FM27 match engine: what we know so far

And here are some suggested genie scout values based on the pooled results.

ST
Pace 100, Acceleration 92, Jumping Reach 79, Anticipation 46, Stamina 43, Finishing 37, Balance 32, Strength 30, Concentration 23, Important Matches 22, Dribbling 20, Pressure 90 (fixed from 18 based on GeorgeFloydOverdosed comment below) , Heading 14

AM (AML / AMC / AMR)
Pace 100, Acceleration 80, Jumping Reach 50, Strength 49, Stamina 45, Anticipation 36, Dribbling 35, Finishing 33, Concentration 27, Long Shots 25, Pressure 120 (fixed from 24 based on GeorgeFloydOverdosed comment below) , Balance 19, Determination 18, Heading 16

FB / WB (DL/DR, WBL/WBR)
Pace 100, Acceleration 83, Stamina 61, Dribbling 41, Anticipation 34, Jumping Reach 33, Concentration 30, Strength 28, Important Matches 21, Long Shots 20, Balance 19

DM
Acceleration 100, Pace 64, Stamina 49, Jumping Reach 38, Strength 33, Agility 25, Anticipation 24

CB (DC)
Jumping Reach 100, Acceleration 86, Pace 62, Balance 34, Stamina 33, Anticipation 30, Concentration 20, Strength 20, Work Rate 16, Important Matches 15, Pressure 70 (fixed from 14 based on GeorgeFloydOverdosed comment below)

GK
Agility 100, Reflexes 91, Aerial Reach 30, Acceleration 21, Concentration 18, Important Matches 11, Anticipation 10
 
Last edited:
Just keep in mind that aside from the calculation bugs that plague Genie Scout, personality attributes are weighted at 20% internally. So Pressure 18 should become Pressure 90.
I've seen repeated claims about this (not from you) that were based on code from an old tool that was wrongly claimed to be Genie Scout. What I haven't seen is actual code from GS. I'm therefore skeptical about what its doing in the rating calculation. It does use personality attributes in the probability of development.
 
I've seen repeated claims about this (not from you) that were based on code from an old tool that was wrongly claimed to be Genie Scout. What I haven't seen is actual code from GS. I'm therefore skeptical about what its doing in the rating calculation. It does use personality attributes in the probability of development.
I know the thread that was on FM Arena you're probably talking about. This is not that.

I arrived at this conclusion about Genie Scout through my own analysis of it. You can deduce it by only weighting one thing at a time, and observing the effects. You'll find personality attributes are weighted and capped at 20%, while the way it calculates things just does not add up.. you can have a player be better than another in every way, and yet that player inexplicably has a lower rating. Even when 2 players are identical (you can do this by weighting only 1 attribute), one will be +0.2% say. The errors are relatively minor, but it matters when you're trying to accurately distinguish to within ~1%, with your own error on top of that - effectively the error ends up being more like 3-5%. And this is aside from the inbuilt proficiency and footedness weights, which are reasonable, but that you cannot change.
 
Last edited:
I know the thread that was on FM Arena you're probably talking about. This is not that.

I arrived at this conclusion about Genie Scout through my own analysis of it. You can deduce it by only weighting one thing at a time, and observing the effects. You'll find personality attributes are weighted and capped at 20%, while the way it calculates things just does not add up.. you can have a player be better than another in every way, and yet that player inexplicably has a lower rating. Even when 2 players are identical (you can do this by weighting only 1 attribute), one will be +0.2% say. The errors are relatively minor, but it matters when you're trying to accurately distinguish to within ~1%, with your own error on top of that - effectively the error ends up being more like 3-5%. And this is aside from the inbuilt proficiency and footedness weights, which are reasonable, but that you cannot change.
I understand your point, but there is a logic behind limiting the impact of pressure. Most of the time pressure has zero impact, and when it does it works by reducing other attributes like composure , decisions, and concentration. Two of those 3 things are literally weighted zero and the third Caio has suggested weights of 20-30. Its just my opinion, but I think putting pressure up at 80-100 would just be be saying that you think consistency 20 and pressure 1 is the same as consistency 1 and pressure 20. Logically that cant be true because pressure of 1 only reduces consistency for a small part of the game, and the consistency 1 player will be bad the whole game.
 
I understand your point, but there is a logic behind limiting the impact of pressure. Most of the time pressure has zero impact, and when it does it works by reducing other attributes like composure , decisions, and concentration. Two of those 3 things are literally weighted zero and the third Caio has suggested weights of 20-30. Its just my opinion, but I think putting pressure up at 80-100 would just be be saying that you think consistency 20 and pressure 1 is the same as consistency 1 and pressure 20. Logically that cant be true because pressure of 1 only reduces consistency for a small part of the game, and the consistency 1 player will be bad the whole game.
I don't know where you've heard those things from, I can only assume from recent reverse-engineering.

Be that so or not, I trust results more than theory, because the game can be faulty but results are self-evident. Good example of this is how recently we're learning how ornate the match engine is, but at the end of the day pace/acc dominates and almost every technical makes no difference to outcomes.

Orion's results from 2022:

Fn2o6rT.png


It remained true for FM24 based on testing myself and HarvestGreen did, so I imagine it's the same for FM27.

And in fact all I've said is that if caiograco finds that Pressure is 18 weight in FM27, then the way to represent that properly in Genie Scout is to multiply it by 5.
 
The numbers I shared are from moving attributes from 12-18. Things change if we go all the way to 1. In particular, pressure works on finishing, which in turn influences whether xG becomes goals. At pressure of 1, most xG becomes wasted chances.

Here is a similar graph from my tests.
1794886778251.webp


What i don't have is tests at a more granular detail, from 1 to 20. And the estimates above could also be updated with more tests, but the patterns is the same as in previous games.
 
I don't know where you've heard those things from, I can only assume from recent reverse-engineering.

Be that so or not, I trust results more than theory, because the game can be faulty but results are self-evident. Good example of this is how recently we're learning how ornate the match engine is, but at the end of the day pace/acc dominates and almost every technical makes no difference to outcomes.

Orion's results from 2022:

Fn2o6rT.png


It remained true for FM24 based on testing myself and HarvestGreen did, so I imagine it's the same for FM27.

And in fact all I've said is that if caiograco finds that Pressure is 18 weight in FM27, then the way to represent that properly in Genie Scout is to multiply it by 5.
"And in fact all I've said is that if caiograco finds that Pressure is 18 weight in FM27, then the way to represent that properly in Genie Scout is to multiply it by 5."

Only if you assume that pressure has some independent effect, but the tests examine attributes one by one. You can reject the idea that pressure works by altering other attributes (without evidence) but if it does then it's not logical to weight it as you suggest. I already pointed out why.
 
Here is how pressure works:
  1. Nerves before the match. Before kick-off, each player gets a "nerves" score from his Pressure (and other attributes). It is bigger in bigger matches.
    • Nerves lower every effective attribute for the whole match.
    • Measured live in the away match at Man City:
      Squad's PressureNervesEvery effective attribute
      Own values17— (baseline)
      140about −2.7
      200about +1.2
  2. Choking on each kick.Every pass, cross and shot first gets a quality roll.
    • The match then sets a nerves level. It is low in open play, higher in big matches, and higher again for penalties, set pieces and shoot-outs.
    • If the player's Pressure dice roll is below a random draw against that level, he chokes: the quality roll is redrawn and comes out worse on average.
    • At Pressure 1, the Pressure roll was always 0 (live). So the choking check fires on almost every kick.
    • This is why finishing collapses with Pressure = 1. Goals fall 0.43 a game below xG, and shots on target fall by 1.9 a game. The chances are created, then missed or badly struck.
  3. Morale on bad events. Players with Pressure under 15 can lose up to 9 morale points when something bad happens, such as conceding. Lower morale lowers effective attributes at the next refresh.
 
Morale on bad events. Players with Pressure under 15 can lose up to 9 morale points when something bad happens, such as conceding. Lower morale lowers effective attributes at the next refresh.
Is this a precise threshold?

I see that you've weighted pressure as less important for more defensive players (or ones furthest from goal), is it because pressure mainly only counts for shot-taking or shot-stopping? If this is the case, wouldn't it in theory possibly be akin to jumping reach, where you only need the 1 or 2 shot blockers and shot takers to have it for most of the benefit?

If it is under 15 pressure, would the morale penalty be deterministic (i.e. 14 gets -1 morale, 6 gets -7 morale) or random?

I am also surprised by your sportsmanship result. It seems quite significant. I thought sportsmanship has negligible effects, and I think I tested this (note: in FM24), though perhaps not adequately. Can you explain the mechanics behind sportsmanship?
 
Last edited:
Asked the agent to look into the code. Here is its summary slightly edit for clarity.

1. Is 15 a precise threshold?
Yes. At 14 or below the morale drop can happen, and at 15 or above it can’t. In practice the cutoff barely matters, because at 14 the drop almost never happens. The Pressure value used is blended with a team-mate’s, probably the captain’s, so a calm captain can help a nervous player. I can dig more into the code if you are curious.

2. Is the penalty fixed or random?
It’s random in two ways:
  • Whether it happens: the chance each time is about 1 in Pressure squared. That’s certain at 1, 1 in 4 at 2, 1 in 25 at 5, 1 in 100 at 10, and about 1 in 200 at 14.
  • How much: a random 0–9 points (4.5 on average), whatever the player’s Pressure.
So low Pressure makes a hit more likely, but it doesn’t make the hit bigger. It only really matters at about 5 or below. We haven’t yet traced which events trigger it.

3. Why is Pressure weighted towards attackers, and is it like Jumping Reach?
Partly. Choking affects every kick, but a worse roll matters most on shots, and it’s strongest on penalties and set pieces. Keepers don’t choke when saving, because saves aren’t kicks. So most of the benefit goes to your penalty taker, free-kick taker and main finishers.

Two caveats:
  • Nerves hit everyone. Very low Pressure lowers every attribute for any player, so a player near 1 is a liability in any position.
  • Above about 13, more Pressure only clearly pays off for attackers (AM most of all). Defenders and keepers gain little.
4. Sportsmanship
The earlier impression is mostly right. It only matters at the very bottom of the scale, and even there the effect isn’t established:
  • At Sportsmanship 1 it costs about −0.26 points a game in my squad save and −0.12 in neutral squads.
  • At 20 it does nothing, and a test using normal values would correctly find nothing.
How it works: the Sportsmanship roll collapses towards 0 only when the attribute is near 1, the same pattern as Pressure. A low roll allows cynical fouls when the player’s Aggression roll beats it, and Dirtiness sets how keen he is to commit them. More fouls mean more cards, more free kicks for the opponent and more broken-up play: about 17 fewer passes and 1.6 fewer shots a game. The engine reads Sportsmanship in at least ten other places that we haven’t decoded yet.
 
I’ve been reverse engineering the FM24 match engine (24.4.2 Steam build, reading the code statically). Nothing here is verified with the in-match harness yet, and FM27 may differ in details. But 95% of the logic seems to be the same for 15 years now anyway

1. Pre-match nerves come from Consistency and Important Matches, not Pressure (FM24)

At match prepare the engine rolls one “nerves” value per player. The roll’s input is a blend of two hidden attributes:

R = (ImportantMatches × (20 − u) + Consistency × u) / 20

  • u depends on how big the match is for this player. The engine computes a match importance I (1–20) from:
    • the competition’s reputation;
    • the competition kind (a friendly is always 1);
    • both teams’ table situation (title race, promotion, relegation);
    • the opponent’s reputation.
  • It compares I with the player’s stature T = (team rep + his current rep + his world rep) / 1350.
  • Below his stature the match is ordinary for him and u is near 20, so Consistency decides. Above it u drops, and Important Matches takes over.
  • For a top-club squad most league games are pure Consistency. Important Matches only counts in genuinely big games, or for low-reputation players.
I found no Pressure input in this roll. The nerves value is subtracted from the attributes for the whole match, with a fixed weight of 20 in the engine’s weight table. That works out to roughly 0.04 attribute points per nerves point.

So “nerves lower every attribute all match” matches FM24. Attributing it to Pressure does not. If Pressure 1 vs 20 moves nerves in FM27 tests, either FM27 changed this, or the test squad also changed Consistency or Important Matches.

2. Where Pressure does enter (FM24)

  • Penalties and set pieces.
    • Execution roll: Q = U[0, skill) − p4. The skill is Penalty Taking for penalties, (3·Free Kicks + Long Shots)/4 for long free kicks and (3·Corners + Crossing)/4 for corners. p4 is a situational pressure term, or a weak-foot term when the off foot is forced.
    • Penalty aim: a “blasted” penalty only goes for the corner if Q ≥ 40 − N, where N is read from the Pressure byte.
    • This fits the “chokes on penalties and set pieces” observation. Penalties, free kicks and corners have no separate success roll; the outcome comes from the simulated ball path.
  • Crowd pressure. Each player gets a crowd term (capped at ±55, weight 20). It comes from his own fans’ share of the attendance, his morale, his Pressure and the team’s recent form. It is added after the soft cap. Roughly +0.8 / −0.6 attribute points for home / away in a 60k ground, about ±0.2 below 15k, and −0.2 for both sides in an empty ground.
  • “Drive” state. Pressure is one input, together with Determination, Professionalism, Ambition, per-match dice, morale, score and minute. It is limited to ±25, which is about ±1 attribute point.
3. The captain blend is real, but small

In FM24 the engine blends five personality values with the captain’s, weighted by the captain’s influence / 50: Pressure, Professionalism, Loyalty, Temperament and Sportsmanship. The influence ranges 1–30, but freshly prepared matches showed 1–5, which makes the captain’s share 2–10 %.

So a calm captain helps a nervous player a little, not a lot. There is also a separate captain term on all attributes:

SituationEffect
With a captainabout −0.5 attribute points for everyone (a constant)
No captainabout −1.4
Which player is captainonly about ±0.1
The captain’s morale and influence come from his relationships with the manager and the club.

4. Things in the thread I can’t confirm from FM24

  • Morale loss after bad events for Pressure < 15 (chance ≈ 1/Pressure², 0–9 points): I didn’t look for this yet.
  • Temperament only feeds discipline and anger behaviour in FM24, not attributes.
  • Sportsmanship has more readers I haven’t decoded either.
  • @meanonsunday’s double-counting point holds for the attribute side in FM24. Pressure does not lower Composure, Decisions or Concentration in the attribute getter. The modifier terms there apply to whole attribute classes or flagged groups, not to specific mental attributes.
5. Practical takeaway (FM24)

  • Consistency matters every week for good players, through the nerves roll.
  • Important Matches only matters when the match outranks the player.
  • Pressure matters for penalty and set-piece takers and in front of big crowds.
  • Weighting Pressure ×5 across the board has no basis in the FM24 code.
Happy to share offsets and function addresses if anyone wants to check this against FM27.
 
Back
Top