Blades’ midfield by numbers

At just past halfway in the season (after the Liverpool game) the Blades sit 8th in the table and have had a pretty successful start to their first PL season in more than a decade.

As it’s found its feet, the team has developed its playing style. It has found it can’t play the same way as in the Championship and has evolved a slightly more conservative style, becoming more direct and taking less touches high up the pitch.

That said, the desire to play high and make chances when possible is still there and is a major factor in keeping opponents away from our goal and goes a long way to explaining the very low shots-against count for this Sheffield United team.

Key also to success so far this season is the midfield. With Fleck and Lundstram as the more attacking elements, and Norwood taking up a more deep-lying role, the engine room of the team has achieved good balance.

If we look closely at those three players we can see where they complement each other, and by focussing in on a couple of those metrics we can also see how parts of their on-pitch performances might be changing as the season progresses.

Left, right and deep-lying

First, let’s look at the big picture.

The obvious division of roles concerns footedness – Fleck on the left, Lundstram on the right – and how high or low the player plays, which puts Oliver Norwood in a deeper role and with a less-defined axis to work on.

Also, as Norwood often covers for whichever CB bombs forward on the overlap, he can find himself on the left or right for extended periods.

Here are the heatmaps that show that. These show these players’ touches in all matches this season, with corner-taking removed so there’s no bias for these for Fleck and Norwood.




Between Fleck and Lundstram we can see that the latter has played a slightly freer role, perhaps less bounded by strong and weak foot bias.

Attacking and defensive strengths

Digging deeper we can look at a few metrics that reflect attacking and defensive activity and see how the players complement each other but may also provide strengths and weaknesses up the left and right axes of the pitch.

In the three radar charts of midfielder performance the right side of the chart indicates attacking metrics and the left defensive. (The ranges are approximate ranges for PL midfielders.)

On the face of it, the three charts look pretty similar.

As we’d expect, Norwood’s bulges out to the left more, ranking higher on all defensive counts, with about double the number of tackles as Fleck and nearly double the amount of ball recoveries as Lundstram, for example.

ONradar copy

JFradar copy

JLradar copy


Looking more closely at Fleck and Lundstram, what’s clear is that the latter is more of an all-rounder. Lundstram puts in more tackles, makes more interceptions and commits about the same number of fouls per 90 minutes. He’s also more box-to-box than Fleck.

Fleck makes more ball recoveries than Lundstram per 90, however. That’s when a player gains the ball from situations where no team has possession or has it played to them by an opponent.

The interesting bit of their respective attacking profiles is that Fleck is the one more likely to have taken players on this season. Lundstram is less of a ball carrier, and more likely to send a pass off after one or two touches.

On the flip-side though, Lundstram is more likely to get involved in passing in the last 20m of the pitch (deep completions) than Fleck. There seems to be a passing hot-spot high on the right in nearly every game this season and Lundstram has usually been part of it. He’s also often the one to receive throw-ins in that part of the pitch.

The time factor

The trouble with the charts so far though is that they are season averages. In fact, players can have a run of games where they do, for example, take players on repeatedly, then a few where they don’t.

So, while we can see which players tend to do more of one thing than another, we need to look at rolling averages (or even per-game figures) to get a more realistic picture.

So, for example, while John Fleck comes out as the take-on meister of our midfield trio, looking at the data in time series shows he came back from injury to his record number of take-ons this season against Southampton on 17 September, but hasn’t quite achieved that since and his average has dropped away.


Meanwhile, however, Fleck’s average number of ball recoveries has been increasing during the season. That could be down to his individual performance or could be down to changes in team tactics. Maybe the evolution towards occasional bouts of energetic pressing has seen our no.4 pounce on loose balls more often?


What’s clear from looking at all these stats and charts is that while they can give lots of insight, they can also mask a lot of things. So while we’ve seen here some key features of how our midfield three tend to play, we’ve also seen that averages can mask a lot.

What we do know, though, as Blades fans is that these three have been a revelation in stepping up to a higher level and being the engine room of a fine first half of a season back in the PL.

Addendum: Defensive actions maps

Here are where the three midfielders have carried out recorded defensive actions so far this season. These pretty much accord with the players’ pass heatmaps, with Norwood the deepest. The other two have a bias towards each flank with Lundstram being a bit busier than Fleck, up and down the pitch as well as across it.




Foster saves Watford. Blades settle for a point.


One look at the shot count, pass count, xG timeline or xG shotmap shows Sheffield United to have dominated this game. Ben Foster made some magnificent saves to deny the Blades.

Having said that, it didn’t feel like that watching the game live. Charting Blades passes per 10 minute period shows a lot of activity early in each half, and that seems to be faily usual for Sheffield United.

The passmap, however, also shows that a lot of passes were in our half and beyond Watford’s front line. They came to play deep and block up the central areas in front of goal with the hope of getting an opportunity on the break via Deulofeu, which they did.



Wilder wins the tactical battle at Brighton

Brighton (0) vs Sheffield Utd (1) review, 21-12-19

This was a game of two halves in a tactical sense.

In the first half Brighton expected to play Potter-ball. IE, passing their way up the pitch, with Gross and Trossard as key creators and Maupay a focus for them in our danger area.

That didn’t happen, though.

The Blades took the game to the Seagulls, pressing high when the opportunity arose, and creating chances. Out of possession we often let their CBs have the ball.

Mid block . . . 
The result was an overload in our favour in front of that, from their final 1/3 towards our goal. Often three Brighton CB were trying to pass the ball to seven of their players while 10 of ours marked them or blocked passing lanes.

In effect we clogged up the middle of the pitch and Brighton sent the ball long to bypass the midfield numerous times, to little effect.

You can see that here in the first half passmap for Brighton.


Half time substitutions are not common but Potter made two, taking off Trossard and Gross and replacing them with big frontman Murray and little pacy frontman Connolly.

It felt to this writer like Potter was saying, “We can’t get the ball up the pitch by playing football so we’re going to have to launch it.”

. . . Low block
But, the Blades, a goal to the good, sat back and absorbed it and let Brighton pass it around outside our final third and dealt with anything that came in.

You can see that in the passmap below. And also where Brighton’s subs played (the squares; the top and right ones are Murray and Connolly)


The chart below shows a timeline of passes bucketed into 10 minute periods and clearly shows when we let them have the ball more.


The PPDA (opponent passes per defensive action) chart also shows how we pressed in limited periods, mostly around the start of the halves. McBurnie’s goal came during a period (23 min) in which Blades’ pressure was low, and was in fact assisted by Henderson’s ball from inside his 6yd box.

PPDA needs to be seen as simply showing what happened. Whether a team pressed a lot and allowed fewer passes to the opposition is only a good thing if that’s what they want during part of a game. There are going to be periods for some teams where they want the other side to have the ball, however, and it’s probably fair to say that’s the case for Sheffield United for chunks of the 90 minutes on Saturday.


Blades at Brighton: An entertaining meeting of styles on the cards

Brighton vs Blades preview, 3pm 21 Dec 2019

BHASHUXaExpected assists and assists for BHA and Blades players. Pascal Gross stands out. xA explained below.

Brighton sit 13th in the PL with 20 points while United are 7th and have 25. Brighton are much improved on last season and have a more attacking character under Potter, with a league position that appears ‘honest’ (beating teams they should, losing to ones they’re more likely to).

The Blades and Brighton have scored the same number of goals (21) but Brighton have let more in (25, Blades 16). Last eight form is 1.38 points per game (United 1.63).


In terms of expected goals/against (xG/xGA), Brighton over-perform their xG while xGA is roughly where it should be. United are the other way on this, getting about the number of goals xG says they should while out-performing xGA.

Of the two teams’ individual players, Blades’ summer target Neil Maupay stands out in terms of xG, as seen here.


The number of shots taken by Brighton is 218 while they have allowed 229, with Sheffield United chalking up 174 and 191 on these counts.

Formation-wise Brighton appear to be very tactically flexible and have employed a wide range of shapes from a 5-4-1, through 4-1-2-1-2, to 3-4-1-2 so far this season. But for the most part 3-4-3 and 4-4-2 have been their preferred basic shape.

All this may point to Graham Potter setting up to suit the opponent at hand and with United almost invariably playing three at the back, BHA could set up similarly.

In terms of style Brighton are happy to play through the thirds or go long, with Pascal Gross a key link in midfield and topping the expected assists (*xA) chart (see top of page).

In terms of directness**, as averaged over all games so far, the two teams are near opposite ends of the scale, with Brighton’s desire to play from the back and build perhaps registering more than a tendency to go long.


Sheffield United’s high ranking in terms of directness probably masks their tendency to play quite high for periods during games.

It should be an entertaining game. For periods Sheffield United will try to take the game to Brighton and the Seagulls will be happy to start from low down the pitch. Both approaches have risks, of course, so it’s a question of who manages those and takes the opportunities that come.

(* xA or expected assists sees players assigned the xG value of shots that results from their pass)

(** Directness is calculated by the proportion of pass distance that is towards the opponent’s goal divided by total pass distance. Average line is the average of all players’ touches in a game.)

Blades vs Villa: xG close but masks the story as visitors fail to hit the target


United won the xG battle (1.13 to 0.89) but that doesn’t tell the whole story. Both teams had about the same amount of shots (7 and 8) but Villa didn’t have one on target and looked fairly toothless for a lot of the game.

Blades’ goals came on 50′ and 73′, with Grealish’s missed penalty at 77′ (not on xG timeline).


This was also a rare occurrence in that the Blades had more possession than the opposition. Below are possession numbers by 10-minute period for the game.


Once again, and bucking last season’s trend, a lot of the offensive action came down the right. You can see who was involved when the ball got to that area in the average position map below — Basham, Baldock, Lundstram and McGoldrick — with Fleck benefitting twice from balls from that area of the pitch as he came in from the left/centre.


Blades come up against a generous Villa, and hope for an early Xmas gift

Sheffield United vs Aston Villa, 14 Dec 2019

(Blades 8th WDDDLW; Villa 17th LLWDLL)

This clash looks set to include some contrast in fortunes so far, and in styles. Villa have been a little more free-scoring than United, with 23 goals for (SUFC = 19).

Meanwhile though, Villa have given up the most shots in the division — 305, against Sheffield United’s 183.

Those numbers are reflected in the expected goals (xG) and expected goals against (xGA) stats, as seen here. That spanking off Leicester last week didn’t help Villa’s xGA line.


We may also see some clash in styles, with Villa tending to sit deeper and look for opportunities to counter while the Blades push on, at least until there is a lead to protect.

That’s most apparent in this chart, which shows how ‘sticky’ teams are up top by measuring their average line against numbers of passes in the last 20m (deep completions).


That Villa tend to build rapidly and not hang around in the opponent’s final third is also shown by the fact their touches in the box outweigh their deep completions (as a per-game average). For Sheffield United those numbers are the other way round, although the recent trend has been for that to reverse with more direct play evident.


Blades at Norwich: Two quick goals allow Wilder’s boys to get choosy about pushing on

Chris Wilder said if there was a Blades swear box it would have been a lot fuller after half time at Norwich. And most people seemed to agree that United went out with extra vigour in the second half.

That resulted in two quick goals around the 50-minute mark, then we let Norwich have more of the ball in the second period but pressed them at opportune moments.

You can see the relative numbers of passes per 10 minute period here.


And here are just Sheffield United passes, with a very noticeable drop-off in the second 45.

plot of chunk unnamed-chunk-12

Here’s where we pressed them in the top 3/5 of the pitch.


Here are some more charts for extra context.


team Sheffield Utd Norwich City
xG 0.605 1.090
shots 9 14
onTarget 3 4
touches 616 769
deepC 19 17
touchesInBox 23 26
passes 460 623
passSucc 0.685 0.767
ppda 7.16 7.67
directness 0.585 0.516
avgLine 47.8 46.7
shots6 NA NA
shotsPen 6 8
shotsOutside 3 6

plot of chunk unnamed-chunk-6

plot of chunk unnamed-chunk-8

plot of chunk unnamed-chunk-15

Preview: Blades vs a leaky Norwich

Preview: Norwich vs Blades, 8 Dec 2019

The starkest stat to start with for this match is the goal difference table. This tells the story of a Norwich team that outshoots the Blades by a little, with almost the same goals scored tally, but which allows lots of shots and lots of them converted.

Norwich goals for is 16 (Blades 17) while the Canaries goals against is more than double the Blades (32 vs 15).


That’s even more apparent when we look at shots against and goals against per game, where Norwich are among the worst in the PL. They’ve racked up a staggering 256 shots against so far.


That’s reflected in the expected goals/against tally too, which measures numbers and quality of shots taken and allowed. While the Blades keep shots against pretty much in line with their shots for, Norwich allow way more shots than they take.

plot of chunk unnamed-chunk-4

Norwich are way less direct, and play much less high than United. That’s shown by their lower deep completions numbers as well as, obviously, their average line, portion of time spent in opponents’ final thirds, and directness numbers.

team Norwich City Sheffield Utd
P 15 15
F 16 17
A 32 15
xG 16.1 18.4
xGA 29.0 17.4
shots 173 158
shotsAgst 256 169
passesPG 494 445
passSuccPG 0.797 0.726
ppdaPG 12.00 9.14
deepCPG 11.3 23.5
touchesInBoxPG 23.2 19.9
avgLinePG 43.5 49.2
opp3rd 0.218 0.297
directnessPG 0.487 0.542




plot of chunk unnamed-chunk-7


I wear the caveat of shame, but I was right about Newcastle’s deep defences

Sheffield United 0 vs Newcastle United 2, 5 Dec 2019

Well, I am today wearing a big pink caveat to atone for the over-confidence implicit in my pre-match blog post.

But I am satisfied that I called the shape of the game correctly. Newcastle demonstrated an almost bizarre unwillingness to press and a defensive depth that would have pleased WW2 Soviet generals*. As a result, some Blades numbers were Man City-like. More of that below.

What can we learn from this most disappointing of defeats? (I’m leaving VAR out of this, btw).

I think Wilder and Knill took the right approach to start — we’d outshot Newcastle by 8-to-1 with 5 on target by the start of the second half — but I think we paid the price for not having the tactics to winkle out a stubbornly dug-in opponent. More of that below too.

They made us look like Man City (except in one key stat)

Last night we broke a lot of records for the season so far. They were:

  • Most touches (883)
  • Most passes (758)
  • Most deep completions (passes in final 20m) (34)
  • Most touches in the opponent’s box (31)
  • Most touches in the final third (285)
  • Second highest average line (55.2m — against Leicester it was 56.2m)
  • Least direct play (39% of pass distance was towards goal — Man City average 36%)

Tellingly, the one stat that wasn’t a record was shots (13). This was fewer than against Southampton, Palace and Tottenham and only one more than against Liverpool and Manchester Utd.

Here’s our passmap. You can see the areas of the pitch that Newcastle occupied in and around their box.


Here are our pass numbers, bucketed into 10-minute periods. There’s only one spell where Newcastle passed more than we did.


Here’s our average player position map. Only three United players average positions was in our own half. By contrast, only three Newcastle starting players average positions was in our half.

plot of chunk unnamed-chunk-15

Newcastle’s remarkable aversion to pressing

This chart shows United’s passes in the top 3/5 of the pitch and Newcastle’s defensive actions in the same area. The lack of pressing is quite staggering and is a result of not spending much time in our half, and being very happy to allow us to come onto them.


This is the Blades pressing map, for a bit of contrast. At a PPDA (passes per defensive action) of 5.82, that’s a reflection of a diametrically opposed approach in which we aimed to retain the ball when we had it and get it back if we didn’t.


Did Wilder get it right?

Knowing that Newcastle would sit in, CW/AK went for a mostly familiar starting line-up, but with Billy and Oli up front. In Billy’s case in particular that seemed to be predicated on the likelihood we would play high and pass lots and get our chances without the need for masses of pace. It was going to be about grinding them down from lots of high possession.

And to a large extent it worked. By the second half we’d had enough chances and shots on target to justify being ahead, and it’s fair to say Newcastle’s keeper played a massive part in that.

So, I think pretty much we got the first stages of the game right.

Could we have changed things? In a way that made a difference?

I think we could. When Mousset came on he was effectivly asked to do the same as the forwards had been doing all night; to try and grind away at getting through a very crowded final third and heavily-fortified penalty area.

And lately we seem to have realised that that approach doesn’t work for us. So, we’ve adapted to use of more direct tactics, as I’ve outlined here.

This works for two reasons. Firstly, against top class opposition we’re likely to be pushed further back up the pitch than we were in the lower divisions. Secondly, against defences that are a cut above what we met in L1 and the Championship it simply doesn’t work to play high and try to fight our way into the box all the time.

Instead, we seem now to more often break from the middle, often using Mousse’s pace to create something in much more open territory. Last night that space wasn’t available, and I think what was lacking was the ability to pull Newcastle’s back 10 up the pitch to create it.

Now, I’m not surprised we can’t just select that tactical approach from the toolbox. It’s not something we have really come up against much in the PL and, well, it may not work anyway, especially when the opponent has a lead to protect.

But I can’t help thinking that a way of pulling opponents out of their trenches so we can use pace in space behind them is a thing we needed to try later in last night’s game.

plot of chunk unnamed-chunk-6

team Newcastle Utd Sheffield Utd
xG 0.455 1.420
shots 6 13
onTarget 4 6
touches 446 883
deepC 17 34
touchesInBox 15 31
passes 285 758
passSucc 0.649 0.825
ppda 43.2 5.82
directness 0.693 0.390
avgLine 41.2 55.2
shots6 NA 1
shotsPen 3 9
shotsOutside 3 3

(* Military history reference. The Red Army was noted for defensive preparations that involved kilometre on kilometre of defensive lines made of trenches, anti-tank weapons, dug in infantry etc, such as at Kursk in 1943)







Blades should have too much for a very cautious-looking Newcastle

Blades vs Newcastle. 5 Dec 2019

Poor at both ends
For the first time this season Sheffield United should, on paper, have too much for the opposition. Caveats based around, “This is the Premier League” and “It’s football; the ball is round” etc do apply, but Newcastle perform badly in some key areas and look like they’ve retreated into a safety-first mode of play.

The long and short of it is they don’t score many goals (four less than the Blades, who are not exactly free-scoring so far this season) and let a lot more in (22 against SUFC’s 13).

That is reflected in expected goals/against since the beginning of the season.

plot of chunk unnamed-chunk-4

Player xG contributions
In terms of creating goal scoring chances we can look at both teams’ players contributions. And it is truly remarkable to see John Lundstram and David McGoldrick sitting high and right on the chart that measures xG and number of shots (below).

plot of chunk unnamed-chunk-5

Ultra-cautious home and away
Formation-wise they tend to play 5-4-1 at home and away, although they have started with 4-4-2 a couple of times. They may go for  5-3-2 at Bramall Lane, however, as they did at Wolves.

Newcastle are a real outlier when it comes to pressing and average line held by their players. They appear to hardly press at all and rely heavily on counter attacking. They are the second most direct team in the league (behind Burnley).


Below is a passmap for their away win at West Ham. Remarkably, they got three goals from so few passes in West Ham’s final third.

It also goes to show they can be a threat — as can any team in this division — but the numbers don’t seem to indicate they are consistently dangerous.

plot of chunk unnamed-chunk-8

Early season lessons?
If we look at their directness over time since the beginning of the season, we can see that Newcastle have settled into their current style. A couple of losses followed by a win against Spurs in August may well have convinced Bruce and his team that sitting deep and being very direct is the answer. We can’t know how intentional all this has been but it certainly is how they’v ended up playing.

There wouldn’t seem to be much of an option for Chris Wilder than to go for it early on against Newcastle. The more a goal against forces them to come forward the more we are likely to punish them.

But please don’t forget the caveats at the top of the page.



team Newcastle Utd Sheffield Utd
P 14 14
F 13 17
A 22 13
xG 13.5 17.0
xGA 25.9 16.9
shots 144 145
onTarget 50 49
shotsAgst 217 163
passesPG 337 422
passSuccPG 0.707 0.719
ppdaPG 19.96 9.49
deepCPG 10.2 22.8
touchesInBoxPG 14.6 19.1
avgLinePG 43.4 48.7
opp3rd 0.208 0.294
directnessPG 0.570 0.553