How much money would it take to bring Arsenal level with the big spenders?

All the great debates about sports take place in bars. Which is funny because bars are the only place where you can’t talk about politics or religion, and in many ways sports fans are more political than most political parties and more religious than many religions.

Like religion, sports have gods and devils, saints and demons, and other assorted dualistic expressions of the culture. The gods of football are folks like Collina, who we believe is the infallible referee whom every player loved. The devils of football are just as broadly described, they are the evil money men who have come to till our perfect pastures with their giant corporate land rapers. Our players are our saviors and our goats; sometimes both in a single game. And among the punditry, which is the priest class if you will, there are dogmatists and iconoclasts. Gods and devils, saints and demons, priests and heretics, we’ve got them all.

But these dualisms are only good for fantasy and for painting broad strokes, they don’t serve us very well when dealing with reality. Which is why we like dualisms; reality is messy and hard to explain.

Some folks, though, want to explain reality.

In their groundbreaking work, Soccernomics, Stefan Szymanski and Simon Kuper did exactly that. They reached into the darkness of English football finance and using huge reams of data showed that, in fact, money can buy you love, at least in the form of a league title. Specifically, Kuper and Szymanski showed that there is a correlation between wages and league position or as he put it in his column on Forbes after Manchester City won the League:   “Money buys success- it’s not a very exciting observation, but one that I have always believed to be the fundamental truth about football. There’s a very clear correlation between wage spending by football teams and league position (read Soccernomics if you want to see some of the evidence).”

Kuper and Szymanski’s idea that wages correlate with league position has actually become dogmatic in its own way. People not only believe that wages correlate with league position but that wages should determine, or come close to determining, league position. Once this idea passed into dogma the Arsenal priesthood then used it as a stick to beat Arsene Wenger with; Arsenal have the fifth highest wage bill as of the 2010/11 season, therefore the club should be more competitive. Or at the very least, simply overhauling the wage structure will create a more competitive team. After all, the argument goes, Arsenal’s wage bill is only £30m less than Man U and they are in the title hunt every season.

Another dogmatic belief among Arsenal supporters and pundits is that Arsenal are perennially just one player short of the squad needed to truly compete for the title. “If only Arsenal would buy that (fill in the position) that they have needed for (fill in the years) then they could win the title” is a refrain that I’ve heard so often that it’s become tinny.

And then there’s the last, and probably most pervasive of all Arsenal’s dogmatic beliefs, that Arsenal don’t buy titles. That Arsenal never were a big spending team and that they didn’t buy players, they made them. It’s a slogan that’s so ingrained in the Arsenal collective unconscious that there’s even a tee shirt for sale in the team shop with a quote from Arsene Wenger “We don’t buy superstars, we make them.”

I know a guy named Zach Slaton who writes for various publications including where Stefan Szymanski is also a contributor. Zach is probably a genius. He’s also a Gooner. And like all really bright people can’t read a book like Soccernomics without challenging some of their conclusions.

For example, when I read this quote from Soccernomics, I simply accepted it as fact:

In fact, the amount that almost any club spends on transfer fees bears little relation to where it finishes in the league. We studied the spending of forty English clubs between 1978 and 1997, and found that their outlay on transfers explained only 16 percent of their total variation in league position. By contrast, their spending on salaries explained a massive 92 percent of that variation. In the 1998-2007 period, spending on salaries by clubs in the Premier League and the Championship… still explained 89 percent of the variation in league position. It seems that high wages help a club much more than do spectacular transfers.

But when Zach read that same quote he knew that it couldn’t be right and he set about proving that it wasn’t.

His first bombshell post on the topic was provocatively titled “Soccernomics Was Wrong: Why Transfer Expenditures Matter, and How They Can Predict Table Position”. In that article, Zach pieced together data that showed Premier League teams were using fewer and fewer academy players (down from 24% in 1992 to just 8% in 2010) which meant that they were more reliant on transfers to fill the slots on the field. Then, using the inflation adjusted transfer fees from the book and web site Pay as You Play he created a model which accounted for every team’s transfers in the Premier League era, and using the average of each year’s dataset showed that team success was correlated to how much their players were worth in terms of transfer fees paid. Teams worth the average amount finished mid table, teams that were worth higher multiples (2, 3, or even 4 times the average) finished higher up the table.

Zach was right enough that as soon as he published the article, Stefan Szymanski contacted him as you can see in the post script if you read the article linked above. And not only that, but the latest version of Soccernomics contains a section addressing many of the points made in Zach’s article. They may publicly disagree still, but the work that Zach put out there is recognized as significant .

Simply getting the attention of one of the authors of Socernommics via a provocatively titled blog post wasn’t the end of this experimentation with Premier League financial data. Working  together with Paul Tomkins and Graeme Riley, Zach wanted a comprehensive model, something that encompassed both transfers and wages and was adjusted for the inflation caused by teams like Man City and Chelsea splurging on huge transfer fees and giant pay packets.  So, he asked Stefan Szymanski if he could use his database on wage data built from public filings with the United Kingdom’s Companies House (think SEC filings for publicly traded companies in the US). Szymanski agreed, and combined with the inflation adjusted transfer data from Pay as You Play, the team created a new model called TTV or “Total Team Value” and its cousin mTTV or the “multiple of the Total Team Value”. Don’t get too scared by these acronyms. The multiple is just that: a multiple of a Team’s Total Value compared to the median team value. If Everton are worth £200m in TTV and are an average team and Chelsea are worth £800m in TTV then Chelsea have an mTTV of 4. Pretty simple, right?

The amount of data that went into the TTV model and, most importantly, the quality of that data is mind boggling. This isn’t just some dude on the internet speculating on player wages or pulling transfer numbers off Wikipedia, this is professional work compiled with academic rigor. And using that data and analysis we can actually tear down some long held beliefs.

Using the TTV model, Zach created this chart of Total Team Values for the first ten seasons of the Premier League. To make the data work within Excel’s line graph function, Zach has coded each Premier League season per the year in which the second half of the season occurs (the inaugural 1992/93 season is coded as “1993”, the second as “1994”, and so on). As you can plainly see, in the early years of the Wenger era Arsenal were one of the top spending teams in the Premier League, ending that decade in nearly a dead heat with Man U with a total team value just over £300m in today’s Pounds (remember, adjusted for Premier League inflation). You can also see Chelsea’s early ambitions mapped out as well, doubling in value from a below average value team in 1997 to a team with a TTV multiple of 2 by 2002. This correlates with the years where they were competing with big clubs like Arsenal for trophies, with the Blues narrowly missing out on an FA Cup in 2002 when Freddy Ljungberg glided past a young John Terry as he made one of his trademark stumbles.

The point is that Arsenal were a buying club in that era. They held the British transfer record when they bought Dennis Bergkamp for £7.5m and they also had the highest paid player in the form of Sol Campbell who was on £100,000 a week after leaving Tottenham. It’s not that Arsenal were exactly like Man City and Chelsea – the club got good deals, spent wisely, and sold for a profit – but the fact remains that Arsenal were the second most expensive team in the land.

The other point here is correlation:  the more that a team spends on players, the greater their chance of winning the League and other trophies and the greater the difference in their spending than the rest of the League, the greater that chance of winning the League grows. The investment that Arsene Wenger made in buying ready-made players and greatly increasing those players’ salary in those first few years at the club correlate to the club competing for and ultimately winning trophies.

I once had a heated argument in a bar over this exact topic. I was a new Gooner and by 2002 the mythology that Arsenal were not a buying club was already one of my well-worn arguments. He was an Everton fan and when I told him that I was an Arsenal supporter he said “Great, the Yankees of Football.” This was heresy as far as I was concerned. The Arsenal I knew and loved didn’t buy trophies! We argued for hours that night, with him bringing up transfers in and me bringing up transfers out. In the end, we simply disagreed and I walked away believing that he was an idiot and holding on to the belief that Arsenal were never a big spending club.

A belief I held until I saw Zach’s data. When you look at Total Team Valuation, Arsenal were undeniably one of the top spending teams during that early Wenger era. Yes, the club got players like Henry and Campbell on the relative cheap, but they spent money – spending that was on par with Man U. Arsenal might not have been the Yankees of football (Man U was) but neither were we the Oakland Athletics as I had been led to believe. Arsenal were the Red Sox.

And then Abramovich happened.

We often talk about how much money Abramovich spent on players since taking over at Chelsea, but the full extent of his rapacious purchasing of players is often lost in the abstract. Instead we need to step back, compare his spending to other teams and to see it in inflation adjusted money, an inflation that he caused in that mad rush to build a team from virtually nothing.

At a time when Arsenal and Man U were leading the League with a TTV around £300m, Chelsea came in and blew the doors off everyone. In one year, they doubled in value. The next year, their TTV was more than Arsenal and Liverpool combined.

It’s important to remember that Total Team Value is basically a measure of how much money it would cost for an owner to put together that exact team and pay them their wages for that one season. Just the team, mind you. By 2007 Abramovich’s Chelsea would have cost £800m while Arsenal was declining, and by the next season would drop below the £300m for the first time since 2002.

So if the model is so good and Chelsea so far above everyone else in terms of TTV, why didn’t they win every year? One answer is the manager. Arsenal and Man U have been blessed with managers who add hugely to the value of the team. According to another one of Zach’s models, Arsene Wenger has averaged 16.8 points per season above what the TTV model suggests that he should and believe it or not, Fergie is actually better at 18.2 points per season above his model.

The reason Abramovich and now Mansour need to spend so heavily is that because among  the Premier League elite teams the competition for players is so fierce and the differences between them so small that just spending money isn’t enough to ensure success. To compete against managers like Wenger and Fergie, teams have to spend loads of money and also have to have a good manager — as Chelsea comically found out this season when Andres Villas-Boas dragged them down into 6th place.*

What is worrying, though, from an Arsenal fan perspective is that the Gunners have gone from the second biggest net spend team in the League in 2002 to the 6th biggest net spend team in the League in 2012. This is dangerous for the club because our fans still believe that we are a team that should be competing for trophies and yet, statistically, when you look at the history of teams with a similar mTTV to where Arsenal is spending right now, Arsenal actually have a 75% chance of finishing out of the top four. It’s only Arsene Wenger’s remarkable ability to outperform that explains the reason why the club has had Champions League football every season under his tenure. Which also explains why Wenger is basically untouchable as far as the board are concerned: he is the human equivalent of adding £100m in players according to this model.

Which brings me to what will almost certainly be the most controversial section of this article: how much does Arsenal need to spend to compete with the top three TTV teams?

Look, it’s a model. Predicting the future with statistics is very dangerous and some say foolish. Not only that, but Chelsea couldn’t literally field £800m in piles of £100 notes and expect to win the title. You have to purchase smartly, you have to have a good manger, and you have to have a lot of luck. The model also doesn’t require that you spend the exact same amount as Chelsea and Man City. In fact, to achieve a relatively high level of competition a team like Arsenal could invest relatively modestly and be competitive. Arsenal don’t need an £800m team to compete.

You also have to remember that unlike in 2002, the Premier League is not a duopoly of Arsenal and Man U: there are three teams at the top of the spending table who all have around a 95% chance of winning the League. Obviously, they can’t all win, so that means that in a normal year they will occupy the top three spots for themselves, leaving the remaining teams fighting it out for 4th or lower.

But still, I was curious, so I asked the question of Zach. What would it take for Arsenal to have a similar percent chance (95%) of winning the League that the three big spenders (Chelsea, Man U, and Man City) currently enjoy?

The quick answer was £215m. That’s the amount of money it would take in transfers and salaries to get into that 95th percentile. That’s if Arsene Wenger didn’t add anything to the team, which we know he does – 16 points per season over his career which has dipped to 14 points lately, which is around £100m in value. That means that if Arsenal wanted to get into the 95th percent, in the Total Team Value model, with Arsene Wenger at the helm, they would have to spend somewhere around £120m.

Zach’s team have created a compelling model, and if the model is correct, then the amount of investment required to get Arsenal back to the top of the League is not simply just to buy one or two players but upgrades on a lot of players. It has to be a sum of quality players that totals £120m worth of upgrades.  There are no devils or saints behind that number, just vetted statistical models.  Where the club finds such money will certainly involve some discussions that will go well beyond the heated rhetoric of political or religious debate, and will require a few saints and many more devils if Arsenal is to build itself into a championship contending team again.

I can already see how the bar argument over what I have presented here will be divided: is a one off investment of £120m “sustainable”? The TTV model doesn’t answer that. This is just about dealing with the reality that Arsenal were a spending club and that in terms of Total Team Value, the club has stagnated since the last title while the other three title-challenging clubs have exploded in value. Is any of this sustainable? Buy me a beer some time and we’ll chat. But if you want to know what I think it would take for Arsenal to be competitive in next year’s title race, truly competitive, I believe in Zach’s model.



*I can’t wait to see what he does to Tottenham.

P.S. Zach Slaton not only contributed to this post in creating the model but also in editing the final article. Follow Zach on twitter @the_number_game and his blog of Forbes.

86 thoughts on “How much money would it take to bring Arsenal level with the big spenders?

  1. Vote -1 Vote +1Cromulus

    Hey Joss,

    Isn’t it odd that your smallest suggestion (that of giving people raises) threatens the integrity of the wage based argument? Imagine, purely as a thought exercise, if last year Wolves simply raised the salaries of everyone on their staff to match that of Shitty’s or Chel&ki’s. They would have had a 95% chance of winning the league :)

  2. Vote -1 Vote +1lari03

    Nice article, just shared it for some other friends to take a look, would love to see more additions to this,now we know how squad management works or how being a soccer manager feels. makes me wish I could do this, lol.

  3. +1 Vote -1 Vote +1Kit

    Hi Tim,

    It was an interesting read but I think the model is a crock and here’s why.

    From the article:
    “It’s important to remember that Total Team Value is basically a measure of how much money it would cost for an owner to put together that exact team and pay them their wages for that one season.”

    And how does the model inventor calculate the transfer value of the players?

    From his website:
    “CTPP = Current Transfer Purchase Price. The player’s original transfer value adjusted for football inflation using the same method – but not the same figures – as the Retail Price Index. (Andriy Shevchenko holds the highest CTPP, his £30.8m transfer in July 2006 now worth £68.2m; the increase down to the difference in the average cost of a Premier League transfer that season – around £2m – and how the average in 2011 is c.£5m)”

    But obviously each player’s transfer value is going to ‘inflate’ at different rates! Possibly even deflate. In the example he cites Shevchenko has a CTPP of £68m — does that make any sense??

    Now let’s look at his calculated CTPP’s for some Arsenal players:

    Walcott £27.4M
    Rosicky £14.8M
    Ramsey £4.9M
    Arshavin £14.6M
    Wilshere (£0 as a trainee)

    I hope I’m not the only one to see something wrong with these CTTP valuations…

    Sure, you can argue that there is an overall transfer price inflation in the market, but by bluntly applying the aggregate indiscriminately, you’ve lost all the meaningful information. The actual market transfer value of a young player will rise as if he meets his potential, fall if he fails, will generally decline as he reaches the end of his contact, and will inexorably be pulled down to zero towards the end of his playing career.

    You alluded to the same problem with the wage valuations in the comments section (the £60,000/week for Denilson is no longer a market rate). I just feel the same problem exists for the whole model.

  4. -2 Vote -1 Vote +1richie

    @Kit the idea’s behind this article are great, indeed the article is great, and the reasoning behind much of it is sound. There is a logic to it, but I’m with you 100% the maths are flawed! I actually think the Zach model is worth saving and I think it could/would work but Zach needs to write his programme figuring in what you could call the Shevchenco syndrome. If a player is sold for £30mil his stock value will only continue to rise if he continues to do the buisness, and will fall as long as he doesn’t. Andre may have improved with age up and until his last big transfer, thereafter with age he deteriorated and so therefore did his value, exponentially.

  5. +8 Vote -1 Vote +1Zach Slaton

    Kit -
    As someone who works in the TPI database, let me explain why it’s not “a crock”.

    Players are re-assigned values when they’re sold. Until then, there isn’t much of a way to re-assess value in between those sales. Tracking player value in between the sales on an individual basis is time prohibitive. This model is about as good as you can get outside of TransferMarkt.

    I bring up that specific resource because it turns out Stefan Szymanski is familiar with the TransferMarkt process for setting player transfer values. They do what you suggest. They sit around and have a number of qualified individuals debate the value they think should be assigned to a player. The group then decides on the value, and that’s what’s assigned to the player in between individual transfer fees. Stefan’s done a preliminary comparison at the club level, and it turns out that the TransferMarkt club totals are very close to those for the TPI. Thus, for the purposes of measuring TTV contribution to league position, the TPI data does a very good job even without having annual adjustments at the individual player level based upon form.

    I am perfectly willing to hear criticisms of the model and examine ways to improve it. As George Box said, “all models are wrong, some just happen to be useful.” The TPI’s models have demonstrated their usefulness. I understand your critique, and it likely accounts for part of the 27% in finish position not explained by TTV variation. The other 73% in variation is explained by TTV variation. The model passes all statistical tests for correlation and linear regression robustness. It sounds a bit better than “a krock” to me.

Comments are closed.