The Art of Data Analysis in FM26 – Position Radars – A FM Old Timer

There’s a certain type of manager who doesn’t fully trust a scout’s star rating or a recruitment analyst’s A+ verdict without knowing why it’s been given. If you’re the sort of player who’d rather look at what someone has actually done on a pitch than what their attributes tell you that they could do, then this post is for you.

When I first wrote a blog showing how to use player radars in previous versions of Football Manager, I criticised SI for mixing the scouts’ recommendation with the analysts’ data analysis. This is something that has been rectified in FM26 but is still hidden away; you have to scroll a lot when you find it; it doesn’t provide many/any negatives for players and doesn’t allow for cross-league comparison. As such, there’s still a way to go for the in-game analysts’ reports.

With Claude and Copilot’s help, I’ve built an Excel workbook that takes a season’s worth of exported player data across the top five leagues (Bundesliga, Premier League, Ligue 1, Serie A and La Liga) and turns it into five position-specific radars – Centre Back, Full Back, Centre/Defensive Midfield, Attacking Midfield/Winger, and Striker. Each one profiles a player against everyone else who plays that role, using StatsBomb-style On-Ball Value (OBV) as the backbone. It’s not a perfect replica of what Hudl Statsbomb or Opta do behind the scenes – it’s a heuristic proxy built from what FM26 actually gives us – but it’s proven a genuinely useful way of cutting through the noise when I’m looking at a shortlist and wanting to use data to do so.

Below, I’ll take you through how to get the data out of FM26, how to get it into the workbook, why you shouldn’t skip the manual entry of team possession data, and then how to actually read what the radars are telling you – with some real examples pulled from my own save.

---

Extracting the data out of FM26

The bad news first: if you’re interested in exporting data out of Football Manager, you’ll know that Football Manager 26 dropped the old Ctrl+P export-to-web-page functionality that previous versions had, which is the method I’d relied on for my FBref-style data dashboard back in FM23. The good news is that the community has already plugged the gap in the form of a free BepInEx plugin called FM26 Player Export, which brings that functionality back

Before you can use it, though, you’ll need to install BepInEx. BepInEx is a general-purpose modding framework for Unity games – FM26 being one of them – and it’s what allows plugins like this one to hook into the game and read what’s on screen. Installing it isn’t difficult, but it’s fiddly enough that I’m not going to try and out-explain a video that already does it well. MustermannFM has put together a clear walkthrough of the whole process, and I’d recommend following it exactly as explained:

Once BepInEx is running (you’ll know because a console window pops up alongside the game when you launch it), grab the FM26 Player Export plugin from FM Scout or Sort It Out SI and drop the .dll into your ‘BepInEx\plugins\FM26PlayerExport\’ folder. The link to that .dll file is also in Mustermann’s video description. Fire the game back up and you’re ready to pull data.

---

The links to the downloads you’ll need

You can download the Excel Player Radar template here:

Rather than have you build the search filter and view required from scratch, I’ve done the legwork. As I’m playing in the Bundesliga with my Bayer 04 Leverkusen team, I’ve created a single Player Database filter set to the top five leagues (Bundesliga, Premier League, Ligue 1, Serie A and La Liga) with a minimum of 1,000 minutes played. That minutes threshold matters more than it might seem – without it, you’ll end up with a long tail of players who’ve made three substitute appearances dragging down (or randomly spiking) the min/max scale that the radars are built on and skewing percentiles for players who’ve barely kicked a ball.

The view that I’ve created has to be used, as it will correctly line up the titles of the player name, club, minutes, etc. for each player with the Raw Data worksheet in the Excel spreadsheet. Don’t make your own unless you’re happy to adjust the Excel spreadsheet yourself, and that will be a lot of work to unpick things. You can download that view here:

Load the filter into your Player Database screen, and with the FM26 Player Export plugin active, do as Mustermann showed you in his video, and this will create a CSV straight into the Football Manager 26 folder in your Documents folder.

One word of caution: because we’re pulling from five entire top-flight divisions with a 1,000-minute floor, over 1,000 players in a typical save. At the minute, the formulas in the spreadsheet are set up for no more than 1,400 players, so you may need to adjust that. As Copilot to adapt it for you, as it will require quite a few changes across a number of formulas. When exporting, don’t be surprised if it takes a little while.

---

Getting the data into the workbook

This part is genuinely the easiest step, provided you do one thing correctly: the CSV that FM26 Player Export produces is semicolon-delimited, not comma-delimited. If you open it in Excel by double-clicking or paste it in without telling Excel this, you’ll get fields like `Best Pos` – which often reads something like `M (C), ST (C)` – spilling across several columns instead of sitting neatly in one, because Excel guesses at the comma inside that value rather than respecting the semicolons around it. Import it properly (Data > Get Data > From Text/CSV, or Data > Text to Columns with Semicolon selected as the delimiter), and this isn’t an issue at all.

Once it’s imported cleanly, select the whole lot – headers not included – and paste it into the Raw Data tab starting at row 2.

There are a couple of gaps in there worth flagging, because FM26 simply doesn’t record everything StatsBomb-style radars show. There’s no Dribbles Attempted, so the workbook can’t isolate failed dribbles as a penalty. There’s no Through Balls, no true Deep Progressions (Progressive Passes stands in for it, which is not an ideal proxy), no Open Play xG Assisted (Open Play Key Passes stands in), and no Dribbled Past, which is why you’ll see Tackle Success% on the radars rather than Tackle/Dribbled Past% like you would on a Hudl Statsbomb radar. None of this is a dealbreaker since it isn’t in game – it just means a couple of the axes are sensible proxies rather than the exact original metric, and that’s noted directly on each Radar Config tab if you ever want to check exactly what’s feeding a given axis.

Once the data’s in, everything else – the per-90 conversions, the OBV calculations, the position flags that decide who shows up on which radar – happens automatically. You shouldn’t need to touch anything else on that tab.

---

Why the possession data actually matters

This is the bit that’s easy to skip but shouldn’t be overlooked. On the Possession tab, columns A and B populate themselves – a dynamic formula pulls every club and league straight out of Raw Data and keeps it sorted alphabetically by league – but column C, Season Possession %, is entirely down to you to fill in by hand, club by club, which can be found in-game within each of the league statistics pages.

Here’s why it’s worth the effort. A handful of metrics on the radars – PAdj Pressures, PAdj Tackles & Interceptions, and PAdj Clearances on the Centre Back radar – are possession-adjusted. Many of you will be familiar with these metrics by now if you’re into statistics, but if you’re not, then the logic is simple: a player can only make a tackle, an interception, or a pressure while the opposition has the ball. A centre-back at a team that dominates possession simply gets fewer chances to make defensive actions than a centre-back grinding it out for a team that sees 40% of the ball, purely because of how their team plays, not because they’re worse at defending. Without adjusting for that, you’d systematically undervalue good defenders at good teams and overvalue busy ones at bad teams.

The workbook does this by scaling each player’s raw per-90 figure by the ratio of the league’s average opponent-possession share to their own team’s opponent-possession share – so a defender at a 65%-possession side gets his defensive numbers scaled up to reflect how few chances he actually gets to make them, and a defender at a 40%-possession side gets his scaled down accordingly. It’s worth pointing out that this is ideal, as it doesn’t account for the game-by-game position figures, so it may not be reflective of the games that the player did play in if they didn’t play every game. Having said that, I can’t imagine you’d want to work out the exact possession statistics per game of each side, because even then it would be flawed if they were a substitute/substituted, given the stated percentage of possession wouldn’t likely be representative of the possession their team had when they were on the pitch.

As you open it, this will be set to 50% by default. If you don’t want to take the small amount of time to find the team possession stats (and it really doesn’t take long at all), then leave it at that. If you do, every PAdj Factor sits at exactly 1, and those metrics are just their raw, unadjusted per-90 values – not wrong exactly, but not doing the job they’re designed to do either. It’s the only manual data-entry task in the whole workbook, but it’s the one piece that turns “busy defender” into “defender who’s actually doing more with less of the ball”.

---

Reading the radars

Each of the five radar sheets works the same way. Pick a player from the dropdown (it only lists players who genuinely qualify for that position, and there’s a Min Age/Max Age filter that narrows the dropdown further if you’re specifically hunting wonderkids or veterans), and everything updates – the bio strip, the chart, and the percentile table underneath it. The dropdown list can be found in the player radar table to the right.

The chart itself is normalised 0–1 against the entire pool of qualifying players at that position, regardless of what age range you’ve set the dropdown to, so the scale doesn’t quietly shift under you every time you narrow your search – only the list of names you can pick from does. Don’t hide this column or you’ll break the radar. The percentile column works the same way, and it’s colour-coded from blue (low) through orange (mid) to dark red (high), so you can tell at a glance whether a number is good without doing the maths yourself.

Here’s how a few real players from my own save actually profile and how to read their data:

Roger Drets (Barcelona, M (C), aged 18) is the one that most obviously justifies the “spot the wonderkid via data” pitch I opened with. He’s sitting in the 99th percentile for both Progressive Passes and Open Play Key Passes among centre/defensive midfielders across the top five leagues and the 86th for Pass OBV – genuinely elite ball progression and creativity for an 18-year-old, let alone one being trusted with minutes at Barcelona. Defensively he’s unremarkable (39th percentile for Defensive Action OBV, a middling 51st for Tackle Success%), which tells you exactly what kind of midfielder he is before you’ve watched a single minute of him: a deep-lying playmaker in the making, not a ball-winner, and one worth watching closely.

Owen Roberts (Man UFC, D (C), aged 32) shows why you shouldn’t read a radar as one single verdict. He’s a monster in the air (97th percentile for Aerial Wins) and reads the game well enough to rank 60th for PAdj Tackles & Interceptions and 86th for Defensive Action OBV overall – but his Tackle Success% sits at just the 20th percentile. Put together, that’s a picture of an ageing centre-back who wins the ball through positioning and aerial dominance rather than through clean, one-on-one tackling, which is a genuinely useful distinction if you’re deciding whether he’s still suited to a high defensive line against pacy forwards. It’s worth pointing out that low PAdj tackles and interceptions aren’t necessarily the hallmark of a bad defender. Many elite defenders rank low on this on the basis that they may be high-pressing teams, so the ball is won back more frequently by their more advanced teammates.

Caner Bilen (Rennes, D (R), aged 20) is a good example of not taking a raw OBV figure at face value. His Pass OBV reads as a small negative number, but that actually ranks in the 88th percentile among full-backs – most players in that position group post negative Pass OBV, since the turnover penalty in the formula tends to outweigh a full-back’s more modest chance-creation numbers, so it’s the percentile that tells the real story, not the sign. Combined with a 91st percentile for Progressive Passes and 96th for Tackle Success%, he looks like an excellent progressive, front-foot full-back. The one flag is a 15th percentile for Turnovers – he gives the ball away more than most peers in the position, a risk that comes with a player this progressive and perhaps a bit inexperienced or with poor decision-making given his age.

Denis Panov (West Ham, AM (R), aged 25) and Curtis Potts (Real San Sebastián, ST (C), aged 27)** make a nice contrast in attacking profiles. Panov is a finisher rather than a creator – 97th percentile for xG, 98th for Shot OBV, 95th for xG/Shot, but only the 8th percentile for Open Play Key Passes and 22nd for Dribble & Carry OBV. He’s there to put the ball in the net, not build the move.

Potts, by contrast, is shooting just as well (94th percentile xG, 95th Shot OBV) while also contributing heavily to the build-up, at the 90th percentile for Pass OBV and 88th for Open Play Key Passes – a much more all-round attacking presence, albeit one who’s weak in the air (25th percentile for Aerial Wins) for a nominal target man.

Identifying a player’s weakness is just as important as identifying their strengths because if your system relies on something that a potential transfer target isn’t good at, then you run the risk of breaking your own tactic. Having said that, a player not doing something doesn’t mean that they can’t do it – it could well be the case that their manager plays them in a particular role that doesn’t lend itself to something that you would prioritise or that they’re in a tactic that restricts their underlying ability. This is where further investigation is needed.

None of this replaces fully scouting the player, obviously – but it tells you what to look for when you do, and it flags players whose underlying numbers are worth a scout’s report even if their attribute profile or reputation wouldn’t otherwise put them on your radar. You’ll still need to find out their personality, preference towards big matches, consistency and their injury proneness, although 90s played go someway towards this. Remember that availability is the most important player metric of them all.

---

Adapting it to other leagues

I’ve built this around the top five for the obvious reason that it’s where the deepest talent pools sit that I am likely to recruit ready-made players from, but there’s nothing hard-coded that ties the workbook to those specific leagues. The Radar Config tabs work entirely off whatever’s sitting in Raw Data and Season Aggregates – there’s no league name baked into a formula anywhere. If your save is based somewhere else, or you simply want to widen or narrow the net, all you need to do is:

– Build (or adjust) your Player Database filter in FM26 to the leagues you actually want, keeping the minutes threshold in place so the data isn’t skewed by bit-part players.
– Export and paste into Raw Data exactly as before.
– Add any new clubs to the Possession tab and fill in their possession percentages – the dynamic club list will pick up anything new automatically; you just need to supply the numbers.

Widening to more leagues just means a bigger player pool to compare against – useful if you’re managing somewhere away from the top five leagues and want a wider net – and narrowing it means a tighter, more relevant one if you’re only ever going to be shopping in a couple of specific markets.

---

I hope this proves useful for those of you who lean towards this way of assessing recruitment – it’s certainly changed the way I look at a shortlist. As ever, let me know how you get on, or if you spot anything that needs tweaking, over on Bluesky.