metaFM

How we test FM27 tactics

Most tactic claims rest on one save, a few matches and a screenshot. Here every tactic plays thousands of full matches under the same conditions, and only the tactic changes. This page explains how.

Tactics ranked
134
Matches played
88,211
AI game plans
3,977
Game version
27.1.0

“Aren't these just simmed games?”

Fair question, and usually the first one. These aren't quick-sim results, and the things that make simmed results unreliable in a save are exactly what this test takes out.

  • Full matches in FM27's own match engine

    Every match runs the game's own detailed match engine, minute by minute, to the final whistle. It's FM27's engine code, not a formula, a lookup table or a highlights estimate. A match that stops before 90 minutes is thrown away, and goals, cards and stats are read straight from the engine's match data.

  • Your tactic, exactly as in the file

    We checked that the engine plays a tactic exactly the way the game itself loads it: across all 53 tactics we compared, home and away, every match played out identically. Every role, duty and instruction in the .fmf is kept, and a file that can't be carried over exactly is rejected instead of being tested in a changed form.

  • Any match can be replayed

    Play a finished match again with the same inputs and you get the same match, event by event. We replay random samples to check it: 60 out of 60 came out identical.

  • Nothing else changes

    Both sides field identical players, and every tactic meets the same 3,977 AI game plans home and away. A better result can only come from the tactic.

  • Thousands of matches beat one great save

    Luck is real in FM. Over one 38-game season, a tactic's points per game can land anywhere in a range about ±0.35 wide on luck alone. At 560 matches that shrinks to about ±0.1, and at the full 8,008 matches to about ±0.03. That's why one brilliant save proves less than it feels like, in either direction.

  • What it doesn't cover

    No one manages from the touchline: there are no substitutions, shouts or in-match tweaks on either side. The AI side plays its club's real game plan, but not the extra opposition instructions and set pieces it adds in a save. Your squad, morale and form matter too. So read a rank as a tactic's built-in edge: the base you then manage on top of.

Don't take our word for it: download the top tactic, play a season and tell everyone how it went in Discuss. Or upload your own and it plays the exact same test.

  1. 1

    The real match engine

    Every match is played by Football Manager's own match engine, on game version 27.1.0. Nothing is estimated from a spreadsheet or a simplified simulator: goals, chances and xG come from full matches.

  2. 2

    Real opponents, the same for everyone

    The opponents are the actual AI game plans that clubs use in an FM27 save: 3,977 game plans from 412 clubs. Each one carries the club's formation, roles and instructions. Every tactic meets exactly the same pool, so no tactic gets an easier draw.

  3. 3

    Every style of opponent counts equally

    Opponents are grouped by formation (14 formations) and each group carries the same weight in the score. A tactic can't rank high just because one common formation happens to suit it. Every plan is played both home and away.

  4. 4

    Only the tactic changes

    Both sides use identical players, and tactical familiarity is held at the same level for everyone. Squad quality, morale, form and injuries are taken out of the picture, so the result measures the tactic itself.

  5. 5

    Staged screening

    Testing every tactic to the end would waste time on clearly weak ones, so tactics move through checkpoints. A tactic that is clearly behind at a checkpoint is eliminated; the rest keep playing, up to 8,008 matches each. A tactic appears on the leaderboard only after 560 matches.

  6. 6

    Points per game, with an honest range

    Tactics are ranked by points per game (win 3, draw 1). Each tactic page also shows a 95% range: where its true level very likely sits given the matches played. When two tactics' ranges overlap, treat them as roughly equal; a gap of a few hundredths is not a real difference.

  7. 7

    The exact file you download

    We test the original .fmf file, and the download is that same file. Every file gets a fingerprint (SHA-256), so a changed file counts as a new tactic and is tested again from zero. A file that was already tested is recognised straight away.

  8. 8

    Every game update starts from scratch

    A Football Manager update can change how the match engine behaves, so each new game version is tested again from the beginning. Older results stay in the version archive.

FM 27.1.0 at a glance

  • 3,977
    AI game plans
  • 14
    opponent formations, equal weight
  • 8,008
    matches in a full test
  • 6,000+
    matches per hour
  • Same day
    new FM patch to first matches

Built for speed, without cutting corners

6,000+
full matches an hour, on a single machine
~75 min
for a complete 8,008-match test of one tactic
½ day
to take every tactic on the board through its first checkpoint after a patch

Our own testing tools run Football Manager's match engine at full speed, so we don't wait for matches to play out inside the game. That speed is the whole point: it's what makes samples of thousands of matches per tactic practical, and big samples are what make a ranking trustworthy.

  • New patch, new results fast. When FM 27.1.0 was released, our lab was running matches on the new engine the same day and started retesting the whole leaderboard right away, best-ranked first.
  • No shortcuts. Every one of those matches is a complete match by FM's own engine. Nothing is estimated, sampled down or simulated by a simpler model.
  • More matches, sharper answers. At 8,008 matches the 95% range of a tactic is narrow enough to tell a real difference from luck.

A tactic's journey, step by step

  1. 1

    The exact file gets a fingerprint

    We take the original .fmf and give it a SHA-256 fingerprint. The file you later download is byte for byte the one we tested; change one setting and it's a new tactic, tested from zero.

  2. 2

    560 matches against the whole AIfirst checkpoint

    That's 20 full rounds against all 14 formations, home and away, with identical players on both sides. Even this first stretch already covers every opponent style equally.

  3. 3

    Compared match for match

    Its points are compared with the best result any tactic got on exactly the same 560 matches: same opponents, same seeds, same players. More than 0.40 PPG behind? It stops here and stays on the leaderboard with a badge.

  4. 4

    1,120 and 4,004 matches0.25 · 0.15 PPG

    The bar tightens as the sample grows, so only tactics that can still reach the top keep playing.

  5. 5

    The full test: 8,008 matches

    286 rounds against every formation, home and away. Ranked by points per game with a 95% range and a full breakdown.

Where the opponents come from

In FM27 every AI manager builds a game plan for their club: a formation, roles and duties, team and player instructions, in and out of possession. We collect those plans from the clubs of a real FM27 save: the same AI you face in career mode, not hand-picked tactics or default presets.

On FM 27.1.0 that is 3,977 game plans across the 14 formations the AI uses, each formation with the same share of the score, so a tactic can't rank high by beating one popular shape.

When FM updates, the AI's game plans are rebuilt on the new engine too, and the new engine is checked against the game before a single result is published.

The same luck for everyone

Football has luck: deflections, missed sitters, a red card. We can't remove luck, but we make it equal. Every tactic plays the same opponents in the same order with the same random seeds, so two tactics face identical luck and the difference between them is the tactic. That's also why checkpoints compare tactics match for match instead of just comparing averages.

What we record in every match

  • Result
  • Goals for & against
  • xG for & against
  • Shots for & against
  • Possession
  • Home / away
  • Opponent formation

All of it is on each tactic's page: per game averages, the home and away split, and results against each opponent formation with the best and worst matchups highlighted.

What the 95% range means

A tactic's points per game is a measurement, and every measurement has a margin. The range shows where the tactic's true level very likely sits. More matches make it narrower.

The ranges overlap: treat A and B as roughly equal, even though B is a few hundredths ahead.

Why testing sites give different numbers

Serious testing sites all use the same FM match engine, yet the points differ. That's expected: each site measures under different conditions. Compare rankings within one site, not raw PPG across sites. A tactic near the top on several sites is the safest pick.

What changes the resultCommon approachesmetaFM
PlayersTemplate players with varied attributes, or cloned squadsIdentical players on both sides, so only the tactics differ
OpponentsA set of manager-made tactics, or default presetsThe AI's own game plans from a real save, all 14 formations weighted equally
SampleVaries by site560 to 8,008 matches, every tactic on the same matches and seeds
SpeedMatches play out inside the game6,000+ matches an hour; a full test in about 75 minutes
ConditionsMorale, fitness and familiarity controlled in different waysFamiliarity held level; morale, form and injuries taken out
Game updatesVariesSame-day support, full retest from scratch, old results archived

Check our work

🔍 Every number is broken down

Results per opponent formation, W/D/L, goals, xG, shots and possession, home and away, on every tactic page.

📏 Honest ranges

Each tactic shows its 95% range. Overlapping ranges mean roughly equal.

⬇️ Test it yourself

Download the exact file we tested (with its SHA-256) and try it in your own save.

🔒 One fixed setup

Protocol and opponent pool IDs are public. If either changes, the whole leaderboard is retested.

🗄️ Nothing deleted

Every FM version keeps its own archived leaderboard.

🚫 No paid spots

No tactic is placed by hand. Updated three times a day (11:30, 16:30 and 19:30 GMT), and more often right after a game update.

FAQ

How can you test so many matches so fast?

Our own testing tools run FM's match engine at full speed instead of waiting for matches to play out in the game. How they work stays private, but every result is a complete match by FM's own engine.

Why does a tactic score differently here than on another testing site?

Different players, opponents, sample sizes and conditions. The engine is the same, the test is not. Compare rankings within a site.

Why 560 and 8,008 matches?

Every checkpoint is a whole number of rounds against all 14 formations, home and away (560 = 20 rounds, 8,008 = 286), so no formation is ever over- or under-represented at any stage.

The #1 tactic didn't win my league. Why?

Here only the tactic changes. In your save your squad, league, depth and the AI's in-match changes all count too. A top tactic is the best starting point, not a guaranteed title.

My tactic stopped at 560 matches. Is that fair?

It was more than 0.40 PPG behind the best result on the very same 560 matches, against every formation. Its numbers are real and stay on the leaderboard with a badge.

Do you test with real squads?

No. Both sides use identical players, so a strong squad can't hide a weak tactic.

How often is the leaderboard updated?

Normally three times a day, at 11:30, 16:30 and 19:30 GMT. In the first days after a game update, when everyone wants the new results, every hour or two. Tactics still being tested show a "Testing" badge.

Can I get my tactic tested?

Yes, for free. Sign in and use "Submit a tactic"; the author always gets credit. →

Why some tactics play fewer matches

Testing every tactic for 8,008 matches would take weeks. So at each checkpoint a tactic is compared with the best result any tactic got on exactly the same matches (same opponents, seeds and players). That is not always the #1 on the leaderboard: over a short stretch another tactic can score best. It stops once it clearly can't reach the top:

  1. 560 matchesstops if more than 0.40 PPG behind the best
  2. 1,120 matchesstops if more than 0.25 PPG behind the best
  3. 4,004 matchesstops if more than 0.15 PPG behind the best
  4. 8,008 matchesfull test

Fewer matches doesn't mean an unfair result: a stopped tactic fell well behind the best result on the very same matches. Its numbers are real, just less precise than a full test.

Tactics stopped early are still on the leaderboard and are ranked by points per game like every other tactic, with a badge showing where they stopped and how far behind the best result on the same matches they were. Stopping depends on how strong the best tactic was when a tactic was tested, so a tactic tested later can stop with a better score than one that passed earlier.

What the numbers can and can't tell you

  • They show how strong a tactic is against the AI's real game plans when everything else is equal.
  • In your own save your players, league, squad depth and the AI's in-match changes all matter too. A top tactic here is the best starting point, not a guarantee of a title.
  • No tactic is placed by hand and no one can pay for a spot. The order comes only from the test results.

One fixed setup for every tactic

The testing tools are our own and stay private, but the setup is locked: every tactic on a game version runs under the same test protocol and the same opponent pool. Their IDs are below; if either ever changes, the whole leaderboard is tested again.

Test protocol
c57c0efd47c7…
Opponent pool
0073a714bc26…