Where the model comes from
Matthew Benham took over Brentford in 2012. He had studied physics at Oxford, worked for bookmakers and in 2004 founded Smartodds, a company that models football matches statistically. In 2014 he also bought the Danish club FC Midtjylland, which became champion for the first time under his leadership in 2015 and turned into a laboratory for data-driven management in football.
Tony Bloom is a professional poker player and the founder of Starlizard, a betting consultancy that works along similar lines. He took over Brighton in 2009 and financed the stadium and training centre. Unlike Benham, Bloom keeps his models entirely under wraps. Brighton never speaks publicly about data.
The origin matters because it explains how both think. A betting professional knows that results are random in the short term. A match ends 1-0 even though the losing side had more and better chances. Over ten matches that can persist. Over forty it evens out. Whoever builds decisions on results decides on the basis of noise. Whoever looks at the underlying process — chance quality, turnovers, positioning — decides on the basis of signal.
That is the basic attitude of both clubs: process over result.
The five principles
Buy players by model, not by reputation
Both clubs assess players with their own models, which make performance data comparable across leagues. A player in the Ecuadorian league is evaluated as if he were playing in the Premier League — with adjustments for league strength, age, position and playing style.
The results speak for themselves. Brighton signed Moisés Caicedo in 2021 for around five million euros from Independiente del Valle and sold him in 2023 for around 130 million to Chelsea. Alexis Mac Allister came for under ten million from Argentinos Juniors and left as a world champion for around 40 million to Liverpool. Marc Cucurella, Yves Bissouma, Leandro Trossard, Kaoru Mitoma, João Pedro — all bought before the market recognised them, all sold at a profit or kept in the squad.
Brentford signed Ollie Watkins for under two million from Exeter and sold him for around 30 million to Aston Villa. Ivan Toney came from the third tier at Peterborough and became an international. Bryan Mbeumo came from the French second division at Troyes and moved to Manchester United in 2025 for over 70 million.
The pattern is always the same: players from markets the big clubs do not watch — South America, Belgium, Japan, the lower English divisions, Ligue 2 — who look better in the model than in public perception.
Sell when the price is right
Neither club has untouchables. If a big club pays a price far above the model value, the player is sold — even when it hurts on the pitch. The proceeds are invested in two or three new players the model sees as undervalued.
That is why Brighton did not collapse after selling Caicedo and Mac Allister. The replacements had already been bought before the departures were confirmed.
The coach as a role, not a saviour
Since 2019 Brighton has had three head coaches who had never worked in the Premier League before: Graham Potter, Roberto De Zerbi and Fabian Hürzeler. Hürzeler arrived in 2024 at 31 from FC St. Pauli and was the youngest coach in the league's history. All three were chosen by profile: playing idea, development of young players, compatibility with the club's squad planning.
Brentford had Thomas Frank from 2018 to 2025, a coach who had risen through the club and accepted the structure: the club plans the squad, the coach trains it. When Frank left for Tottenham in 2025, his assistant Keith Andrews took over.
Both clubs have shown that a stable structure survives a change of coach. The coach is important but replaceable. Squad planning is not.
Look for small advantages systematically
The best-known example: set pieces. Brentford and Midtjylland recognised early that around a third of all goals come from set pieces and that hardly any club trains them systematically. Midtjylland was one of the first clubs to employ a set-piece coach. Brentford worked for a time with Gianni Vio, an Italian set-piece specialist, and scored an above-average number of goals from corners, free kicks and throw-ins over the years.
The logic: where everybody invests equally, there is no advantage. Where nobody invests, the advantage is cheap.
Process KPIs instead of the league table
Neither club assesses coach and team primarily by points. Under Benham, Brentford spoke openly about keeping a "table of justice" — a ranking based on chance quality showing where the team would stand if chance played no part. A coach who is doing badly but has good process numbers is not sacked. A coach doing well with bad process numbers is not automatically celebrated.
Brentford: closing the academy
In 2016 Brentford did something no other English professional club had dared: it closed its youth academy.
The reasoning was sober. Brentford sat in London between Chelsea, Arsenal, Tottenham, Fulham, West Ham and QPR. The chance that a talented twelve-year-old would choose Brentford of all clubs was small. The academy's costs were high, the return — measured in players who reached the first team — close to zero.
Instead Brentford built a B team for players between 17 and 21. Recruitment: players released by other academies, often at 16 or 18, often from the big London clubs. Plus players from abroad who were undervalued in their leagues. The B team played no league but friendlies against professional sides, U23 teams and foreign clubs — as many and as demanding as possible.
The idea behind it comes straight from the betting business: selection in English academies is imprecise. A sixteen-year-old is released because he had a bad month at the wrong moment, because he is a late developer or because another player in his position was better at the time. The market for released players is cheap, because everyone assumes the academy decided correctly. Brentford did not believe that.
The B team produced players such as Chris Mepham (now Bournemouth) and served many others as a springboard. The most important effect was structural: the club had a channel for young players without carrying the cost of an academy.
Brighton went the other way and expanded a category-1 academy in Lancing, which produced Ben White, Evan Ferguson, Jack Hinshelwood and captain Lewis Dunk. Both models work. The difference lies in the honest assessment of your own situation.
Brighton: De Zerbi and the build-up
For coaches, Roberto De Zerbi's time at Brighton from 2022 to 2024 was the most interesting chapter. De Zerbi had Brighton build in a way that had not been seen in the Premier League.
The principle: the centre-backs held the ball at their feet, often for several seconds, and waited until the opponent came. Only then did they play — into the space the pressing opponent had opened. The ball was deliberately used as bait. Observers called it "inviting the opponent in".
What is methodologically interesting: De Zerbi trained these patterns with high repetition and in clearly structured forms. Build-up situations were staged, the opposing team played set pressing patterns, and his own team practised the response. That was less free than positional play under Guardiola and closer to a script. De Zerbi has said in interviews that his players repeat the build-up patterns until they run automatically.
Guardiola described De Zerbi as one of the most influential coaches of recent years. That is a strong statement about a coach who never finished higher than sixth at Brighton — and it shows how the club works: it hired a coach whose ideas did not fit the English pattern, because the profile was right.
The criticism
The models are not public. Everything said about the data work at both clubs is reconstruction. Brighton never talks about it. Benham used to talk and has become quieter. What exactly is measured, and how, nobody outside the clubs knows. That makes the model hard to assess — and hard to copy.
Success is also luck. Both clubs know that best themselves. Caicedo for five million is not proof of a model. It is proof of a good hit. Only the sum over ten years and many transfers is proof — and that speaks for both.
The ceiling is real. Brighton will never be champions. Brentford will never play in the Champions League. The model lifts a club from the third tier into the upper half of the Premier League. It does not lift it to the top, because the clubs at the top have long applied the same methods with ten times the money.
The squad is never finished. Whoever sells when the price is right has a new team every year. For supporters that is wearing. For the playing idea too: a system that needs to be settled does not tolerate an annual rebuild.
Brentford without Frank. Whether the structure really copes with Thomas Frank's departure in 2025 has yet to be seen. The thesis that the coach is replaceable is being tested right now.
Transferability to amateur football is limited. There is no data. There is no transfer market. What transfers is the way of thinking, not the tool.
Application: what you can take from it
This section is our interpretation.
Brighton and Brentford are not a collection of drills for coaches. They are a school of decision-making.
Assess the process, not the result
A youth coach who judges whether his training works by match results is judging noise. Instead count what you actually train: how many chances did we create after winning the ball in midfield? How often did we win it back within five seconds of losing it? That is your table of justice. A 0-2 with good numbers is a better match than a 2-1 with bad ones.
Find the cheap advantages
Brentford trained set pieces because nobody else did. Ask yourself: what does nobody in your league train? Throw-ins. Kick-offs. Behaviour after conceding. The goalkeeper's goal kick. Ten minutes a week on something nobody trains is worth more than another ten minutes of passing exercises.
The released player is undervalued
Brentford's B-team logic applies to any scouting: a player sent back by a bigger club has usually not got worse. He had a bad month, was a late developer or stood behind someone else. If he comes to you, look at him yourself instead of adopting the other club's verdict.
Separate the roles
Brighton and Brentford work because the club plans the squad and the coach trains it. At an amateur club that means: the head of youth decides the line, the coach implements it. A coach who brings his own playing idea and his own players every season makes the club dependent on him.
Repetition is allowed
De Zerbi's build-up work contradicts the dogma that everything must be open and game-realistic. Some patterns — building against a press, winning the ball after a kick-off, rest defence — may be trained like a script until they run automatically. Only then are they available under pressure.
Accept chance
That is the hardest lesson. Good training cannot prevent a defeat. Bad training cannot prevent a win. Whoever understands that does not change everything after a loss — and does not change nothing after a win.
What to take away
- Brighton and Brentford belong to betting professionals who run their clubs the way they place bets: on signal, not on noise. Process over result.
- Players are bought by model from markets nobody watches, and sold when the price exceeds the model value.
- Coaches are chosen by profile and are replaceable. Squad planning is not.
- Small advantages are sought systematically — set pieces are the best-known example.
- Brentford closed its academy and built a B team out of released players. Brighton expanded its academy. Both work, because both were honest about their situation.
- The models are not public, the success has a ceiling, and what transfers to amateur football is the way of thinking, not the tool.
Anyone who documents a season of training has after a year what Brighton and Brentford draw from data: a view of the process, independent of the weekend result. Coach OS records the sessions. The questions you ask of them are yours.
FAQ: frequently asked questions
Sources and further reading
- Christoph Biermann: Matchplan — on Midtjylland, Brentford and the data revolution, with interviews with Benham and Rasmus Ankersen.
- Rasmus Ankersen: lectures and interviews on the work at Midtjylland and Brentford, 2014–2021.
- Interviews with Paul Barber, Dan Ashworth and David Weir on the structure at Brighton.
- Michael Lewis: Moneyball — the origin of the thinking, from baseball.
- Interviews with Roberto De Zerbi on training methodology 2022–2024 in Italian and English sports media.
A note on the evidence: the data models of both clubs are not public. Transfer fees are base amounts according to media reports and vary by source. Statements about the data work are reconstruction from interviews and observation.