The Declining Fortunes of David Moyes (An m£XIR analysis)

Three weeks ago I began taking requests for m£XIR analyses.  The first post based upon such requests was a timely one given the appointment of Alex McLeish as Aston Villa’s new manager, and explained why Villa fans and management must have a realistic outlook of a 7th or 8th place finish next season. Following the request of Dan Kennett, I now pivot to the opposite end of the managerial spectrum by analyzing the tenure of David Moyes at Everton, who is currently the third longest serving manager in the Premier League behind Arsene Wenger and Alex Ferguson.

The Data/Technical Explanations

As in all previous analyses using the m£XIR model, all transfer data is taken from the 2011 Transfer Price Index and is denoted via “2011 CTPP” (Current Transfer Purchase Price). Ordinal logistic regression models of the likelihood of match outcome (win, draw, loss) versus match venue (home/away) in relation to the ratio of the 2011 CTPP starting XI costs (m£XIR) were created for each season’s data, using all teams in each season.

Here’s how it works. We took the cost of each starting XI (adjusted for inflation) for both Everton and their opponents in every match from March 16th 2002 to the end of the 2011, and then, by taking into account the difference in value between the two teams, and the advantage/disadvantage of the fixture being home/away, we compare the actual result with the m£XIR model’s predicted outcome.

When these individual match outcomes are then summed over a season, one begins to understand how well a manger is doing cumulatively against match-by-match expectations. This cumulative residual – a greater or lesser number of points than was expected – is what is used to judged manager performance across a season. Does he do better or worse than expected?

The image a couple of paragraphs below summarizes this cumulative point difference in the form of the bar graph data at top, as well as other key statistics that help one understand the reason for various trends within the data. A season point-per-match (PPM) differential is also provided, which is simply the cumulative point difference at the end of the season divided by the number of matches (38 in each of Moyes’ seasons). Zero differential to the m£XIR model indicates a manger is performing on par with the expectations set by the match venue and his ratio of £XI to the opposition’s. Falling below the x-axis, or having a season PPM differential less than 0, indicates under performance versus expectations. Landing above the x-axiss, or having a season PPM differential greater than 0, indicates over performance versus expectations.

The second major set of data in the graphic summarizes Moyes’ £XI in every match, as well as averages and standard deviations for each season and his entire tenure. The gray bar running through the middle of the data is its interquartile range (IQR), which represents the bounds of the middle 50% of the data, and is based upon the full range of £XI values throughout Moyes’ tenure (Q1 = £44.4M, Q3 = £64.1M).  The match-by-match m£XIR data helps us understand any major shifts in resources available to Moyes that could impact expectations within the m£XIR model, or increased variability, as it was shown in the Villa post that increased variability correlates with increased failure to meet expectations. Finally, several of the final rows of data provide summaries of each match outcome (win = black bar, draw = no bar, loss = red bar), venue (black bar = home, red bar = away), and the usual summaries of a season’s performance.  The graphic contains a good bit of data that may be best viewed by opening the original image, and referencing the chart when reading the text below.

The Results

Based upon the graph above David Moyes’ tenure can be divided into two parts – a tumultuous first four years and a more stable second five years. Each era has unique characteristics to them, and thus are analyzed separately.

The First Four Tumultuous Years

David Moyes took over Everton on March 14, 2002 in a bid by Everton to avoid relegation. Everton were sitting 16th in the table with 30 points, a -8 goal difference, and were level with the 18th place club on points (Bolton). Moyes’ 4-1-4 record in the final nine matches was enough to secure 15th position for a club that had fallen on hard times, avoiding the relegation that went to Ipswich, Derby and Leicester.

Even though Moyes overhauled the squad in the Summer of 2002, his £XI at the start of the 2002/2003 season was still 25% cheaper than the final match of the 2001/2002 season. This was due to modest investments in the form of defender Joseph Yobo (£5M original purchase price, £10.3M 2011 CTPP) and goalkeeper Richard Wright (£3.5M original purchase price, £7.2M 2011 CTPP), combined with loans for Li Tie, Rodrigo, and Li Weifeng.

Midway through the season Moyes’ signed Brian McBride on loan, who contributed four goals through the end of the season. Moyes’ achieved the lowest variability in £XI during his entire Everton career during that initial season, improving Everton’s table position by eight places and their PPM Differential by 0.6 over the previous season, for which he was awarded the LMA’s Manager of the Year award. During that first full season Moyes outperformed the m£XIR model by 0.26 points per match, via his second-least expensive team that hovered just below his career-long first quartile for much of the season.

The success was short lived. A paltry £2.75M (£7.1M 2011 CTPP) was invested over the summer of 2003 (Kevin Kilbane, Nigel Martyn, and James McFadden), of which the trio had a combined 55% utilization rate (i.e. the games they started). The season was a near disaster for the club, especially a collapse over the last 10 matches (1-4-6, a drop of more than 6 points versus the m£XIR model) that saw Everton barely avoid relegation.

The 2004-2005 season may have been Moyes’ finest. Prior to the season the club was overhauled. Tim Cahill and Marcus Bent were brought in over the offseason (£2.4M combined original purchase price, £5.6M 2011 CTPP), while a number of players from the previous year’s squad, including Wayne Rooney, were let go or sold. With a starting XI in the first half of 2004-05 that only cost 62% of the final 18 games of the prior season, Moyes’ Everton would amass an 12-4-3 record through the first 19 matches of the season; sitting in third position with 40 points, a +8 GD, and 15.1 points more than his team’s cost would suggest he should have had (a 0.79 per match over performance).

After such a hot start Moyes’ Everton side only earned 21 points from the remaining 19 games, falling back to a more pedestrian 0.30 points per match overperformance. This despite the fact that he broke Everton’s transfer record in January via the purchase of James Beattie (£6M original purchase price, £13.9M 2011 CTPP). Everton’s fourth place finish qualified them for the Champions League play-in round, making the 2004-2005 Everton side the least expensive Champions League qualifier in the history of the Premier League.

Further investments were made in the summer of 2005. A total of £17.3M was invested (£40.2M 2011 CTPP) to push Everton’s squad cost over the £100M barrier in 2011 CTPP for the first time. Just like the 2003/2004 season, a prior successful campaign was followed up by an abysmal one. Everton didn’t make it out of the Champions League play-in round, and struggled to a 5-2-12 record the first half of the year, putting them in a relegation battle in 17th table position courtesy of an under performance of nearly nine points to the m£XIR model. As has become par for the course, Moyes was able to turn things around to some extent during the second half of the season, briefly going positive between matches 26 and 33 before settling in slightly to the negative.

Moyes’ investment levels during his first four years did not seem to correlate to much season-long success. Nor did his utilization rates. Moyes’ lowest utilization rates (the percentage of his squad, in terms of what they cost, that made it into the XI) come in these first four seasons – 46.8%, 42.9%, 39.4%, and 47.4% respectively, much lower than the league averages those years (49.4%, 50.0%, 51.1%, and 47.1%).

Variability in £XI from match-to-match did go up each of the four years, nearly doubling from his first to his third year. After investing over £84M (2011 CTPP) in the club’s transfers, Moyes’ had an average finish of 9.75 over the four seasons and the team only looked to be marginally better at earning points commensurate with the expectations set by transfer expenditures. Not an auspicious start for a manager who has lasted longer than nearly any other than Ferguson and Wenger.

The More Stable, But Declining, Second Five Years

The £XI trendline in the graph above demonstrates the monumental shift in Everton’s starting XI transfer costs that took place over the summer of 2006. Moyes spent £13.6M (£30.1M 2011 CTPP) on Joleon Lescott and Andrew Johnson (the latter for yet another club record fee – £8.6M at the time or £19M 2011 CTPP). Moyes was also able to achieve a much higher utilization rate than in years past – an average of 62.1% throughout the season. What had been a squad that had only started an average of 44% of its transfer expenditures the four years prior had now increased it’s utilization rate by almost 20 percentage points!

With increased £XI comes increased expectations, with Moyes’ results versus those expectations in the 2006/2007 season looking like two bell-shaped curves reflected over the half-season mark. Moyes was up nearly 4 points versus the m£XIR model at the quarter season mark, but by Match 19 he had dropped back in line with expectations.

The acquisitions of Tim Howard (£3M or £6.6M 2011 CTPP) and Anderson de Silva (£0.7M or £1.5M 2011 CTPP) further cost Everton, but just like the beginning of the season the club saw a resurgence at the start of the second half of the campaign. The £XI fluctuated a good bit more in the second half of the season, as the normal wear-and-tear on players and experimentation by Moyes led to changing starting lineups. Still, Moyes was able to outperform the expectations set by his expenditures by nearly 5 points at season’s end. This represented an improvement of nearly eight points and five table positions over the previous year’s disastrous campaign, and a spot in the next season’s UEFA Cup competition.

Yet again Moyes spent fairly big in the summer of 2007. He bought four players for a total of £23.25M (£33.6M 2011 CTPP): Leighton Baines (£6M/£8.7M 2011 CTPP), Phil Jagielka (£4M/£5.8M 2011 CTPP), Steven Pienaar (£2M/£2.9M 2011 CTPP), and yet-another Everton record transfer in Yakubu (£11.25M/£16.2M 2011 CTPP).

While the entire 2007/08 season represented Moyes’ highwater mark in terms of average £XI and season PPM differential, Everton’s most impressive form came at the beginning of the second half of the season. The team went 7-1-2 in the first ten games of the second half of the season. At that point Everton was fifth in the table, level with Liverpool on points and only two points behind Chelsea for third. This was Everton’s peak against the m£XIR model – 15.98 points to the good for a PPM differential of 0.551. Everton would finish 2-4-3 and remained in the fifth table, seeing their season PPM differential fall back to 0.34.

While it didn’t equal Moyes’ Champions League qualifying campaign of 2004/05, the 2007/08 campaign did represent a number of bests for his tenure at the club. Moyes achieved his highest average £XI (£66.4M), his highest point total (64), and his best season PPM differential (0.34). Most importantly, he broke the seasonal boom/bust cycle of his first four years.

The next three seasons saw Moyes solidify Everton as a solid upper half of table team, finishing 5th, 8th, and 7th. While the end results may have been acceptable to Everton’s management, they have demonstrated a steady decline in Everton’s performance. While Moyes’ point totals remained relatively stable in the 2008/09 and 2009/10 campaigns, the resultant table position had slipped to 8th by 2009/10. Moyes’ 7th place finish in 2010/11 hid the reality that he achieved his third-lowest point total of his nine year tenure, and would have only been good enough for 8th in 2009/10 and 6th in 2008/09.

Each of the last three seasons have seen Moyes’ dig an early hole versus the m£XIR model, with the hole getting deeper and lasting longer with each passing season. The first two seasons of this three season period also saw Moyes’ spend reasonably big in the transfer market in the hopes of qualifying for the Champions League: £26M in 2008/09 (£27.2M 2011 CTPP) and £21.7M in 2009/10 (£27.9M 2011 CTPP). By the time the 2010/11 season rolled around, Everton’s (poor) finances caught up with them and Moyes’ transfer expenditures fell to a paltry £1.25M (both transfers saw a total of two matches of action). The 2010/11 season also saw Moyes’ highest £XI variablity since 2006/07. Overall, the three season trend cannot look good to Everton and their supporters.

Conclusions

The reality is that Moyes’ magic at Merseyside may be coming to an end. Moyes spent four of his first six seasons to the positive of the m£XIR model, steadily building a team that moved from inconsistent boom/bust cycles to one that moved towards consistent Europa League qualification. Unfortunately, seasons 7 through 9 have seen Everton slide steadily backward despite significant investment the first two years. The team now faces a cash crunch with an established manager known for Europa League, but not Champions League, qualification. It’s a tough sell for a club desiring to attract top talent.

For Everton to truly compete with the Big Six they need a much bigger stadium and other sources of revenue to support the spending that is required. In a drive to balance their books and look for additional investors, Moyes and the club will have to go back to the days of a starting XI that costs around £45M and enjoy the financial benefits of weekly wages that are also at lower levels.

The question is whether or not Moyes’ can recreate the magic of 2002/03 and 2004/05 while additional revenue sources are acquired, or perhaps whether he would even accept such a challenge. One might imagine that he would not be too enamoured with the idea of rebuilding Everton and going back to “the good old days” after he spent the last nine seasons building up the team and undoubtedly avoiding opportunities to manage at other clubs. Given his reputation within the league and his record on paper, he might find a few eager suitors at other clubs, both domestic and foreign. The domestic avenues are likely going to require a step back, as the managerial positions at the Big Six seem pretty secure at this point.

Moyes’ finds himself in a tough position. The club he’s helped to build up may now be facing a slow decline back to expectations filled with boom/bust cycles. His options inside the Premier League are a bit few-and-far-between right now unless he wants to start a building project at another club which has the cash to push beyond its current level.

It will be interesting to see how 2011/12 goes for him and Everton, as the last three seasons indicate he may find himself in the negative at seasons’ end versus the m£XIR model for the first time since 2006.

Zach Slaton is the author of A Beautiful Numbers Game blog, and occasional contributor at the Transfer Price Index and The Tomkins Times blogs. You can follow him on Twitter and Facebook.

6 Responses to “The Declining Fortunes of David Moyes (An m£XIR analysis)”

  1. Rob February 15, 2012 at 2:26 pm #

    My apologies if I am misunderstanding something as this is my first exposure to this site (although I am a big fan of this type of analysis which I find is relatively lacking compared with the N. American sports).

    Wouldn’t net spend be a more relevant factor? The only major expenditures Moyes has made came on the back of selling players: most notably Rooney and Lescott (although he got a good price for Andy Johnson and some others). If you remove two players who have gone on to play Champions League football it’s not logical to suggest simply spending money should result in an improvement in performance. In fact I would argue the contrary: to take a proven player and replace them with an unproven player (even if you spend all the money you obtained for the existing player) is probably more likely to result in a decrease rather than increase in performance. Sure, sometimes it might work out, but it’s very hard to predict how a new player might perform in a team (compared with the likely great performance which earned that player a high priced transfer).

    Everton sold Pienaar recently for very little — they may have gotten some money to spend on transfers but nobody who watches Everton would suggest for a second we were likely to improve as a result of losing Pienaar and adding another player. His recent return on loan highlights how he has a far greater value to this team than he warranted in a transfer fee (or could be replaced with a new player from an equal fee).

    While this is understandably hard to quantify at least a net approach would begin to address some of these issues. Everton are an interesting case since they sold Rooney and Lescott for around 50 million combined however their net spend over Moyes era is practically zero. Meanwhile other teams may have had less in total player fees paid but a higher net spend.

    To suggest Everton “spent” 26 million here or 21 million there is misleading because in order to spend that money they lost players whose value to that team may possibly have been greater than the value of the new players. Meanwhile another team could spend 10 million (which is “less”) but without selling any key players in order to buy and therefore the higher net spend would result in a stronger squad.

    • Zach February 16, 2012 at 4:34 am #

      Rob -

      Thanks for stopping by and for your feedback.

      To a certain degree, net spend is factored into the TPI. We calculate the squad or starting XI cost (depending on which metric we’re using) in current transfer prices. That means we take the last transfer fee paid for each player on that starting XI or squad, and then inflate it for each season’s inflation rate since their transfer to their current team. Each player’s inflated cost is then added together to generate the squad/starting XI total costs. So, a team’s net spend is reflected somewhat in their current squad/starting XI values.

      Where there is a bit of a gap is in that the fee someone pays for Everton’s players doesn’t feed back into the the direct cost of the player’s value at Everton prior to the transfer except via the inflation factor calculation. So, in the case of Rooney or Lescott you’re right. Their new clubs valued them way more than their current CTPP value would indicate in prior seasons.

      This is counteracted a bit by looking at the relative size of total transfer values at clubs (MSq£ and M£XI metrics). These metrics take the absolute values (Sq£ and £XI) from each season for each club and divide them by the average values of those metrics for each season. Thus, the “multiple of…” metrics. So, if Manchester United buys Rooney from Everton for a large transfer fee, the next season their £XI will go up, Everton’s goes down, the average Sq£ and £XI goes up, and thus Everton’s M£XI and M£Sq goes down (and ManU’s goes way up). The same can be said of the match-based m£XIR metric.

      I’d suggest reading the “2011 Update to the MSq£ to get more details.

      No metric is perfect, but we’ve found this one does a pretty good job. I’d recommend keeping an eye out for Soccernomics 2nd Edition. We’ve received word that Kuper and Szymanski take this theory seriously and are relatively complimentary.

      • Rob March 22, 2012 at 1:44 am #

        Thanks for the reply. I will check out that 2011 update article and also Soccernomics (as I very much enjoyed the 1st edition).

  2. Rob March 22, 2012 at 1:45 am #

    Oh and FWIW I’m ordering a Kindle version of your book too. ;)

Trackbacks/Pingbacks

  1. Soccer: The Declining Fortunes of David Moyes (An m£XIR analysis) » Stathead » Blog Archive - July 15, 2011

    [...] The Declining Fortunes of David Moyes (An m£XIR analysis): At Pay As You Play, Zach Slaton uses the relationship he found between transfer fees and win probability to judge how much bang for his buck the Everton manager has gotten in recent seasons. [...]

  2. A Comprehensive Model For Evaluating Total Team Valuation (TTV) | Pay As You Play - May 2, 2012

    [...] resources bring to individual matches.  Along the way we’ve profiled a number of managers and clubs, come up with a list of the best and worst performers versus our models, commented on [...]

Leave a Reply