fmsave

Read your Football Manager 26 saves from disk into Python records, DataFrames, CSV and JSON.

Documentation: guides and the full reference.

Install

fmsave needs Python 3.12 or newer.

uvx fmsave info path/to/career.fm   # try it without installing
pip install fmsave
pip install "fmsave[pandas]"        # adds Table.to_pandas()

Quickstart

import fmsave

with fmsave.open("career.fm") as career_save:
    my_club = career_save.managed_clubs()[0]  # empty when you are between jobs
    squad = career_save.players().where(club_uid=my_club.club_uid)

    for player in squad.sorted_by(lambda player: player.ability.current, reverse=True)[:5]:
        print(player.name, player.age, player.ability.current, player.attributes.finishing)
        # "Alex Example" 24 148 16

    squad_frame = squad.to_pandas()  # needs fmsave[pandas]
    squad.write_csv("squad.csv")

A player carries his names, birth date and age, nationality, club, height, positions, all 24 attributes, the eight personality attributes, current and potential ability, reputation, transfer value, condition, traits, his contract and any unserved ban. Records and tables are immutable and keep working after the save is closed.

Each reader returns a Table: where(...), filter(...), sorted_by(...), find(name=...), by_uid(...), by_id(...) and coverage, plus to_dicts(), to_columns(), to_pandas(), to_polars(), write_csv(...), write_json(...) and write_jsonl(...).

What else it reads

Twenty-six readers in all, each returning a Table of records:

• People: players(), contracts(), suspensions(), staff(), staff_lists()
• Clubs: clubs(), managed_clubs(), finances(), sponsorships(), facilities(), stadiums(), affiliates(), job_vacancies()
• Competitions: stages(), competitions(), fixtures(), league_tables(), competition_rules(), transfer_windows(), player_match_stats()
• Injuries: injury_types(), injuries()
• Your own club's work, which only the club you manage stores: training(), mentoring(), tactics(), set_pieces()

Every field is either verified against the game or unconfirmed; ask with fmsave.field_status(fmsave.Player, "contract.wage"). Values come from the save as stored, a value fmsave cannot read is None rather than a guess, and money is kept in the unit the game stores it in, which is not the currency it displays.

Each reader also measures what it decoded against loose bounds drawn from a couple of careers. A save unlike them can miss one and still be read perfectly well, so a missed bound is a warning and you get the table anyway. Pass strict=True to fmsave.open to have one raise instead.

What it cannot read

• Names the game renders from its own installed database. Competitions, leagues, nations, cities and all but a couple of hundred grounds. fmsave ships none of these and reads nothing from your game install. Competitions carry database_id, the id every outside name source is keyed on, so you can supply your own map. See Competition names.
• Links the save does not store, such as which competition a rules block belongs to, or which ground a club plays at. Worked out from the fixture calendar where possible, left empty where not.
• Meanings for numbers the game never displayed, such as staff job titles or the settings inside a tactic. These come back as raw numbers rather than as labels fmsave guessed at.
• What the save keeps only in part or not at all: scores for about three quarters of a career's matches, injuries older than about two years, and today's availability, line-ups, staff attribute values, card counts, club debt and asking prices.

Competition names

competition_names = fmsave.read_competition_names("competition-names.csv")
with fmsave.open("career.fm", competition_names=competition_names) as career_save:
    for competition in career_save.competitions():
        print(competition.database_id, competition.name)  # 12345 "Example League"

The file is UTF-8, two columns, database_id then name; a leading header row is skipped. Any mapping of database id to name works too. Without a map, name and every denormalised competition_name is None. About one competition in ten carries no database id and can never be named, and one the game created during a career carries an id no outside source holds.

Command line

fmsave info career.fm
fmsave export career.fm players --managed-club -o squad.csv
fmsave export career.fm players --nation 7 --columns name,age,club_name,contract_wage
fmsave validate career.fm --json

export writes any of the twenty-six tables as CSV, JSON or JSON Lines, and needs a scope: --club, --managed-club, --competition, --nation or --all. Nested groups flatten to contract_wage-style columns and a coded value gives a label column plus a _code column, so JSON is the lossless format. validate runs every reader and reports its checks, counts and coverage, holding no names, uids or text from the save, so it is safe to paste into an issue.

Where saves live

• macOS: ~/Library/Application Support/Sports Interactive/Football Manager 26/games/
• Windows: usually Documents\Sports Interactive\Football Manager 26\games\
• Steam Cloud and other stores may keep saves somewhere else.

If the game is running, copy the save and read the copy.

Safety

fmsave is read-only. It never modifies saves, makes no network connections, and does not read game memory. Using it to gain an advantage in shared online careers may break platform or community rules.

Support

fmsave is a hobby project, maintained on a best-effort basis. Report problems through GitHub issues, and never attach save files. See CONTRIBUTING.md for how to report a wrong value without sharing real data. To request removal of any content, open an issue with the rights request form.

Disclaimer

fmsave is an unofficial fan project. It is not affiliated with, endorsed by, or sponsored by Sports Interactive or SEGA. Football Manager, Sports Interactive and SEGA are trademarks or registered trademarks of their respective owners.

License

MIT. See LICENSE.