributors to team stability — asking not just how many moments a player creates, but which behavioural conditions improve when they're on the pitch.

For coaching: preventative behaviours may be as coachable as passing or pressing, rather than treated as instinctive traits some players simply have.

For analytics: the challenge isn't replacing event data but extending it — measuring behavioural influence before it appears as a visible outcome, if that turns out to be possible at all.