FM24 Companion: a free tool built on the test harness work, plus what my own test runs found

Forsti37

New member
Joined
Nov 10, 2026
Messages
7
Reaction score
0
Hi all,

I’ve spent the last weeks turning the research from this forum into a companion app for FM24, and I measured a lot
along the way with the harness. Both are free and open source (GPL-3.0):
GitHub - Forsti37/FM24-Companion: Companion app for Football Manager 2024: reads the running game and turns match-engine research into buying, development and team-selection decisions

First, thanks. None of this would exist without:

  • @LightningFlik : FM-24-Match-Harness and the development-function thread (20900). Being able to run FM24’s own
    engine and development routine with fixed seeds changed everything.
  • @blackbird : the 93,000-match attribute curves (15934 #860). Their shape is still the backbone of my model.
  • EBFM: Matchday Mechanics.
  • @harvestgreen22 : the training catalogue (测试40) and the feet/pressure/training tests. Your training data ships with the app, with credit.
  • @GeorgeFloydOverdosed : the summary thread and the GK grid.
  • kLamAHS/fm24bot and the Squad Analyzer: their public notes on the game’s memory structures.
What my harness runs found in a total of 4.5 million (4545049 as an exact number) simulated games (one long-running 24.4.2 save; paired seeds, 2k screening / 10k confirmation per arm.
Caveat: the save runs the FM TWEAK v1.4 engine mod, so match numbers may differ a bit on vanilla.)

  1. The published attribute curves are ~2× too strong in my runs. Whole XI 12 → 13 in everything: +11.3 season points
    measured vs +22.4 predicted. Pace/acc still lead, but scaled 0.67 / 0.59; jumping reach 1.41; work rate only 0.33.
  2. Position matters. A striker’s pace is worth 0.27× the team average, a CB’s jumping reach 2.1×. One player 12 → 20
    in everything: AMR +14.7, DC +9.4, DM +8.2, ST +7.9 points.
  3. Condition hits about 3× harder than earlier tests said, sharpness only about ⅓ as hard. Whole XI at 90 % condition:
    −11.7 points a season; 90 % sharpness: −4.5. Rotate.
  4. Set pieces don’t decide matches. Corners, FKs, penalties, long throws (5–20), height, and the trained set-piece
    routine familiarity (0 vs 100, 12k matches): nothing measurable.
  5. Development is match points. Under 125 a month (~4–5 full games) growth is braked; LightningFlik’s function
    ran ~2× too fast at 17–21 in this save.
  6. Every player has a hidden development ceiling (1–5) that caps CA at 70–100 % of PA. It is re-rolled on a transfer, a
    renewal and the first pro contract, and the odds depend on the new club’s facilities. A low-ceiling youngster
    should get his pro deal early; a ceiling-5 one should stay on his youth contract.
  7. Team training barely matters. Sessions fill 11 categories capped at 50; three Attacking sessions a week saturate
    them, and every schedule I tried is within ±0.2 rating points of pure rest. The “block weight is split over its
    attributes” rule from the training file doesn’t hold.
  8. The individual focus is the lever, and it fades toward 20. Quickness adds +0.80 pace a season at 8 but only +0.16 at
    19; Agility & Balance and Ball Control still add ~0.5–0.8 near 20.
  9. Level matters a little, minutes a lot. A pro-level game gives ~10–30 % more match points than an academy game, and
    friendlies count at the lowest level.
Full write-ups with tables:
FM24-Companion/docs/findings/README.md at main · Forsti37/FM24-Companion

The app
reads the running game (Steam 24.4.2) and gives you:

  • 0–100 ratings per position (weights editable)
  • depth chart for first team, B team and U18
  • best XI for today
  • growth, best training focus and a friendly planner
  • transfer upgrades now and at peak, release clauses, bargains and expiring contracts
  • renewals and sell/loan candidates
Optional and off by default, it can also manage individual focus, schedules, saved lineups, shortlists and the
transfer/loan list in the game. Keep a backup save and use it single-player only.

Download (latest release):
Release FM24 Companion 0.10.2

Feedback, bug reports and your own harness numbers are very welcome. Especially vanilla-engine runs to compare with my
modded ones.

Disclaimer:
The biggest part of this is vibecoded, but I have a completed education as a software engineer and am currently studying informatics while working as a software engineer. I've spent the last 2 days reviewing everything again as I really liked how it changed the way I play the game and I thought it would be interesting for you guys as well
 
Last edited:
It doesn't work for me. Even though it reads properly ingame date in upper right corner, it doesn't show any data whatsoever
 
Is there a way to make it show in scouting only players who are interested to join ?
 
It doesn't work for me. Even though it reads properly ingame date in upper right corner, it doesn't show any data whatsoever

What game version are you on? I've released a new version that should show a clearer error message in the UI.
 
Last edited:
Is there a way to make it show in scouting only players who are interested to join ?
I've tried to recreate the function the game uses, but I've yet to crack it. If you know any other tool that can do it I can maybe derive it from that. The problem is, that the game calculates it on demand and it isn't saved anywhere.

However you can create a shortlist with all the players that are interested in joining and filter through it in the app or let the app manage your shortlist and filter through it in the game.
 
I've tried to recreate the function the game uses, but I've yet to crack it. If you know any other tool that can do it I can maybe derive it from that. The problem is, that the game calculates it on demand and it isn't saved anywhere.

However you can create a shortlist with all the players that are interested in joining and filter through it in the app or let the app manage your shortlist and filter through it in the game.
I found this for FM 24 but there's like 40 factors that go in to generating the final willingness score; it's not even worth sharing my notes because the evaluation is worthless without recreating FM 27's 100% guaranteed. I don't own 27 either so I can't point you to the functions that govern it. Claude or ChatGPT should be able to find it for you though if you've got Ghidra or IDA Pro with an MCP server and you say something like this:

"I want to understand the mechanism that governs whether a player wants to sign with me or not. The Football Manager process is attached to Cheat Engine and IDA Pro, the MCP servers of which are both available. In my game (running), Eric Maxim Choupo-Moting (UID: 35015916) and Nicolás Tagliafico (UID: 14008030) are regarded as being "Extremely interested" in joining. Solly March (UID: 29090530) is only considered "Interested", which is two levels below this."
 
There is a recommend team schedule section in This week in FM card on development tab and Team schedule and friendly card that show change in the next 8 sessions base on what schedule are currently set, which schedule option should i follow, is it not that important? What is the different between the two?
 
There is a recommend team schedule section in This week in FM card on development tab and Team schedule and friendly card that show change in the next 8 sessions base on what schedule are currently set, which schedule option should i follow, is it not that important? What is the different between the two?
Good catch, the two disagree, that’s on me. Follow the "Team schedules and friendlies" card. The “Recommended team schedule” line under "This week in FM" is a leftover static tip from an older study. It doesn’t look at your save and nothing applies it; I’ll remove it in the next release.

If you activate the "Weekly Schedules" option in the settings, the app automatically writes the training schedules into the game. If you do the trick of leaving training schedule management to your assistant, so your players don't get mad, he'll sometimes override it on a Sunday but the app fixes it by Monday.

How much does it matter? Less than you’d think. In harness tests, three Attacking sessions a week are enough to fill FM’s training categories (about +6 CA a season vs. resting), and more adds nothing. Physical sessions and set-piece Routines showed no measurable effect. The things that really drive development are each player’s Additional Focus (also set automatically if you activate that in the app) and match points: about 125 a month for full growth. That’s why the same card tells you how many friendlies to arrange.
 
Back
Top