Monday, 5 December 2016

Has Neil Warnock improved Cardiff?

When Paul Trollope was sacked as Cardiff manager the team was languishing in 23rd place heading for a relegation fight.

Eight weeks later with Neil Warnock in charge, the club is still languishing in 22nd place now very much in a relegation fight with almost half the season gone.

So was it worth it? Has Warnock improved results and performances?

First off, Trollope’s team had amassed just eight points in those 11 games – a rate of 0.73 points per game. Extend that over a full season and it becomes around 33 points – nailed on relegation and quite possibly bottom in some years.

Warnock has flipped those numbers, 11 points in eight games, good for 1.375 points per game. Again if we extrapolate that over a whole season we get 63 points – comfortably top half of the table and potentially just a win or two from being in the playoffs.

Most importantly however, if we extend Warnock’s 1.375 ppg over the whole of the season from him joining we get 48 points.
That in itself should be enough to keep Cardiff out of the relegation zone (just), but added to Trollope’s eight points the Bluebirds should be on course for a steady bottom half finish.

Now this haul has come over a small sample of just eight games – a lucky win here or there can wildly skew the ppg totals – so can we be sure Warnock has improved performances to justify these results?

I'm exiled at the moment, so just watching the couple of televised matches under Warnock it’s been a far better looking Cardiff team than with Trollope, but does that translate into actual solid data? (Remember, appearances can be deceiving.)

Well, yes it does.

Attacking efficiency

Let’s consider the attacking side first.
Warnock has made Cardiff’s attack quite a bit more efficient. Despite cutting down the number of shots Cardiff take per game by roughly 1.5 when at 11v11, the number of those that are not blocked by defenders and that then make it on target has remained almost unchanged.

(click the image for a bigger version)

Looking at locations (under all game states), Warnock has cut down almost two shots per game from outside the box and instead added an extra shot from inside. The biggest increase has been headed attempts from inside the six-yard-box. (Somewhat reassuringly, this is true to form for Warnock and meets what he did at Rotherham last season, as I mentioned in my preview piece upon his hiring.)

While headers are far less effective for goal scoring than shots, if you are going to make them a focus of your attack then you want to do so from as close range as possible. Whether this has been initiated by the types of opportunities created, better crossing, or position of attacking players, it is a welcome improvement that has already yielded two goals from this area.

Overall, although there are improvements, as I wrote previously, this is still a below average Championship attack which relies too heavily on headed attempts and still needs further work to open defences and create chances regularly.

Defensive stability

Given the attacking shortcomings this means the defence has to be above average just to keep its head above water. Warnock has improved on what was a roughly league average unit to make it a well above average group under his charge.

Warnock’s removed a full 2.5 attempts from inside the box during his reign and given up just 1.5 more from outside – a more than fair exchange by anyone’s reckoning.
He’s improved an already strong aerial defence, but most encouragingly Cardiff have cut out almost a full shot per game from both the central and side areas of the 18-yard-box.

Cardiff now concede just 2.25 (foot) shots per game from the danger zone of six-yard-box and central 18-yard-box, well below the league average of 2.95 per game.
Incidentally, the Bluebirds also take 2.25 shots from the danger zone area each game, so a deficit of almost one shot per game in this area under Trollope has been equalised.

In all, Cardiff’s defence has changed from allowing more shots from better locations than the league average under Trollope, to allowing fewer shots from far more distant locations under Warnock.
(I was hesitant about Warnock’s defensive potential in my earlier piece, but I am very pleased to say he has far outperformed my prediction.)

Overall performance

Looking at the overall shot shares under the two managers, we can see Warnock has made noticeable improvements in terms of all shots, unblocked shots, shots on target and goals for.
The team is at least league average in these markers now and better still when the scores are level.

Shooting % (the rate at which shots on target are scored) is typically around 30% within the Championship – although this can be subject to significant short term volatility.

Trollope, perhaps being a victim of poor shooting locations, had a rate of just 22%, while Warnock has had shots being scored at an above average 40%, again perhaps influenced by the improved (though still not great) locations. I would not expect this to continue long term, though the short term correction is much appreciated.

Likewise, goalkeeper save% (the rate at which goalkeepers save shots on target) can be volatile and also dependent on shot location.

Again Warnock’s improved defence may well tell part of the story in this improvement.
However, it is well worth noticing that both Trollope and Warnock’s save% are well below the 70% league average, and it is hard not to wonder if selling TWO first team goalkeepers in the summer without recruiting a suitable replacement may be significantly hindering Cardiff’s progress.

Encouraging signs

Perhaps most encouragingly for Cardiff fans, this improvement in shot shares is seen throughout all score states.
This means the team is able to hold its own or even control the balance of shots when the game is there for the winning, not just when several goals down and the opposition are sitting back. And when ahead it means Cardiff are as likely (or more so) to score a second than concede an equaliser.

Also encouragingly, Warnock has faced a reasonably stiff set of fixtures during his time, including five of the top six teams in terms of 11v11 shots on target share – one of the better statistical predictors of team quality and future success.

The signs for Cardiff are more positive under Warnock and with the whole bottom half of the table (Rotherham excluded) so tight, a couple of wins could easily see the Bluebirds as high as 15th.

Of course football has its shock results (such as Blackburn winning at Newcastle last week) and being dragged into a relegation battle can have unanticipated effects on players and management, but I am now far more confident that Cardiff will be sitting away from the relegation zone in May than I was under Trollope.

Thursday, 17 November 2016

Are Brighton legit and have Gary Monk and Steve Bruce turned Leeds and Aston Villa around?

The Championship is now more than a third played and we’re starting to see some decent statistical trends coming through, although the table doesn’t fully reflect those yet.

In fact, much of the middle is still quite a mess, but at the very top and bottom, the quality, or lack of, is showing.

In most meaningful categories Newcastle are head, shoulders, and sometimes an entire upper body above the rest of the field.

Meanwhile, at the other end, Rotherham look like a League One team already: it would take a serious burst of good fortune to even bring them up to parity with their fellow relegation candidates. Sorry Millers fans.

There are three quite intriguing games on TV this weekend as the major leagues return from the international break, so let's look in a bit more detail at the match-ups involved.

Brighton vs Aston Villa

Behind Newcastle, Brighton have put some distance between themselves and the competition, but while they are a statistically good team, this has been fuelled by the highest save percentage and second highest scoring percentage (just behind Newcastle) in the division. The combined PDO total of 123 (league average is 100) is actually off the (my) charts.

The problem maintaining this form will come if and when this hot streak runs out.

The Seagulls are only a midtable side when it comes to where they take their shots from, although they are one of the strongest defensively. This shows with Brighton just 19th with 29.65% of all shots being on target, while they are best at restricting teams as fewer than one in four (24.57%) opposition shots hit the target.

The other real key to Brighton’s success comes at when they are getting their shots on target. At level score situations Brighton have the third highest shots on target share in the division – 0.625 (or 62.5% of all shots on target).

Combine this with a 40% scoring rate (league average is 30%) and it is easy to see why they are such a dominant force so far.

When you are playing with the lead so often and can force teams to take more than half their shots (99 out of 182) from outside the 18 yard box, you have a very good chance of being successful.

Brighton’s opponents on Friday night, Aston Villa, have had a pretty difficult season so far which resulted in the sacking of Roberto Di Matteo with Steve Bruce replacing him.

What difference has that made? Well, Villa were actually a pretty much league average shots team throughout Di Matteo’s brief reign. He played a more open game than Steve Bruce has shown so far, which resulted in more shots both for and against per match.

But the killer statistic for him was Villa not being able to hold a lead due largely to a shocking 11% shooting % when one goal up. Indeed, Villa only scored ONE goal when leading under Di Matteo, and that has not improved since – they’ve only managed TWO shots on target when leading so far under Bruce.

It’s worth remembering at this juncture that Steve Bruce’s numbers especially are a very small sample size, but they can provide some interesting indicators.

Aside from taking a more defensive stance overall resulting in fewer shots for and against, the one real improvement Bruce has made is in shots on target share, from 0.51 to 0.56. Some of this may be down to shots being taken slightly more centrally in the box compared to Di Matteo’s team, but otherwise it is hard to spot what may be responsible for this. Perhaps it is just a case of Villa’s potentially potent strikerforce finding form.

Which brings us on to the other major boost for Bruce; shooting and save percentages – a fourteen point increase when combined. These factors, which can be subject to high levels of variation, have really powered Villa’s rise, although the aforementioned shots on target improvement will have helped too.

At level score situations Villa are red hot at the moment, and have goalkeeper Pierluigi Gollini rocking a 90% save mark in Bruce’s term. This is likely to cool off and when it does it seems likely Villa will find themselves stagnating in mid table unless other factors improve.

Ipswich vs Nottingham Forest

On the face of it Forest fans should be the most concerned in this fixture, but Ipswich followers have good reason to also be looking over their shoulders.

Life has not been fun at Portman Road this season and while the shot share numbers look generally above average, particularly when scores are level, the locations are far less desirable.

Ipswich have taken the fewest shots (just 30) within the dangerzone (six yard box and centrally in the 18 yard box) and sixth fewest within the 18 yard box as a whole. Instead, McCarthy’s team (as always, it seems) has been surviving on an aerial attack and stout, if not outstanding defence.

This is not necessarily a secure position to be in as a heavy aerial attack typically converts at a much lower rate. And should the goals dry up this way, it could be all too easy to get sucked in to the relegation battle.

Forest are already there however, with below average shot numbers being further hurt by below average shooting and save percentages. Their numbers are slightly more encouraging at level score game states but they are not able to maintain this when leading.

And although the attacking shot locations are generally mid table, they have conceded more six yard box shots than any other team (11) – the only side to hit double figures in this regard.

There are several sides worse than Forest, but the problem is they already have a head start and life can seem so much harder when you are already staring the relegation zone in the face.

Leeds vs Newcastle

The Champions-elect (yes, I said it) visit one of the surprise packages of the season in the final game of the weekend.

I asked before the season if this was the best squad to ever play in the second tier of English football and the answer was generally positive. After a slow start the Magpies have now hit their stride and look a cut above everyone else.

Of course upsets, injuries etc. happen, but the gulf in class is noticeable visually and in the numbers.

Leeds are a strange one indeed and I’m at a bit of a loss to explain their current lofty playoff place perch.

After a very poor start to the season their overall 11v11 shot numbers have pretty consistently improved to at least league average, and even slightly above. Signs that Gary Monk’s system is bedding-in perhaps?

However, their shot locations are very mid-table and they’ve even spent the same amount of time winning as losing (396 minutes) with 767 minutes at tied – all almost exactly league average.

Leeds do a pretty good job of making sure that the few shots they do take make it on target and the win over Norwich was probably their most impressive statistically of the season, following wins over two much more lowly teams, Wolves and Burton.

So perhaps this is just them hitting a run of form, a sign of a fairly weak schedule so far and how close the table is – just three points separate Leeds in sixth with Preston in eleventh.

If Monk can sustain this improvement however, then they could cement a top half place and push towards the playoffs – that would be quite a turnaround from the last few years of statistical awfulness.

How they perform at home against Newcastle, and not necessarily the result, could give a decent idea of where this team is heading.

PositionTeam 11v11 Corsi 11v11 Fenwick 11v11 SoT 11v11 Goals Rate 11v11 Sh% 11v11 Sv% 11v11 PDO Points
14 Aston Villa 0.495 0.517 0.524 0.52 26.15 72.88 99.04 21
13 Barnsley 0.473 0.467 0.443 0.52 37.14 72.73 109.87 21
7 Birmingham 0.478 0.464 0.455 0.54 35.71 74.63 110.34 25
23 Blackburn 0.432 0.428 0.432 0.39 29.63 64.79 94.42 13
12 Brentford 0.489 0.492 0.507 0.58 28.38 79.17 107.55 22
2 Brighton 0.532 0.574 0.578 0.76 42.37 81.4 123.77 34
9 Bristol City 0.519 0.5 0.529 0.54 34.38 66.67 101.04 24
18 Burton 0.529 0.54 0.508 0.49 27.27 70.31 97.59 18
21 Cardiff 0.497 0.528 0.477 0.37 26.92 57.89 84.82 15
16 Derby 0.54 0.532 0.541 0.48 18.18 76.79 94.97 20
8 Fulham 0.54 0.546 0.5 0.59 37.1 74.19 111.29 24
3 Huddersfield 0.552 0.519 0.542 0.5 29.31 65.31 94.62 29
15 Ipswich 0.47 0.496 0.51 0.48 28 68.75 96.75 21
6 Leeds 0.485 0.522 0.526 0.53 31.67 68.52 100.19 26
1 Newcastle United 0.613 0.6 0.648 0.74 43.04 72.09 115.13 37
5 Norwich 0.54 0.511 0.511 0.5 38.03 60.29 98.32 27
20 Nottingham Forest 0.493 0.465 0.479 0.43 34.33 58.9 93.23 16
11 Preston 0.462 0.467 0.481 0.55 35.38 72.86 108.24 23
17 Queens Park Rangers 0.471 0.482 0.517 0.45 28.33 62.5 90.83 20
4 Reading 0.515 0.483 0.459 0.53 31.15 76.39 107.54 28
24 Rotherham 0.346 0.332 0.366 0.31 32.08 59.78 91.86 7
10 Sheffield Wednesday 0.572 0.576 0.555 0.46 22.54 66.67 89.2 24
22 Wigan 0.481 0.472 0.473 0.42 26.42 67.8 94.21 14
19 Wolverhampton Wanderers 0.512 0.523 0.504 0.39 23.44 63.49 86.93 17

Thursday, 13 October 2016

Can Neil Warnock turnaround Cardiff City?

It's been quite a while since my last post, here's hoping I can make it more regular for the rest of the season.

The Severnside derby kicks off the return of the Championship.
Like most local derbies it draws a decent amount of attention and this one has the added Wales vs England element too.
However this one is likely to attract a bit more scrutiny given the events at Cardiff during the international break.

Paul Trollope lasted 11 games in charge and in some respects its surprising he lasted that long.
I'm not usually one for sacking a manager early but in Trollope's case I'm not convinced it was going to work out.
He'd already switched from his initial 5-3-2 formation and tactics, and credit to him for that, but according to the data things were getting worse, not better.

It’s a shame it ended this way as statistically speaking, the season appeared to have started fairly well.
For the first six weeks Cardiff had a decent overall shots on target share above 50% at 11v11 and at level score situations.
However, the Bluebirds' all shot share (Corsi/TSR) only once made it above 50% either under all conditions or at level score situations.

But (poor) shot quality matters.

And when we look deeper into the data we can see why the results did not reflect this limited early season promise.
Trollope's attack, whether by design or simply through not having any other options, had been focused on headed and long range attempts at goal.
While a varied attack can prove profitable (see Brighton), Cardiff have exchanged a sizeable amount of shots from prime positions into headed attempts.
This is not a good exchange to make.

By week 11, Cardiff had taken the third fewest footed attempts at goal in the danger zone (six-yard box and centre of the 18-yard box) in the division - only Rotherham and Ipswich have taken fewer.
Despite being middle of the pack defensively, Cardiff have the joint fifth worst danger zone shot difference - again Ipswich and Rotherham are two of the teams below them.

Looking at the headed attempts at goal, the situation is give-or-take a near perfect mirror image.
Cardiff have the directed the third most headers at goal (behind only Barnsley and Aston Villa), and have conceded the third fewest.

While it may delight commentators to regularly swing the ball into the box, such a one-sided attack does not usually prove effective as headers are much less likely to be scored than attempts from the feet.
Sadly, Cardiff have the second lowest % of their shot attempts being taken from the centre of the 18-yard box (Ipswich lowest).

In fact, Cardiff have been making almost two thirds (63%) of their attempts on goal as headers outside the six-yard box or shots from outside the 18-yard box.
It is entirely possible that taking so many attempts from poor locations is playing a part in the awful shooting % figure.
By contrast, Cardiff have forced just 56% of attempts at goal from these poorer locations.

So instead we have Neil Warnock joining the show.

Personality-wise, Warnock has had something of a love-hate relationship with Cardiff fans. He’s generally been complimentary about the club and its fans, but during his travels around the league he’s managed to rub a lot of Bluebirds up the wrong way.
Still, this isn’t a popularity contest – this is, of course, a results driven business and following guiding Rotherham to safety last year and his wealth of previous experience, Warnock would seem to be an obvious choice.

But how much of last year’s great escape was down to Warnock’s influence and is he likely to have a similar effect on Cardiff?
Well, having gone through the data I’m not convinced Warnock had that big an effect on Rotherham’s survival.
He took over from Neil Redfearn after 30 games with Rotherham battling Bolton, MK Dons and Charlton to avoid relegation.

Credit where it's due?

In the end the Millers completed the task relatively easily (nine points above safety), but I suspect that may be as much down to the awfulness of the other three teams as to Warnock’s magic.
As you can see in the chart below, at level score Rotherham’s all shot share (Corsi/TSR) and unblocked shots share (Fenwick) remained pretty unchanged from Warnock’s start to the end of the season.
There was a slight uptick in shots on target share (green line) but this still never broke the 45% mark – hardly earth shattering but useful at the bottom of the table.

The biggest changes, however, came in the rate Rotherham scored their goals and kept them out.
The Millers’ combined shooting % and save % (PDO) was its lowest of the season (just) when Warnock took over – 17 points below league average.
By the end of the season Rotherham’s PDO had made up 15 points of this difference – with the save % being the main benefactor by more than 10 points.
And it’s so much easier to win games sneaking the odd goal when you’re not shipping them constantly.
Combined with a smaller but important increase of five points in shooting % and Rotherham were just a shade under league average in these key metrics by the end of the season.

So could Warnock have instilled a tactical change to improve these measures?

Well, in all his changes gave Rotherham one extra shot from the centre of the 18-yard box per 7.5 games… so two more shots from this area during his spell in charge than Redfearn would have expected.
But there were ten extra headers inside the six-yard box compared to Redfearn and shots taken from outside the 18-yard box were cut down significantly.
So it is possible this could have accounted for the increase in shooting %.

How about defensively?

Well, aside from a very small drop in the number of headers allowed inside the six-yard box, this does not make great reading.
Warnock’s team conceded more shots per game from the centre of the 18-yard box (+0.5), the sides of the box (+0.24) and outside (+0.9) than Redfearn’s side averaged.
The increase in shots outside the box would be far less of a concern if it meant shots inside the box were being pushed out, but this did not happen.
Of course, the situations may have been different – perhaps Warnock’s defensive system meant opposition players were closed down more frequently when taking shots resulting in poor quality chances.

However, there’s little evidence in my data to suggest anything other than a more normal save % and shooting % driven recovery.
Indeed, at level score under Warnock 35% of shots on target taken were hitting the back of the net and more than 86% of shots on target conceded were saved – both way above league averages of 30%-70%.

For Cardiff fans, it seems that it might be a case of more of the same as regards the aerial based attack, although there may be a focus to not shoot from distance so much.
Also, it is probably fair to say that Cardiff's squad is more talented than Rotherham's (especially given the free agent signings made in the last week) so one would hope Warnock can get more out of this talent.
But perhaps its most notable that in this young season Cardiff have the second lowest shooting % and save %, to give a PDO of just 78.44 – lowest in the entire division by some way.
It’s safe to say the team is due something of a rebound, how much we will have to wait and see.

Monday, 4 April 2016

Why did Middlesbrough buy Jordan Rhodes and is he worth £11m?

Last week Ted Knutson answered a question from Marco Jackson about Middlesbrough’s purchase of Jordan Rhodes.

I’d wondered at the time of the purchase if it was money well spent, especially when Boro were reportedly pursuing Ross McCormack too – a long-time favourite of mine for several reasons.
But Ted’s answer that Rhodes’ data had dropped off and he was posting below average numbers intrigued me. In what way had it changed?

So, I dove in to the WhoScored database to check out his stats.
WhoScored only has data as far back as the 13-14 season, Rhodes’ second at Blackburn, so that does take out his most prolific season in the Championship.
But 13-14 saw him post pretty similar goalscoring numbers to the previous year so it’s probably a fairly safe comparison to make, and we have two consecutive seasons of data at least.

First up, here are Rhodes’ radars from the 2013-14 and 2014-15 full seasons at Blackburn.

As we can see, there’s not too much difference really.
Identical NPGs/90 in each season, his all shot conversion % is higher in 13/14, this dips a bit in 14/15 so he takes more shots per game, but his shooting % from shots on target rebounds instead.
Also noticeable that his key passes/90 fell away almost totally last season, but they weren’t very high to begin with.
All-in-all pretty good numbers for a pretty damn good Championship striker.

So let’s have a look at this season at Blackburn and the data Middlesbrough had to work with when making their decision.

Yikes! Everything went down aside from his all shot conversion %.
Most worryingly, his shots/90 was down by almost a full shot: that’s not the way to correct a scoring slump and falling shooting %.

So what went wrong? And why, bearing in mind these worrying numbers, did Middlesbrough still pursue Rhodes and eventually pay north of £9m?

A striking headache

First we need to understand the attack Rhodes has been a part of the past few years.
As I detailed in an earlier post, both Gary Bowyer and presently Paul Lambert favour an aerial dominated attack – the most aerial attack in the Championship in fact.
In the last two seasons Bower’s Blackburn averaged the most headed attempts /90 in the whole division (4.1 in 2014-15 and 4.0 in 2013-14) with only Ipswich for company (3.9 in 2014-15) and everyone else a full one shot per game behind.
This season they are still ticking along at around 3.7 headed attempts/90, with Ipswich closest at 3.1.
The difference is stark.

Rhodes is a more than capable header of the ball, but he is no Peter Crouch or Andy Carroll. In fact, he’s not even Rudy Gestede, or perhaps more importantly, his partner is no longer Gestede – and that’s the biggest problem as far as I see it.

Rhodes lost his strike partner for around £6m last summer and Rovers have failed to replace him - and its easy to see why.

Gestede was a particularly important player for Blackburn largely because of his physical presence and he was a shot monster - being top in headed attempts and within the top six of all shot attempts by players with meaningful time on pitch both those seasons.

He rarely directly assisted on goals for Rhodes though. Indeed, examining the videos of Rhodes’ goals for the last two seasons it’s hard to spot a Gestede assist in there. (There is, however, some hilariously comical defending at times.)
And the numbers bear this out. But his influence is shown by a moderate key pass total.

However I suspect Gestede was more effective as a decoy, distraction or just a threat in his own right and as a result the pair worked as a unit very well.
His presence alone would have drawn the most attention from opponents’ biggest physical defenders – particularly at set pieces – while he was still able to remain effective. Gestede accounted for 4.2 and 4.0 shots/90 those seasons, more than half of which (2.7 and 2.3 respectively) were headed.
This allowed Rhodes more freedom from imposing defending and a good foil to work off and the pair managed at least seven shots/90 between them both seasons they were together.

But then last summer Villa came calling and being still under a transfer embargo Rovers were unable to resist the cash. That is quite some offensive output to replace in one go.
Rovers tried to do so in the summer (and winter) transfer windows and from within, but have pretty much failed.

Supporting cast

Bengali-Fode Koita arrived as did Nathan Delouneso (who has since departed on loan to Bury) as did Tom Lawrence who returned to Leicester City and is now out on loan at Cardiff City.
(It’s probably unfair to put Lawrence in this group as a direct replacement for Gestede, he’s certainly not that type of player, but he was a forward brought in during the summer to bolster the attack.)
Koita was perhaps the least ineffective of the initial trio, managing 1.0 headers on goal/90 in a limited 700 minutes.

But the most effective replacement came from within – Shane Duffy has doubled his headed attempts /90 output this year (from 0.7/90 to 1.4/90) in more than double the playing time too.
Unfortunately Duffy is a central defender.
This obviously means his threat and support to Rhodes is limited to pretty much set pieces only.

Rhodes has done his part, seeing his headed attempts increase by almost a third from last season, but without an effective strike partner for most of the season it’s been a lonely time for him and opposing teams have been able to focus their defensive tactics on shutting him down, without a proficient partner in crime to worry about.

All of which means his actual foot-based attempts have dropped by almost a full shot /90.
And yet Rhodes remained, to all intents and purposes, Blackburn’s top shot taker this season.

The Middlesbrough dilema

So was it still a risk for Middlesbrough to go after Rhodes? Yes. I was querying it myself, especially at the sums involved.
But taking this wider range of data into mind I believe it was a far more calculated risk  – along with a lot of hope/expectation that in a better playing environment Rhodes would return to his form of the previous few years.
He’s a proven goalscorer at this level and clearly Middlesbrough believe in that record rather than half a season in a poorly functioning attack - this is obviously evidenced by the £2m of performance related payments in the deal.

And while the cash may seem a large sum, it’s worth remembering Rhodes was bought by Blackburn for £8m and is still only 26 – pretty much in the prime of his career for the next few years.
So yes, a gamble, but not as big a gamble as it seemed at the time I would suggest.

And the question that follows of course… has he performed as Middlesbrough expected?

Well, we know he’s had trouble finding the net, scoring just his second goal this last weekend against QPR, but what about his other numbers?

Well, believe it or not, this radar is probably prettier viewing than the previous one.
Yes, his conversion% and shooting% are through the floor, but his shots/90 is through the roof - he’s managed to out-perform both those seasons at Blackburn considerably.

This is just a small sample size so far and of course Middlesbrough would have expected a few more goals from their (potentially) £11m man, but he certainly seems to be putting in the work and getting the chances, it’s just a matter of a couple falling right for him I suspect.

Here are some more of Jordan Rhodes numbers over the last three seasons to give you an idea of his production at Middlesbrough:

Football is not an exact science as we all know, but after looking at this data so far I think Middlesbrough fans should be a lot happier about their investment.
Perhaps a slightly scuffed shot bobbling over the goalkeeper after a defensive howler (as it did at Loftus Road) is just what Rhodes needs to get himself going again.

Finally, I'd just like to add that of course I don't have access to as much data on the subject as I'd like and with Ted's experience at Brentford I'm sure he picked up some deeper data of Rhodes than just this.
Also thanks to the StatsBomb teams for sharing the radar drawing tool with me too.

And if anyone is interested, the list of Blackburn's top shooters this season is an interesting one to say the least and speaks volumes about why the club finds itself well off the pace despite good shot share numbers.

Blackburn top shot takers this season (min 300 mins) Age Position (s) Apps Mins Total/90 Out Of Box/90 Six Yard Box/90 Penalty Area/90
Bangaly-Fodé Koita 25 AM(LR), FW 14 702 2.7 0.8 0.3 1.7
Jordan Rhodes 26 FW 25 2179 2.6 0.3 0.5 1.9
Tony Watt 22 FW 9 379 2.4 0.5 0.2 1.7
Chris Brown 31 FW 13 474 2.1 0.4 0.4 1.3
Ben Marshall 24 D, M(LR) 37 3234 1.9 1.3 - 0.6
Shane Duffy 24 D(C) 37 3251 1.8 0.2 0.1 1.5
Simeon Jackson 29 AM(L), FW 11 316 1.7 0.6 0.6 0.6
Jordi Gómez 30 D(C), M(CLR), FW 12 896 1.7 1 0.1 0.6
Tom Lawrence 22 AM(CLR), FW 21 1188 1.7 0.7 0.1 0.9
Danny Graham 30 AM(L) 11 965 1.5 0.1 0.1 1.3
Chris Taylor 29 AM(CR) 12 557 1.5 0.3 - 1.1
Craig Conway 30 AM(LR) 31 2478 1.3 0.8 - 0.5
Elliott Bennett 27 D(CR), M(CLR) 15 1042 1.2 0.7 - 0.5
Elliott Ward 31 D(CL), DMC 4 301 1.2 - - 1.2
Hope Akpan 24 DMC 31 2496 1.2 0.4 0.1 0.6
Darragh Lenihan 22 DMC 16 1152 1.2 0.8 - 0.4
Matt Grimes 20 Midfielder 8 385 1.2 1.2 - -
Nathan Delfouneso 25 AM(LR), FW 15 727 1.1 0.4 - 0.7
Corry Evans 25 DMC 26 2028 0.9 0.6 - 0.3

Thursday, 17 March 2016

Middlesbrough vs Hull City: The Championship's big guns

Friday night features the Championship’s two best teams statistically speaking – Middlesbrough host Hull, in a game that could go a significant way to deciding the promotion fates of these two teams.

However, were the season governed purely on shot-based analytics, rather than the reality of the league table, both teams would be comfortably clear in the top two.

Pos Team Played Won Drawn Lost GF GA GD Pts
1 Burnley 37 21 11 5 60 30 30 74
2 Brighton 37 18 14 5 50 34 16 68
3 Middlesbrough 36 20 7 9 47 23 24 67
4 Hull 36 19 9 8 50 22 28 66
5 Derby 37 16 13 8 50 35 15 61
6 Sheffield Wednesday 37 15 14 8 53 36 17 59
7 Cardiff 37 15 13 9 48 40 8 58
8 Ipswich 37 16 10 11 46 43 3 58
9 Birmingham 36 15 10 11 40 34 6 55
10 Preston 37 13 13 11 35 34 1 52
11 Queens Park Rangers 37 12 14 11 46 42 4 50
12 Wolverhampton Wanderers 37 12 11 14 45 50 -5 47
13 Leeds 36 11 14 11 36 41 -5 47
14 Nottingham Forest 37 11 13 13 34 35 -1 46
15 Reading 36 11 11 14 41 42 -1 44
16 Blackburn 37 10 13 14 36 36 0 43
17 Brentford 36 12 7 17 48 58 -10 43
18 Huddersfield 37 11 9 17 48 52 -4 42
19 Bristol City 37 10 10 17 34 58 -24 40
20 Fulham 37 8 13 16 56 63 -7 37
21 Milton Keynes Dons 37 9 10 18 31 49 -18 37
22 Rotherham 37 10 6 21 43 62 -19 36
23 Charlton 37 7 11 19 34 66 -32 32
24 Bolton 37 4 14 19 36 62 -26 26

While Middlesbrough have continued on their good form from last season, though arguably at a lesser level in a lower quality league this term, it is Hull that have been a cut above the division.
We can see that Hull started off relatively slowly (given their current standards) but were still above average (50%) in all three 11v11 shot share metrics.

But once Steve Bruce’s team found their stride around week 10 things really took off and the Tigers have never looked back – continually dominating games and outshooting opponents at a near 2:1 pace.
Middlesbrough too had a comparatively sluggish start to the season, but similarly, although to a lesser extent, improved after week 13 and have been hovering around 55% 11v11 shot shares since then.

To emphasise how good these shot share numbers are, the clubs are within the top four with regards to all shot share (Corsi/TSR), unblocked shot share (Fenwick) share and are actually the best two when it comes to shots on target share.

It’s also worth noting as well that there is virtually no difference between the teams now in shooting %, save % and as a result PDO (our catch-all metric for 'luck' and variance). They are both around league average at converting shots on target in to goals (circa 30%), way above the league average (circa 70%) in saving shots on target, and have a PDO of around ten points above average (100) suggesting they have been on the positive side of luck and variance - but no more than each other.

But returning to our shot share metrics, what gives Hull such a significant (approximately 5% points) advantage in all three?

Aitor Karanka’s team has been known for its defensive strength throughout his tenure and Middlesbrough went an ungodly amount of time without conceding a goal earlier in the season, so has Steve Bruce managed to top this?
Well yes. Bruce’s team have quietly bettered Karanka’s defence – at least in terms of shots conceded (by 22 in total; eight fewer from the danger zone, excluding headers).

Combine this with the best attack in the league in terms of numbers and location (most total attempts – 561; most six-yard box shots – 23; second most shots from the centre of the 18-yard box – 113; most danger zone shots – 136; best danger zone shots difference – 65) then Hull City is a pretty formidable beast.

This of course is not totally unexpected given Steve Bruce remained in place after the club’s relegation and retained most of his players as well.

Middlesbrough’s attack is the real difference between the two sides however.
Boro are pretty much league average in total shots taken and while they are slightly above average in terms of attempts from the danger zone, this is nothing to write home about.

Cracking Karanka's shell

However, this is all driven by Karanka’s game-state tactics.
At even score Boro can drive the bus as much as anyone, in fact they are third in Corsi and Fenwick (just behind Hull and Reading in both cases) and fourth in shots on target share.
However, when taking the lead Boro become a far less threatening side, claiming just 47% of all shots taken.

Hull, in contrast, do ease off a little, but still take almost 53% of all shots when one goal up.
And when two goals to the good Hull really drive home their advantage. Boro however remain in the Karanka shell.

2+ Corsi 1+ Corsi 0 Corsi 1- Corsi 2- Corsi
Hull 65.7% 52.7% 58.7% 66.7% 67.4%
M'boro 47.9% 47.3% 56.7% 58.3% 66.7%
Difference 17.8% 5.4% 2.0% 8.4% 0.7%

These trends are by-and-large replicated for both Fenwick and shots on target share as well.
Indeed, such is the magnitude of Karanka’s switch in style that Boro’s Corsi drops off more than any other team in the top six and only four other sides (Ipswich, Preston, Leeds and Bolton) do worse when taking the lead.

It’s likely this approach by Karanka will not have escaped many regular Championship watchers, but it is reassuring to see this visual theory validated and eye-opening to appreciate how pronounced it is.

With Burnley pulling away of late and set to have at least a four point lead over one of these teams after the match, it is a crucial one for their respective promotion challenges.
However Middlesbrough face (on paper at least) a much tougher run-in, including both other top four rivals – Burnley and Brighton – and it may be an imperative for Karanka’s side to claim all three points in this home fixture.

If Boro do manage to take the lead, it will be telling to see if they revert to their normal tactics and can manage to hold on, or for once appreciate the importance of the game and push on.

Karanka's future

As a final aside, I was intrigued by this week’s discord emanating from the Riverside Stadium and questions over Karanka’s future at the club.
It’s impossible to say what would happen if Karanka were to leave Teeside – would a replacement be better or worse, more or less likely to secure promotion. But what is clear is the job Karanka has done in taking Boro to the brink of promotion for two successive years – no easy task.
Stresses and strains happen in high intensity jobs such as managing a football club with high expectation and scrutiny, and it’s surprising these flash points don’t happen in public more often.
If fences have been mended and Karanka achieved a desired break from the office for even one weekend, then it is probably for the best for the club and its promotion chances.

Data table

Position Team 11v11 Corsi 11v11 Fenwick 11v11 SoT 11v11 Goals Rate 11v11 Sh% 11v11 Sv% 11v11 PDO Points
9 Birmingham 0.468 0.439 0.459 0.53 27.69 79.08 106.78 55
16 Blackburn 0.531 0.561 0.549 0.51 23.29 72.5 95.79 43
24 Bolton 0.502 0.502 0.456 0.38 24.31 66.28 90.58 26
17 Brentford 0.487 0.476 0.458 0.45 28.29 71.11 99.4 43
2 Brighton 0.52 0.528 0.52 0.6 33.33 75.94 109.27 68
19 Bristol City 0.492 0.518 0.504 0.39 24.11 61.87 85.98 40
1 Burnley 0.449 0.466 0.517 0.66 37.91 79.02 116.93 74
7 Cardiff 0.491 0.485 0.513 0.56 29.94 75.17 105.1 58
23 Charlton 0.401 0.404 0.448 0.34 22.54 64.57 87.11 32
5 Derby 0.539 0.534 0.554 0.56 29.75 70.87 100.61 61
20 Fulham 0.467 0.457 0.474 0.49 33.74 68.51 102.25 37
18 Huddersfield 0.551 0.546 0.514 0.47 30.26 63.89 94.15 42
4 Hull 0.6 0.617 0.625 0.72 29.71 80.95 110.67 66
8 Ipswich 0.501 0.508 0.52 0.53 28.93 72.11 101.04 58
13 Leeds 0.481 0.449 0.418 0.45 29.46 73.72 103.18 47
3 Middlesbrough 0.541 0.552 0.572 0.69 29.11 82.2 111.32 67
21 MK Dons 0.436 0.43 0.416 0.37 25 70.06 95.06 37
14 Nottingham Forest 0.502 0.51 0.5 0.47 18.79 78.52 97.32 46
10 Preston 0.496 0.504 0.502 0.52 27.42 73.98 101.4 52
11 QPR 0.515 0.52 0.527 0.54 28.66 72.34 101 50
15 Reading 0.604 0.598 0.558 0.49 25.81 66.67 92.47 44
22 Rotherham 0.474 0.458 0.449 0.4 29.08 64.74 93.82 36
6 Sheff Wed 0.507 0.508 0.538 0.57 32.17 71.54 103.71 59
12 Wolves 0.452 0.449 0.436 0.45 31.78 70.06 101.84 47

Tuesday, 8 March 2016

Championship Week 35: Why Blackburn and Reading's good numbers are hiding fatal flaws

So there’s been a bit of tightening up around the Championship table over the last few weeks.

Pos Team Played Won Drawn Lost GF GA GD Points
1 Burnley 35 19 11 5 54 27 27 68
2 Middlesbrough 34 20 7 7 47 20 27 67
3 Hull 34 19 7 8 48 20 28 64
4 Brighton 35 17 13 5 49 34 15 64
5 Derby 35 16 12 7 47 30 17 60
6 Sheffield Wednesday 35 14 13 8 50 36 14 55
7 Cardiff 35 14 13 8 47 38 9 55
8 Birmingham 34 15 9 10 40 32 8 54
9 Ipswich 34 15 9 10 42 40 2 54
10 Preston 35 12 13 10 33 32 1 49
11 Reading 34 11 11 12 40 38 2 44
12 Queens Park Rangers 35 10 14 11 41 42 -1 44
13 Wolverhampton Wanderers 35 11 10 14 43 49 -6 43
14 Brentford 35 12 7 16 48 55 -7 43
15 Nottingham Forest 34 10 12 12 32 31 1 42
16 Leeds 34 9 14 11 32 40 -8 41
17 Blackburn 34 9 13 12 33 32 1 40
18 Huddersfield 35 10 9 16 44 48 -4 39
19 Fulham 35 8 13 14 53 58 -5 37
20 Bristol City 35 9 10 16 31 55 -24 37
21 Milton Keynes Dons 35 9 8 18 30 48 -18 35
22 Rotherham 35 9 5 21 39 59 -20 32
23 Charlton 35 6 10 19 32 66 -34 28
24 Bolton 35 4 13 18 33 58 -25 25

Looking at the “good-lucky” graphs there’s two teams which have had really strong shot numbers all season but yet find themselves in mid-table obscurity – Blackburn Rovers and Reading.

Indeed Blackburn are currently closer to the relegation zone than the playoffs, and have been in danger of being dragged into that dogfight for much of the season.
That risk was significant enough for Gary Bowyer to be given his P45 with Paul Lambert replacing him.

Reading, on the other hand, started the season on fire hovering around the automatic promotion spots – but then an awful run of form saw them drop into mid-table and earned Steve Clarke the sack.
After two games for interim manager Martin Kuhl, Brian McDermott arrived for his second spell in charge of the Royals.

But fortunes have not really improved and the club apparently made its intention clear to settle for obscurity this season by selling Neil Blackman to promotion chasing Derby.
Of course, Blackburn took a bit more persuading but eventually sold their leading man Jordan Rhodes to Middlesbrough for somewhere between £9m and £12m, depending what you read.

From the aforementioned chart we can see that below average PDO scores have had quite a big say in these two teams’ positions.
But how have they been generating such good shot metrics and is there any likelihood of a last minute push up the table.

For Reading, shot volume certainly isn’t a problem – 535 attempts on goal while only conceding 329 leads the Championship in both categories.
So surely it’s just a matter of time? Well, maybe not.

Breaking down where Reading are taking their shots from gives a much better insight into their poor league position.
Only Leeds United and QPR have taken fewer shots inside the danger zone than Reading (76).
In fact, just 14% of Reading’s shots come from the danger zone – the lowest proportion in the league. In contrast, the Royals have taken 55% of their shots from outside the 18 yard box – the highest in the league.

But was this a tactical strategy heavily leaned on by one of the managers. With Kuhl taking charge of just two games, I’m going to exclude those games and look just at how Clarke and McDermott have performed.

Well, the one really notable change offensively is a slightly more aerial approach under McDermott. His Reading team take one fewer shot per game outside the box than Clarke’s, but have replaced that with one more header from outside the six yard box - in effect, exchanging one low % effort for another.

Overall, when we compare the two managers, McDermott has seen a dip in all shot metrics
The only increase has been a significant jump in save percentage, which has boosted Reading’s PDO to 94, still well below a league average of 100.

11v11 Corsi 11v11 Fenwick 11v11 Sot 11v11 GF% 11v11 Sh% 11v11 sv% 11v11PDO
Clarke 0.64 0.64 0.62 0.54 0.28 0.63 0.90
Kuhl 0.39 0.34 0.30 0.25 0.17 0.79 0.95
McDermott 0.61 0.61 0.56 0.50 0.24 0.70 0.93

Defensively McDermott’s reign has not been a success either as the team are conceding more shots per game from every location except headers outside the six yard box – further evidence of McDermott’s greater emphasis on the aerial perhaps.

So despite these continued well above average shot shares, given the poor locations the vast majority of attempts are coming from, aside from a spell of 25 yard screamers hitting the top corner, I see little hope for Reading to push towards the playoffs.

For Blackburn, who also have very good overall shot share numbers, the problem is that they too are one-dimensional in the wrong respect.
Overall Rovers have a league average attack of 438 (12th) attempts and a pretty stout defence, allowing just 385 attempts (4th best). However they rely on an aerial attack which again produces efforts on goal with a lower likelihood of scoring.

Blackburn have taken the fourth most headed goal attempts within the six yard box (17) and by far the most headed attempts outside the six yard box 117 – Bolton are second in this category with 96 while the league average is just 63.
The aerial strength is also reflected defensively. While they have conceded 10 headers from close range (9th best) they have allowed just 36 from further out – again the best mark.

All this means they have comfortably the best headed attempt difference in the league at +88. (Reading are second at +44.)
Unfortunately the aerial attack is taking just too much priority. Almost one in three (31%) of all Blackburn’s goal efforts are via the head – again way clear of 2nd place (Bolton, 24%).

Lambert joined Blackburn with a reputation rather in tatters after a calamitous spell at Aston Villa and on the surface this isn’t helping as his 11v11 shot share metrics are mostly lower than Bowyer.

11v11 Corsi 11v11 Fenwick 11v11 Sot 11v11 GF% 11v11 Sh% 11v11 sv% 11v11PDO
Gary Bowyer 0.53 0.57 0.58 0.56 0.22 0.76 0.98
Paul Lambert 0.54 0.55 0.53 0.48 0.24 0.71 0.95

However, when we look at it in game state, Lambert’s team is better at driving play when tied than his predecessor, but other game states, especially when leading, are far more mixed.
Ultimately though, it is the continued reliance on an aerial attack that will scupper any hopes Blackburn harboured of a playoff push.

Finally, here's the 11v11 data table to play around with yourselves:

Team 11v11 Corsi 11v11 Fenwick 11v11 SoT 11v11 Goals Rate 11v11 Sh% 11v11 Sv% 11v11 PDO
Birmingham 0.47 0.442 0.464 0.55 28.13 79.73 107.85
Blackburn 0.538 0.562 0.554 0.52 23.13 73.15 96.28
Bolton 0.49 0.493 0.439 0.37 24.62 67.47 92.09
Brentford 0.487 0.477 0.466 0.47 28.86 71.35 100.2
Brighton 0.524 0.529 0.522 0.59 33.33 75.19 108.53
Bristol City 0.494 0.523 0.511 0.38 23.13 60.94 84.07
Burnley 0.453 0.466 0.518 0.66 35.62 80.15 115.76
Cardiff 0.489 0.481 0.509 0.56 31.08 74.83 105.91
Charlton 0.403 0.407 0.45 0.34 22.7 63.95 86.65
Derby 0.536 0.529 0.557 0.58 29.93 72.65 102.58
Fulham 0.457 0.449 0.466 0.5 34.44 69.94 104.38
Huddersfield 0.547 0.546 0.511 0.47 29.17 65.22 94.38
Hull 0.591 0.61 0.63 0.74 29.94 81.63 111.57
Ipswich 0.506 0.514 0.528 0.53 28.19 71.43 99.62
Leeds 0.481 0.451 0.415 0.43 29.13 72.41 101.54
Middlesbrough 0.54 0.552 0.573 0.7 29.3 82.91 112.21
MK Dons 0.438 0.429 0.409 0.37 25.71 69.74 95.45
Nottingham Forest 0.511 0.519 0.513 0.47 18.84 77.86 96.7
Preston 0.497 0.506 0.5 0.52 28.07 73.68 101.75
QPR 0.517 0.519 0.516 0.51 27.78 71.11 98.89
Reading 0.612 0.605 0.572 0.51 25.83 67.26 93.08
Rotherham 0.478 0.462 0.45 0.39 27.61 64.63 92.25
Sheff Wed 0.503 0.507 0.53 0.56 32.84 70.59 103.42
Wolves 0.453 0.448 0.429 0.44 31.45 70.3 101.75