Academy Rankings 2025/26: How the research was done

Defining the player population

Every edition begins by establishing the population of professional players whose Academy histories will be researched. Using data from FBref, we identified every male player listed as English or Welsh who made at least one appearance in one of the 23 covered senior competitions during 2025/26.

That produced a population of 2,067 players: 1,897 English and 170 Welsh. One senior appearance is enough for a player to enter the study, but it does not give every player the same value; the strength of the competition and the minutes he played are taken into account later in the scoring.

The increase from 1,814 players last year came from two expansions. Welsh players were included for the first time, allowing Cardiff City, Newport County, Swansea City and Wrexham to be assessed through the English and Welsh players they developed. And competition coverage grew from nine leagues to 23.

The competitions covered this year are the Premier League, Championship, League One, League Two and National League in England; the Bundesliga, La Liga, Serie A, Ligue 1, Belgian Pro League, Primeira Liga (Portugal), Danish Superliga, Süper Lig (Turkey), Eredivisie (Netherlands), Swiss Super League, Segunda División (Spain), Super League Greece, Austrian Bundesliga, Scottish Premiership and Ligue 2 (France) in Europe; plus Major League Soccer (USA), Liga MX (Mexico) and A-League Men (Australia).

This captures many more players who came through English or Welsh Academies before continuing their careers overseas, or outside the competitions covered previously.

FBref provides the fixed starting population, together with each player’s senior competition, appearances and minutes. What it does not provide is the complete Academy history behind each player. Establishing that requires a separate research process – and it is by far the hardest part of the study.

Reconstructing Academy histories

Identifying the eligible players is relatively straightforward. Establishing where each of them developed is considerably harder. There is no single complete and verified database of Academy histories, particularly for less prominent players further down the pyramid.

Transfermarkt is a valuable starting point, but its community-maintained records can be incomplete and do not always give the context the Rankings need. An entry may lack dates, or refer to a grassroots club, a trial, a training period or some other involvement that does not amount to time within a professional Academy. For that reason Transfermarkt was used as a source of research leads rather than as definitive evidence on its own.

In previous editions, investigating those gaps meant extensive manual searching. Repeating that across 2,067 players would have taken hundreds of hours. For this edition we developed an AI-assisted process, using OpenAI’s Codex models, to search, compare and reason across public sources – in effect, a research assistant that can read and cross-check sources itself rather than simply extract data from a fixed field. This allowed the same detailed investigation to be run consistently for every player in scope.

Starting from a known FBref player and any available Transfermarkt leads, the process searched the wider public web for evidence from sources including official club biographies, Academy and scholarship announcements, governing-body records, first-person interviews and reputable news reports. It could then compare those sources, check that they referred to the correct player and identify possible Academy spells that were missing from the initial leads.

This is where this kind of AI offers a real advantage over conventional automated data collection. The relevant information does not appear in a consistent field or use standard language. A source might say a player “came through the ranks”, “joined the youth set-up”, signed a scholarship or appeared for an Academy age group. Each statement has to be interpreted in the context of the player’s identity, age, career chronology and the type of involvement being described. The process therefore helped not only to find sources, but to classify Academy involvement and extract the dates or duration the evidence supported.

The AI helped find, compare and interpret evidence; the evidence itself came from public sources. Every Academy association retained for the Rankings was linked to supporting evidence and uncertainty and conflicting information were preserved rather than replaced with unsupported assumptions. Difficult or ambiguous cases could trigger an additional research pass and a varied sample of player records and their supporting sources was also checked manually.

For an Academy to receive credit, the evidence had to show that the player was part of a professional club’s formal Academy or youth system before his 18th birthday. Trials, training periods, grassroots clubs and post-18 development squads did not qualify on their own. Dates and durations were recorded only to the precision the sources supported.

The result was one source-linked research record for each player, containing his qualifying Academy history and the available evidence about the timing of each spell. Only once this research was complete were the player records passed into the separate ranking calculation.

Calculating player value

With the Academy histories established, the next stage was to assign a value to each player. The central principle is that the standard of senior football reached remains the primary measure, with playing time providing additional information about the player’s level of involvement.

Each player receives one competition weighting, based on the strongest covered competition in which he appeared during 2025/26. If he appeared in more than one covered league, only the highest weighting is used, so his value is never duplicated.

Competition weights were rebuilt from Opta’s league-strength ratings. We chose a range of 1.0 to 3.0 to balance two aims: rewarding Academies for producing professional players at all, and recognising the higher level some of those players reach. The National League was set at 1.0 and the Premier League at 3.0; every other competition was placed proportionally between them according to its average Opta rating, then rounded to one decimal place. We also tested reasonable alternative ranges and confirmed that the broad results did not depend on this exact conversion.

Minutes played were introduced as a secondary factor for the first time. A minimal appearance receives approximately 90% of the relevant competition weighting, rising gradually to the full value at 2,700 minutes – the equivalent of 30 complete matches. Minutes beyond that point do not increase a player’s value further.

This deliberately limits the influence of playing time to a 10% range. A player making a single brief Premier League appearance receives approximately 2.7 points, while one reaching the minutes threshold receives the full 3.0. A Championship player sits between approximately 2.07 and 2.3. Regular involvement is rewarded, but minutes can never outweigh a substantial difference in competition strength.

The result is one value for every player, reflecting primarily the level of football he reached and, to a lesser extent, how much he played. The next step is to decide how that value is shared when more than one Academy contributed to his development.

Sharing player value between Academies

If a player developed at only one qualifying Academy, that Academy receives his full value. Where more than one Academy contributed, the value is divided between them rather than duplicated.

For this edition, the allocation reflects the relative length of each Academy spell where sufficiently reliable and comparable evidence of duration is available. Spells are grouped into three broad bands:

Each Academy’s share is calculated from its tenure weighting as a proportion of the player’s combined total. The broad bands recognise meaningful differences in the length of development without pretending that contribution can be measured precisely month by month.

Duration-based allocation is used only when comparable evidence is available for every Academy credited with developing that player. If the duration of any spell remains unclear or disputed, the player’s value is divided equally between all of his qualifying Academies. This prevents the Academy with the most detailed surviving public record from gaining credit simply because another spell is less well documented.

Every qualifying professional Academy remains part of the calculation, including foreign clubs and other Academies outside the published table. A ranked Academy cannot inherit the share belonging to another club simply because that club falls outside the scope of the Rankings.

Finally, every player-value share credited to an Academy is added together to produce its final score and position. Those same contributions are grouped by the senior competition in which the players appeared to create the breakdown shown in the full table. The competition figures are parts of an Academy’s total, not additional points.

Interpreting the Rankings

An Academy’s final score reflects the number of English and Welsh professional players it developed, the strength of the senior competition they reached, their level of involvement, and the share of their development attributed to it. It is a measure of demonstrated Academy productivity within the senior football landscape covered in 2025/26.

The Rankings do not assess the current quality of an Academy’s youth teams or the potential of the players presently within its system. A low score reflects the number of graduates in senior football this season, not the standard of the Academy’s current work. Nor can any study built on public information guarantee that every historical detail has survived. But retaining source-linked player records, and applying the same evidence and scoring rules across the full population, makes each stage of the process consistent and traceable.

The complete results are in Academy Rankings 2025/26: The complete table, available to TGG Members.