Defenders are the least glamorous position in fantasy Premier League games (apart from goalkeepers of course), but they are an integral part of any successful team. Depending on the formation deployed, they can contribute between three to five players in the first XI. Supposing a realistic score of 8 ppm, then that is a theoretical range of 24-40 points every week. Defenders are by far the deepest position in terms of quality; you can likely find suitable pickups in free agency. Eight or ten team leagues could even stream one or two defenders according to matchups if you like to live more dangerously.

That being said, it is still worth taking a slightly deeper dive into this position to try and identify which stats are more lucrative than others. For example, are aerials more profitable than interceptions? Should you outright ignore successful tackles? (Spoiler alert: yes, you should).

For the purpose of this analysis, a ‘defending’ score was estimated using WhoScored 2016/17 data and was based on how many minutes each player played. The stats that went into the defending score are: Aerials, Tackles, Interceptions, and Effective Clearances. Similarly, an ‘attacking’ score was constructed using these stats: Shots per Game, Dribbles, Key Passes, Crosses, and Dispossession.

Once the defending and attacking data has been analysed, a total score will be built which includes the two scores above plus goals, assists and clean sheets. Goals against have been omitted due to the fact that it is very complicated to ascertain who was on the pitch when goals were conceded.

By using multiple regression analysis, it is possible to try and predict which stats will likely contribute more to the total defending or attacking scores. A pool of 96 defenders from all of the current Premier League teams was used for this purpose.

A quick look at the data indicated that there were outliers that were primarily from the promoted teams; Brighton, Newcastle, and Huddersfield. Since their stats are from the Championship and are not completely compatible with Premier League performance, they were excluded from the analysis going forward.

Defending Score

The first review indicated something that is probably quite surprising; the number of successful tackles does not really affect the total defending score. The flat trendline shown in the graph below shows that as tackles increase, the total defending score only increases slightly. The effective clearances numbers were divided by four since in Togga an effective clearance is only worth 0.25 points.

Without getting into the nitty-gritty of multiple regression analysis, it was found that the effective clearances stat was statistically insignificant. Even though the trendline is positively correlated (i.e. more clearances = higher defending score), it did not add any additional information to improve the prediction of the total defending score and therefore was eliminated.

The model did show that the number of aerials had the biggest impact on the total defending score. In fact, of the top 20 defenders in the analysis, 18 out of 20 are centre-backs. Therefore, it is safe to say that centre-backs are key to acquiring ‘defending’ points. Also, it is worth noting that the defending score is not biased towards the top teams. In the top 20 players shown below, a total of 12 different teams are represented.

C. Azpilicueta B. Mee R. Shawcross G. McAuley
N. Otamendi M. Keane N. Monreal J. Fonte
J. Vertonghen G. Cahill S. Mustafi K. Walker
L. Koscielny V. Lindelof D. Luiz V. Dijk
S. Cook A. Williams D. Lovren C. Dawson

Attacking Score

Moving on to attacking points. The assumption here would be that full-backs will dominate this score if centrebacks dominate the defending score. After a preliminary look at the statistics, the dispossession stat was excluded due to it being insignificant in trying to maximize the attacking score. This is expected since they contribute negatively to the attacking score.

The good news is that all of the other variables affect the total attacking score positively and are all significant. However, there isn’t one variable that stands out as more significant than another. The crossing stat came out highest in the analysis but it was very close. The table below shows the top 20 defenders in terms of attacking score.

J. Milner* L. Baines* C. Daniels* D. Rose*
P. Aanholt* C. Soares* A. Smith R. Bertrand
J. Holebas* C. Fuchs* B. Mendy* H. Bellerin
N. Clyne* K. Walker M. Lowton* K. Naughton*
A. Valencia* V. Moses M. Alonso S. Coleman*

* in the top 20 for successful crosses.

As expected, all of the top 20 defenders for the attacking score are fullbacks. Even without the goals and assists points, James Milner comes out on top. Again, interestingly there seems to be no bias towards the top clubs with 14 different clubs represented.

A closer look at the crossing stat shows us that of the top 20 attacking defenders, 14 of them appear in the top 20 for crossing as well. This perhaps highlights the significance of crossing in attaining a larger attacking score.

One other interesting facet of the analysis was the dribbling stat. It was found that a dribbling value of 1.1 per game would most likely give you a larger attacking score. Obviously it would be very hard to predict who only successfully dribbles about once per game. The key takeaway would be that dribbles are important, but more doesn’t necessarily mean better.

Total Score

Finally, if we model the defending and attacking scores along with the goals,  assists, and clean sheet scores to try and maximize the total score, we find that the defending score contributes by far the most amount of information.

The top 20 defenders in the data set are outlined below. Milner has been classed as a defender in the analysis. That means he scored all of the clean sheet bonus points etc. that he missed out on in the Togga system last season making him the top scoring defender. There are nine full-backs and eleven centre-backs from a total of 11 different teams which highlights again that there seems to be no bias towards the top clubs.

J. Milner S. Cook S. Kolasinac M. Keane
C. Azpilicueta A. Valencia G. McAuley L. Koscielny
M. Alonso G. Cahill M. Lowton N. Clyne
K. Walker J. Vertonghen C. Dawson S. Mustafi
C. Fuchs N. Otamendi L. Baines B. Mee

It is interesting to note that the top 5 defenders are full-backs (Cesar Azpilicueta was a full-back at times for Chelsea last season before their switch to 5-3-2 so he is defined as both a full-back and centre-back in this article). Milner is also somewhat of an anomaly. His goal return (7) went unmatched and he was amongst the highest scoring in terms of key passes and successful crosses which appears to be the perfect recipe. If Milner was the starting left-back for Liverpool this season, he would be a top pick for highest scoring defender so long as he also kept PK duties. But to everybody’s annoyance, Jurgen Klopp has decided to put him back in mid-field.

Conclusion

To sum up, there is a definite emphasis on the defending score portion in trying to predict a high scoring defender. Aerials in particular appear to be one of the most crucial stats in predicting a higher score. Whilst dribbles are important, more of them do not appear to be better.  Goals and assists do not weigh as heavily as one might imagine and should be treated as more of a bonus than anything else. And playing for teams near the top of the table neither helps nor hinders your chances of having a high scoring defender with the exception of the most elite full-backs/wing-backs (but not James Milner anymore).