Many a time during his reign the debate has been on the lips of many West Ham fans over who should and shouldn’t be played under Sam Allardyce, with a particular focus on the strikers this season.
It is no secret that Allardyce has a preference to playing Carroll and Nolan together and in its prime, the partnership has shown to be a very effective combination. However, with Nolan’s arguably weaker performances this season and the ever-growing frustration amongst some fans, with the style of football at Upton Park, there appears to be more and more demand for the partnership to come to a halt and for Allardyce to play others instead.
There seems to be a bit of debate flying around at the moment over the reliability of statistics in football, a debate which has actually lead to the sacking of Mark Warburton recently at Brentford. It is entirely up to you on how you interpret these results, take them with a pinch of salt or take them as fact that Nolan needs to go, it was just something I wanted to investigate.
My Recent Analysis
For those who haven’t seen it, I recently produced some data infographics investigating the effect of players on the team’s results, measured by their effect on the average number of points per game (PPG). To save you from reading the article again, I’ll summarise for you. It found that when Carroll (-0.14 PPG) and Nolan (-0.29 PPG) play, we average a lower PPG than when they don’t. No where near as bad as the effect Cole (-0.99 PPG) and O’Brien (-1.23 PPG) have, the two worst results that I found, but still a negative effect nonetheless.
I wanted to follow this analysis up with something a little more in-depth and that would take into account relationships between players. Given all the heated arguments I hear about who should and shouldn’t be played with whom at West Ham, I think it’s quite fitting. It also needed to be something that would equally provide a single value for each player for easy comparison.
Friends With Benefits
The analysis I proceeded to conduct, I call “Friends with Benefits”. Click on the images below to expand.
It compares the average performance rating (*) for all of the games that a player has started, against their average performance rating for the games when they have started alongside a given player. An overall “player effectiveness” is then calculated as the average effect they have on all players in the current squad.
In a similar fashion to my last analysis it showed that Joey O’Brien doesn’t seem to do anybody any favours and in fact seems to trigger a weaker performance out of all of the other defenders, resulting in a -0.251 player effectiveness. Tomkins, Noble and Valencia actually lead the front line in terms of lifting others around them with an effectiveness of +0.062, +0.047 and +0.045 respectively.
Looking at the defenders in a little more detail, it was very interesting to see that James Tomkins appears to have a very positive effect on Andy Carroll’s performance (+0.407); one can only assume that this is related to the accuracy of Tomkins’ long balls forward. Similarly, he has a positive effect on most of the other defenders too, which are detailed in his graphic. His overall “player effectiveness” however isn’t anything huge (+0.062) but it is nonetheless positive.
Collins (-0.062) actually has the reverse effectiveness of James Tomkins, and interestingly has a very negative effect on Carroll’s performance ( -0.380), but a positive impact on Sakho (+0.145) and Valencia (+0.151).
I haven’t produced any other graphics for the remaining defenders as, well, they were a little boring with no stand out results between them, so moving on…
I’ve only collected together two infographics here for the midfielders, Nolan and Noble. The others didn’t really show anything too fascinating, so I won’t bore you with the details. Noble and Nolan are interesting characters to compare. Our club captain versus, arguably, the player that should be our club captain. It is no secret that Noble is not the most talented, skilfully, on the pitch, but the Upton Park faithful generally adore players with the grit and determination to fight for a win over natural ability; find someone with both and you get a very popular player at the Boleyn.
Sakho, Valencia, Song and Downing – these are the top 4 players that Noble influences, and all in a positive way. Just because Noble himself may not be putting in a high rating, there is little arguing that having a positive effect on these 4 players is probably more significant and beneficial to the team.
Moving on to Nolan, it is no shock, I’m sure to anyone, that Carroll (+0.210) benefits by having him on the pitch with him. One thing that may shock people though is that Song (+0.241) also benefits quite a bit by having Nolan in the team. Possibly a false-positive result, but intriguing nonetheless.
I heard someone recently mention that playing Nolan and Noble together is a bad idea, the detailed reasons for which I don’t remember, but this data would suggest they have a case. Nolan actually benefits from having Noble on the pitch, with an increased average rating of +0.059, whereas Noble suffers with a decrease in his rating of +0.170. Summing these together we can calculate a “Partnership Effectiveness” of -0.111. For context, the score for the Noble and Song pairing is +0.214, with both players benefiting when the other is on the pitch.
This analysis was definitely the most interesting of the lot. Given the smaller number of strikers we have, I managed to put together the graphic above, which details the effectiveness for each partnership that has started upfront this season.
It certainly came as no surprise to me that the most effective strike partnership is Valencia and Sakho (+0.483), with Sakho (+0.383) getting the better deal out of the two. At the other end of the scale, the worst partnership we seem to have, according to the figures, is Sakho and Carroll (-0.924).
If I’m honest, this did surprise me a little, and I can’t really put my finger on exactly why but please feel free to offer thoughts on this as I’d be interested to hear.
It was also interesting, and actually no surprise, to see how Carroll and Valencia both perform with each other. Carroll getting the better end of the deal with an average increase of 0.290 in his player rating and Valencia experiencing an average decrease of 0.239.
So what does it all mean?
I’d like to think that the results here show the complexity of team selection in football, and generally any team sport. If we simply consider the “best” players in the team the strike partnership of choice should be Sakho and Carroll as they have the 2nd and 3rd highest average rating in the squad, behind Downing (7.15), at 7.13 and 7.10 respectively. However, when they play together, their performance suffers, so actually the “best” players are not always the way to go.
Partnerships and the effect one player has on another is vital in considering the effectiveness of your team, it sounds obvious I know, but it’s satisfying to see the numbers back it up. Yes, one quality player can change a game and win a team the match, and I have no doubts in my mind that Carroll is good enough to be that player.
Good, consistent performances require players to work together, something I hope that Allardyce begins to realise more, because it sure doesn’t seem like it with his some of his selection methods.
* The player ratings have been calculated as the average of scores given to the players by SkySports and WhoScored.com