The Stat Man: Assessing Lee Johnson and the crucial season rescue attempt at Sunderland

You needed to scroll quite far down the list of favorites with the bookmakers before you reached the name of Lee Johnson. Even two nights before his appointment, he was about the fifth favourite; and while many were wondering whether the white smoke would appear for Cowley, Cook, or perhaps even for a last-minute change of heart from Gus Poyet, Johnson was quietly going about beating them all to what is currently the biggest opportunity in British football.

Aside from his excellent performance in an interview (something we have all noticed from his media appearances), what might have prompted Jim Rodwell and Kristjaan Speakman to opt for the former Barnsley and Bristol City boss?

Clearly, there is much “work to do” (to use the Sunderland AFC Twitter platitude that has had far too many outings in recent seasons). The construction of a squad designed to implement Phil Parkinson’s slow, methodical and cautious style of play is going to take time to convert into something more dynamic, exciting, and - most importantly - automatic promotion-worthy.

What specific targets should Johnson and Speakman set to transform the 2020-21 season from being one in which Sunderland seem destined to be in the chasing pack behind the play off places, into one in which they are looking back over their shoulders at everyone else?

Join me as I take a high-level look at both of these questions.

---

Why Lee Johnson?

Of course, I don’t know exactly what it was about Lee Johnson that won him the Sunderland job. But in the new data-driven operation that rumour suggests we are about to become, I’d be surprised if some of the following wasn’t considered.

Squad quality conversion

Head coaches and managers have squads at their disposal with varying degrees of quality. It is their task to at least deliver the expected level of performance on the pitch in accordance with the quality available to them. Better head coaches and managers will exceed expectations.

The table below shows the current League One sides ranked by squad quality.

What this reveals, is that the minimum expectation of the current Sunderland squad is that they occupy a mean league position of second. There would be a little movement up and down a place or two, but the expected position would be second place.

To provide something to contrast against, I assessed Phil Parkinson’s career in the period from the beginning of the 2016-17 season to the current day to generate a ‘quality conversion’ score. That is, a number that describes his ability to convert squad quality into attacking threat. The result was -0.17.

The number being negative means that attacking performance is below what should be expected given the players available. Phil Parkinson’s Sunderland was an 8th to 10th placed League One side - which seems about right, based on his result of -0.17.

The chart below shows Lee Johnson’s result alongside Phil Parkinson’s.

This suggests that, if Johnson were to continue the pattern of previous success, the performance of his Sunderland side would exceed the expectation derived from squad quality. i.e., an expected league finish of at least second place and most likely higher.

Improvement rate

To rescue the 2020/21 automatic promotion campaign, rapid improvement is needed. Improvement can come through recruitment, youth development, and tactical astuteness. Those factors are not addressed separately here - only overall improvement is discussed.

The chart below provides a visualisation of the rate at which Lee Johnson was able to improve his sides compared, again, to Phil Parkinson.

To be fair to Phil Parkinson, he was improving Sunderland, but much too slowly. We certainly couldn’t have afforded to give him the time it seemed he needed to create an effective automatic promotion machine.

Game management

An inability to see out a game was something for which Phil Parkinson was heavily criticised in his last few weeks in the Sunderland job. But, as we will see, this is generally a part of the game in which he is strong.

I measured ‘game management’ using the proportion of occasions on which each of the following occured:

• A lead was extended.
• A lead was maintained.
• Came from behind to draw.
• Came from behind to win.

The results are presented in the next table.

Phil Parkinson’s numbers are that of a cautious manager. 72% of leads are held, but victory results from a losing position on only 3% of occasions. Going ahead against a Phil Parkinson side is close to a guarantee that you’re not going to lose the game.

Lee Johnson’s results, on the other hand, suggest a much less cautious approach. Only 64% of leads are held for the full three points, but conversely, 12% of games in which his side goes behind, are won. Remember ‘Bristanbul’? That kind of game will be more common with Johnson at the helm, in both directions.

It won’t be boring!

Blooding youth into first team action

Another criticism levelled at Phil Parkinson was his over-reliance on aged and experienced players, when many thought there were better, younger, and more energetic youth options available. All too often, a young player would impress in a midweek cup game and the fans would be left bemused as they didn’t even make the bench for the next league game.

I processed every minute played by every player who has been managed by Parkinson and Johnson in the last five years. Those minutes were then categorised as either being played by someone aged under, or over 23 (‘youth’ or ‘developed’, respectively). The results are presented in the chart below.

It seems that the young players have good reason to be excited about the arrival at the club of Lee Johnson and Kristjaan Speakman!

There are many attributes on which the previous performance of a manager or head coach can be assessed. But just from the four mini-assessments above, I have to wonder whether the job would have been given to Parkinson had a similar exercise been carried out in October 2019...

---

How to rescue the 2020-21 season

It became clear in recent weeks that Phil Parkinson’s overly cautious approach, which was built on keeping clean sheets and scoring a goal per game, wasn’t going to result in automatic promotion. In fact, even the most optimistic models were predicting a finish in the lower play off places.

As I type, Sunderland sit in ninth place in the table, a full ten points behind the leading side, Hull, who have played the same number of games. Consider also that a third of the season is gone, and we’re looking at a mighty difficult situation. Welcome to Sunderland, Lee!

One of the best predictors of automatic promotion that I have seen is ‘goal quotient’. I have written about this before - it sounds fancy but it isn’t, it’s simply goals for divided by goals against. I obtained 30 seasons of data and attributed successes (automatic promotion) and failures (any other league position) to each side involved and calculated their goal quotient.

From this, I fitted a logistic curve that explains the probability of automatic promotion given goal quotient. The curve is shown below.

What this illustrates, is that sides with a goal quotient less than 1.75 have a less than 50% chance of automatic promotion, and sides with goal quotient greater than 1.75, have a greater than 50% chance of automatic promotion.

It also suggests that when the goal quotient goes below 1.50, the probability of automatic promotion drops quickly below 10%. But with a goal quotient greater than 2.00, the probability goes above 90%.

With that in mind, consider that Sunderland’s current goal quotient is 18/13 = 1.38, which gives us a current automatic promotion probability of 0.06. That’s a 6% chance of automatic promotion. This result - along with that of several other sides in with a realistic shot at promotion - is illustrated below.

This does not paint a pretty picture for Sunderland fans. On the other hand, Hull, Lincoln, and Portsmouth fans have reason to be optimistic. I’m thrilled for them...

To mount a genuine push for automatic promotion, Sunderland are going to need to score more goals. Significantly more goals. The next plot shows how the probability of promotion will change in two scenarios.

• Scenario 1: Sunderland score an average of two goals per game for the remainder of the season but, as a consequence of being more attacking, they concede an increased average of one goal per game.
• Scenario 2: Sunderland score an average of two goals per game for the remainder of the season and retain the current goals against rate of 0.87 per game.

In Scenario 1, Sunderland end the season with an automatic promotion probability of 0.61. That is much better than the current 0.06, but still not much more than a 50% chance.

In Scenario 2, Sunderland would end the season with an automatic promotion probability of 0.86. We would be very unlucky to miss out on promotion in that circumstance.

Without doubt, Sunderland needs to become a much more attacking side that gets close to six shots on target per game. Whether that can be achieved, without sacrificing too much solidity at the back, remains to be seen.

In contrast to footballing life under Phil Parkinson, it’s going to be anything but dull. Hold on to your hats.

---