Moneyball in Football Manager 2019: II

moneyball logo

Right, time for part two of Moneyball in Football Manager 2019. For any of you just catching up with this save now, in Part 1 we outlined what my theory behind Moneyball is, the rules I’ll be following throughout this save, and who I’ll actually be playing as, Glasgow Rangers. As I said in the last update, I selected Rangers because I feel given their tentative financial position, they’re an ideal option to test out pure Moneyball with, and see whether I can utilise this concept to not only topple Celtic at the top of the SPL, but return Rangers to European football and competing at a continental level.

If you remember, the conditions that I’ve applied to my save come largely from the concepts outlined in Moneyball by Michael Lewis, but also from the excellent football economy books Soccernomics by Simon Kuper and Stefan Szymanski, and The Numbers Game by Chris Anderson and David Sally. If you need a reminder, below are the rules I’m using:

  1. Players are undervalued for a range of biased reasons such as age, personality, appearance. None of these matter. All that matters is production on the field, which can be analysed through statistics.
  2. Several countries are overvalued in the market, such as England, Holland and Brazil and there is better value for money off the beaten track.
  3. Wage spend correlates to success far more than transfer expense.
  4. Prioritise identifying and improving your weak spots in transfer windows. This is the best way to improve your team.
  5. Strikers cost more than they should and can be poor value for money.
  6. Ensure that all potential signings are thoroughly scouted – the wisdom of crowds – before their statistics can be analysed. 
  7. Don’t buy players who performed well solely at international competitions. These are a perfect example of ‘small sample size’, where stats have no true value to them as they can’t be trusted to be replicated. 
  8. Long term, producing your own players and developing through the academy is more cost effective. 
  9. Football is won by scoring more goals than the opposition – like baseball is won by scoring more runs – and therefore attacking stats must be a main focus.
  10. Some of the best value can be found in identifying players who are no longer wanted by their clubs, and are likely to vastly undervalue them.

One thing I do want to discuss before we move on to cover how things are going with Rangers so far, is discuss rule number 9. I’ve had a few people quite understandably ask me about this rule, and what my theory behind it is, so I thought I’d give this some explanation without boring anyone. One of the major tenants of Moneyball with the Oakland A’s, was that because they had discovered the stat of On-Base Percentage to be so undervalued in the market, they focused heavily on this. The theory also grew from this one very simply line that is specifically mentioned in the film:

Your goal shouldn’t be to buy players, your goal should be to buy wins, and in order to buy wins, you need to buy runs.

Now, rather simplistically – I understand – football is also a game that is won by scoring more than the other team, just goals rather than runs. As a result I decided to prioritise attacking statistics to focus on winning games. However, I decided to have a look and see whether my theory – that attacking was more important to focus on than defending – was actually backed up by statistics. After all, statistics are what I’m focusing on here. I discovered this statistical analysis done in 2014 on the Premier League since it’s inception. Now, this of course is the Premier League and not the SPL, but the statistical analysis shows that scoring a goal has nearly twice as large an effect on your eventual points total as conceding. Therefore that indicates that attack is more important than defence. When the top PL teams are analysed, the effect of scoring comes down, but is still 1.5x more important rather than 2x more important. The only analysis where not conceding is even close to scoring in terms of importance is home/away analysis. Even then, scoring is still more important than not conceding at home, although there is only 0.10 of equivalent difference. Away, the importance of attacking and scoring was huge:

scoring in away games is far and away the best thing a team can do for its expected point total. It is worth a whopping +0.83 points, which means that in the history of the EPL, every away goal was very nearly worth on average as much as playing an extra game (+0.86 points)

Now, I believe this statistical analysis more than supports my theory that I should be prioritising attacking – and scoring goals – more than defending. I should be clear that this doesn’t mean that I’ll avoid looking at defensive stats – much like how the A’s analysed the stats of Relief Pitcher Chad Bradford before acquiring him – but they’re not my main focus. Tactically, I also looked for any possible relationships regarding possession, and found this article on givemesport that ran a statistical analysis that showed there was a relationship between possession and winning, and so I’ve also taken this into account when planning tactically. Basically, I need to play attacking, possession football in order to have the best statistical chance of winning games. Now, on with the analysis of Rangers so far.

Screen Shot 2019-02-09 at 18.27.42

Tactically, I’ve followed the above guidance, and have been using a fairly basic 4-3-3 structure. I initially started off playing a 4-2-3-1, but I wasn’t totally happy with how it was playing, and I certainly didn’t feel like it got the best out of Lucas Piazón. This is the current 4-3-3, designed to play attacking possession football, but it is tweaked for each game, and as a result I wouldn’t feel confident advising anyone to follow this role for role or instruction for instruction. So, how well has this system and our players served us so far this season?

Screen Shot 2019-02-17 at 16.55.25Well, this is how we currently stand in the league. In all honesty, I’m quite surprised that we’re top of the table, but I’m still not confident that we’ll be able to hold onto first until the end of the season. Interestingly, we stand 3rd in terms of goals scored, 1st in terms of goals conceded, with only 14 (and three scored in the last game before I’m writing this), which somewhat doesn’t match up with the statistical evidence I’m basing my theory from, but we do match the evidence relating to possession, as we lead the league on possession per game, with 63% – with the closest to us being Celtic with 53%. The reason why I’m unsure that we can maintain our 5 point gap at the top is that we have played Celtic three times so far in this save, and we have yet to beat them, losing, and drawing in the league, and losing in Extra Time in the League Cup Final.

However, this hasn’t been the only successes we’ve had so far this season. As mentioned above, we also have managed to get to the League Cup Final, losing 3-2 to Celtic in Extra Time after an infuriating catalogue of errors that led to their 3rd goal. In addition, we’ve also had a great run in the Europa League so far.

Screen Shot 2019-02-18 at 08.37.57

Quite surprisingly, we absolutely dominated our group, a strong group containing Spartak Moscow, Partizan Belgrade and Sporting Lisbon. Realistically, I’d have been happy with challenging for 2nd in this group, but probably going out after finishing 3rd. In the end, we easily finished 1st in the group by 6 points, and our reward has been a First Knockout Round tie against German club Hoffenheim. That is certainly going to be a challenge, but after our performance in the group stages, I’m quietly confident that we’ll be able to give Hoffenheim two great games, and we certainly won’t make it easy for them – I hope.

With all of this, we arrived at January 1st, and it was time to really get cracking. In reality, I’d identified our weaknesses by about November, and made the deals in November for January, making only one transfer itself in January. These transfers were made as a result of statistical analysis – Moneyball – but also a simple analysis of realising we had key weaknesses I hadn’t noticed at first that were largely a result of injuries making life harder for ourself. One transfer that I didn’t mention last time was that I sold Andy Halliday to Hibs for £200k, slightly below his true value of £250k, but a negligible amount. Halliday has since gone on to average 6.66 for Hibs, so that’s a clear win there. I also loaned youngster Glenn Middleton to Dundee United, and he went on to average 7.41 for them, dominating from the left wing. Graham Dorrans was also sold to Hibs for £525k after averaging 6.93 for our side. That’s below his true value – according to the game – but he wasn’t performing well for our club, and had become a disruptive influence in the dressing room, so I was more than happy to let him go. In reality, I was happy with £525k for Dorrans. Youngster Aidan Wilson went out on loan to Alloa, and Carlos Peña was sold for £950k to Mexican second division side Dorados – his value is now £875k – after averaging a shocking 6.4 whilst out on loan, and I was also chuffed to get rid of his £25k per week wages. Striker Kyle Lafferty became unhappy that he wasn’t playing – largely because of his 6.7 average rating – and therefore was sold to FC Zurich for what will eventually amount to be £1.2m, which is well above his true value of £975k. I also cancelled the loan of Ryan Kent, who had averaged 6.98 for us, but had chipped in with some important goals for us, but lost his position in the side early on in the season. I cancelled his loan to make way for the returning Glenn Middleton – rule 8 – who had performed so well at Dundee – again, 7.41 average rating – that I believed his statistical performances deserved a chance in the first team. My last transfer was totally unplanned, but when Spurs came in for our first choice goalkeeper Allan McGregor on deadline day, I knew I could get above market value for him, and exploit the money available in the Premier League. In the end, I managed to get what will amount to £1.6m for him, well above his value of £1.1m. This has meant that Wes Foderingham has become our first choice keeper again – he was at the beginning of the season – and I need to plan for goalkeeper moving forward. All in all, I’ve made £5.25m from transfers since the beginning of the save, and my first sale of the save, Daniel Candeias – for £1.4m – has gone on to be valued at £1.1M and average 6.47 in Ligue 1. So overall, I feel I’ve come out of my sales pretty well so far.

Screen Shot 2019-02-17 at 10.44.02Now, the incoming transfers. I’d analysed that our major weakness was in the centre of the park, both through a lack of quality in that area – hence Halliday and Dorrans being sold – but also a lack of cover as a result of injuries. In particular a position of concern was defensive midfield, with Ross McCrorie playing there most games in the first half of the season – tiring him out – because the player I thought would be his rotation option – Jordan Rossiter – spent most of the season injured, and has only played a couple of games because he’s injured so often. As a result my first signing was a defensive midfielder. I specifically used statistics here because I searched for a DM that had averaged over 4 tackles and 4 interceptions per game, with an average rating of at least 7.00. With that as my search, up popped Óscar Gil at Spanish Segunda team Santander. Gil has spent the majority of his career in the 2nd and 3rd levels of Spanish football, so is an ultimate example of a player being undervalued. After my scouts overwhelmingly loved him, he was signed for £475k, with £14k per week wages. This transfer uses rules 2, 3, 4 and 6. It’s an ultimate Moneyball example, finding an undervalued player from a lower leagues, but who statistically fits perfectly, and signing him for peanuts. This signing also allows me to play McCrorie at centre back more often moving forward.

Next up was the player who is shown in the image above, Álex Fernández. This came through my analysis of central midfield being a weak area, and I feel we lacked a little bit of creativity in the centre of the park, with Ryan Jack being my main creative player. I searched for players who had an average rating of above 7.00, showing that they’re in good form and that form is replicable, and up he popped, playing in the Segunda in Spain. My scouts loved him – again, wisdom of crowds – so I signed him for £525k, and awaited his arrival. His value is now reflected by the game as he’s gone up to £3.6m. Again, this is perfect Moneyball, we’ve found an undervalued player off the beaten track, scouted him well, and signed him for next to nothing. The only problem is that mere days before joining the team in January, he injured his cruciate ligament, and so we won’t be able to play him for 8 more months. I was gutted with this one, but I still feel he’ll be a fantastic signing in the long term. He still fulfils rules 2, 3, 4, 6 and 9 of my Moneyball principles.

Next up were two signings I largely made for depth. After Lafferty left, I also cancelled the loan of Umar Sadiq – to free up wage space – meaning I needed to sign a striker to back up Alfredo Morelos. I had one in particular I was looking at, but my scouts spotted Lawrence Shankland for me, available from Ayr for pretty much nothing. He’d been averaging 7.03 for Ayr, scoring a goal every game, and at a fee of £90k, he’s well worth the risk. My scouts really liked him, and he should be a great rotation option for us over the next several years, and hopefully make us a big profit in the future. Shankland fulfils rules 1, 2, 4, 5, 6 and 9. My next signing was one I had gone back and forth on for several months in game. I had seen Syrian Aias Aosman available at 2. Bundesliga team Dynamo Dresden for quite some time, and had been scouting him for the majority of the season. My scouts absolutely loved him, and highly recommended that I sign him. However, when I saw his average rating – 6.64 – I knew I couldn’t sign him because statistically he wasn’t proving he was good enough. However, I then delved closer into his form, and discovered that any time he was on the field for a good length of time, he averaged around a 7.00 rating and was putting in good performances. I then reasoned that as long as I gave him game time, his stats showed that he’d be able to do the same thing for Rangers. The best thing about Aosman is that he was available for next to nothing, and so after a bid of £110k, he was ours. Aosman matches rules 2, 4, 6, 9 and 10 – Dresden had no interest in him anymore. With these signings, we’re well positioned for the second half of the season.

Screen Shot 2019-02-18 at 09.51.00

Young Glenn Middleton smashes in his first goal for the club, in our recent 3-1 win against Livingston.

So, all in all, it’s been a really good season so far. We’ve performed well on the pitch and sit 1st in the league at this point. We’ve never beaten Celtic yet, but that will come in time. I’m still not sure whether we’ll end the season top of the table, but either way this shows the progress we’ve made as a club in just 6-7 months to be up there challenging first season. I’m starting to bring in my Moneyball transfers, and I’m ecstatic with all of them so far. Financially, the club looks healthy with around £16m in the bank, and I still have £1.6m to spend, which is going to be left alone – remember, if I don’t have any weaknesses to fill as of yet, spending money would be wasting money. I’m also currently negotiating a contract with a goalkeeper (who is averaging well above 7.00) to come in for next season – who is also a great example of a Moneyball transfer – so we’ll be strengthened there next year. At this point, everything is looking extremely promising, and the concept is working well. The only problem I have at the moment is I’d like us to score more goals to go along with our possession, but that will come with time.

This brings us to the end of the second update of the Moneyball in FM19 save. Until next time, thank you very much for reading, and as always, should you have any questions regarding this save, the concept of Moneyball, or Football Manager in general, please feel free to ask in the comments section of this blog, or contact me via Twitter (@JLAspey). Thank you again, and I hope you’ve enjoyed reading the second update of this save. I’ll be back as soon as I can with the end of season report, to see whrther we’ve managed to hang on to 1st place. Thank you again for reading.

7 thoughts on “Moneyball in Football Manager 2019: II

    1. It’s pretty basic atm. I send my scouts to areas where they have good knowledge. I’ll then analyse the stats of the players they recommend to see if they’re suitable. Sometimes I’ll search for statistics, using it like a database too. Working well so far.

  1. So you kind of look for stats on positions for example a winger you’d search assists and cross completion something along them lines?

  2. There is a pretty huge statistical flaw in the PL goals vs conceding analysis. The analysis shows that on avg each game played gets you 0.86 points each goal scored gets 0.82 and each conceded loses 0.44. So say you lost each game you played 1-2 then you would still be expected to gain 0.86+0.82-0.88 or 0.8 points per game. This is clearly ludicrous.

    The problem is that you can only do this analysis when the independent variables are uncorrelated which in this case is not true. In general good teams are good at both scoring and defending (or attacking is really the best defense) so the attribution to scoring and defending goals is pretty spurious coming out of this regression.

    Also would rather do this on a match by match basis rather than overall games played goals scored etc because football points are a non linear function of goal difference. ( meaning a goal diff of negative anything is worth 0 , a goal diff of 0 is worth one point and the first goal after is worth an extra two points in each game. Goals after that aren’t worth anything.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s