I decided to try something different for this post.
I always thought that are some affinities between investing and Fantasy Football and wanted to write about my approach to it since a while. What prompted me to do it now is not because I won (yeah! it helps) but because I had a very different feeling back in September when the season started. First few premises. Football is that sport that is played with your feet and a ball, hence the name football; I do not know why Yankees insist to call football a game played 99% of the time with their hands but if you are looking for that, bye bye. I do not particularly like football but I play fantasy football with friends / schoolmates since I was in high school because I love it: this was the 27th consecutive season (despite my wife made me promise three year ago that I would stop once we had our first baby…guess what? Did not happen). As for FF, I just like the social aspect of football: watching a game with friends at the pub, playing in the park; you will never find me watching a game at home, I find it extreeeeeeeeeeemely boring. My method works even if you do not know the names of the players…like you can make money with index funds even if you do not know which companies are in those indexes.
The Rules
A league is normally contested by 8 teams (since some years we do it with 10, making it slightly harder but funnier): in the game, you are the owner and the manager of your team. Each team has 23 players: 3 goalkeepers, 8 defenders, 8 midfielders and 6 strikers. For us, the players pool is based on the Italian Football League (Serie A) but you can use whatever League. At the beginning of each season we do a ‘draft night’: each team has a budget (1000 credits) and league’s participants bid players one by one like in an auction, higher bidder wins. This way, each player can only join one team; in our league, the order in which players are auctioned is 100% random.
When the auction is completed, we create a season that mimics the Serie A season, a quadruple round-robin (because we have 10 teams instead of 18) tournament where a team faces each other team four times in a random order. Each game, a team presents 11 players on the field out of 23 of its roster (plus 7 on the bench) and points are assigned based on the real performance of players in Serie A. I will not go in any deeper detail because if you are reading so far you already have an idea of what I am talking about.
The similarities with investing
This is the most interesting part of the process, at least for me; I will present the model itself later in the post, this part is more relevant to the general audience of this blog. Like investing, I look at FF as a ‘problem to be solved’.
You have to find a measure for ‘fail value’
Each player has a cost like a stock has a price: considering the bidding process in FF, the ‘market’ (other players) set the price for each player and you can decide to ‘buy’ (bid slightly more) or leave it. A player value is based uniquely on his future performance, which similar to stock is unknown to the game participants. In stocks you pay today for future cashflows (as explained by the Discounted Cash Flow model), in FF for future goals. The most common mistake is to take a player (stock) no matter the cost. Think about Tesla, to justify the end of 2020 price the firm has to produce almost impossible future results and yet people were buying it because…it is Tesla (or Peloton, or Snowflake, or…). In this sense the stock market is more complicated than FF because you do not have an end of season set in stone: Tesla price can stay irrational, away from its intrinsic value for years; I am using Tesla just as an example here, it is possible that today price is actually correct and the company will have that spectacular future economic performance.
If you buy something overvalued, you might lose money or underperform your benchmark, lose in comparison to what you would have gained if you invested in a different way. In the same fashion, your overpriced football player will bring you less goals/bonuses and you will underperform in your league, a.k.a. you will not win.
Luck is huge factor
I am not fooling myself thinking my method works because I won this year. There are multiple metrics, including how it performed in the past, that made me conclude that I might be onto something. In the stock market lots of investors can be right for the wrong reasons. You have to understand the reasons that led you to the specific result you are experiencing; this process is hard and the right evidence might manifest after decades. As Annie Duke put it, avoid ‘resulting’.
No one knows all the players like no one knows all the stocks
The universe of possible choices is simply too big to be an expert on every single item (be it a player or a stock); you can have your favourite team like you can be a specialist on a sector but you need a process, guidelines to drive you in the right direction. In FF you can acquire players on the cheap because even if everyone knows the basic characteristics of each (starter or bench, role, etc) what makes the difference is the relative value of a player compared to all the others. There are two cognitive mistakes in what I would call a typical FF league (this is based on my experience, it might be that you play against other PhDs, in which case you still need this to close the gap against your competitors). The first one is that competitors dedicate not enough time to the bottom half of their roster (i.e. defenders from 5 to 8): those spots are typically considered as fillers and anyway players that would rarely be on the field. Here not enough attention is paid. The second one is that competitors dedicate too much time to star players, so much that they think they can quantify how much Top Player A should command has premium over Top Player B. Reality is that, given the complexity of the game (a player can suffer injuries, clashes with his manager, be sold during the season, etc), you are always better off taking the cheaper of the two, whoever your ‘market’ set it to.
In this regard, FF is way easier then the investing game because your process has to be better than the one of your competitors (a limited set); in the stock market, no one is expert on every name but every time you buy a stock, you are competing against experts on THAT name. This is why the ‘active’ game is so much more difficult than the passive one. If you become an expert on a sector but then that sector falls out of fashion, you can overperform that sector peers and still lag the overall market.
You have to trust the process
If you were a Value investor in the last ten years you know what I am talking about. It is really hard to stick to what you know works when the whole world is going in the opposite direction. In this context, FF is actually harder than the stock market because you are in the middle of the auction process, while I am 99% sure you do not trade stocks from NYSE floor. If you have 3 or 4 friends bidding Cristiano Ronaldo to the moon, it is human to want to be part of the action; if you see others fighting over someone you did not know existed 5 seconds before, I bet you feel like the sucker at the poker table.
When you try to acquire undervalued assets, at first you feel like your team (portfolio) is full of crap. At the end of this season auction last September, I was exactly in that mood. I had a solid midfield because everyone else was fighting for strikers (obviously) and defender/wingers that might score; I did not have any superstar. FF is a social topic among Italians, it is not uncommon to discuss about your team with friends that do not play in your league; when I was describing my team, the typical reaction on the other side was a “poor you idiot” at best. After the first three weeks, I was sitting at the bottom of the ranking, questioning if my process was busted. This is totally normal when you try to build a ‘contrarian’ portfolio, you avoid the hot names and buy the discarded ones. But it is really hard to overcome the social aspect and the risk of becoming a pariah. You become this meme:
The Process
My strategy is based on two pillars: the fair value of each role and the players ranking. First an important disclaimer: we are talking about a stupid game, I think devoting more than an hour of your life to tinker about this is time wasted. I personally wasted toooooooo much time (but def not enough to make this process remotely close to perfection).
The Ranking
For each of the four roles (Goalkeeper, Defender, Midfielder and Striker) you have to rank all players from top to bottom; since there are hundreds of them, what FF players normally do is to prepare their own list of circa 10% of the names for each role, very suboptimal. The added issue in this task is that you do not have to rank a player in a vacuum, his value depends on how his team manager plays him and if his manager is able to keep his job for a whole season. The variables and possible combinations are high. My solution is to resort to the wisdom of the crowd.
There are several websites that organise FF competitions that have to rank players: since they have to keep the game as fair as possible, their list has to reflect their best estimate of each player characteristics; they also re-adjust player values to reflect competitors choices, like in the stock market: if a player is taken by more than X% of teams, that player valuation will rise compared to the others, otherwise the organiser risk to have a not balanced game (and a financial risk if it is a betting company). We do not care about each player value, we care about his position compared to the others: if Gazzetta values Ronaldo 40 (out of 250), it does not mean that you want to bid it up to that value; you just want to know if Ronaldo has a higher/lower value than Lukaku and what’s the relative difference in values. These website rankings are better that your own but they are still pretty rubbish.
My preferred choice is a list that I usually find on a specialised forum called Gruppo Esperti. Each year these guys put together a quite detailed excel list that value each player based on ten inputs, all focused on FF. It is still not an 100% quantitative method but I like their approach. Since the explosion of Moneyball, quantitative approaches are emerging also in the world of football; there are some websites that produce really cool player analysis but they are usually behind a paywall, making them out of scope for my goal. You have also to pay attention to the fact that their objectives are not 100% aligned with yours: in the ‘real’ football you need disruptors, players that can stop the other team play and steal the ball, whereas the only thing you care in FF is a player ability to score or to facilitate a goal.
The easiest and most useful proxy of a player value is his salary. Since some years I started to integrate the Gruppo Esperti list (70% weight) with player salaries (30% of weight of my overall ranking). The football world is so competitive that in aggregate a baller salary is a good predictor of his value: some might be overvalued and others undervalued but those errors get washed out when you look at hundreds of them. You want players that actually play in your team, if a real team invested a lot in a particular guy you have higher chances that they will try hard to make it work before ditching him; if the player is out of his team project, he might still play so that the team can increase his value when they look to sell. If you are curious about the quantitative analysis behind the salary thesis, I suggest you the book Soccernomics.
The Fair Value
In Italy, a typical FF league is made by eight teams. Over the years, I developed a quite strong opinion (I did not properly tested it though): leagues participants are pretty good at identifying the top eight players in each role but inside that cluster they rarely rank correctly the players by price/future value. This is understandable because there are several ‘random’ factors that can impact a player performance: injuries, coach preferences, adaptability to the Italian League if they were playing elsewhere before; this last point became even more important in recent years: Italy is not the place to be anymore in the football world, exceptional players are sold to other, richer European teams (sometimes even China) and Italian teams acquire players that had sub-par years in other big Leagues. Real team rosters change more frequently than in the past and this reduces the number of ‘staple’ players in the pot.
All of these means that you are better of bidding the cheapest player in each 8-players cluster. Let’s take a real example: at the beginning of the latest seasons, the top 8 strikers in mine (and I suppose almost everyone else list) were Lukaku, Zapata, Immobile, Ibra, Ronaldo, Dzeko, Caputo and Belotti. The average FF participant spends a lot of time trying to forecast if Ibra will be better than Ronaldo or Lukaku and therefore try to pay them accordingly; you are instead better of assuming an agnostic approach and take the cheapest one, whatever the name that your group bid less.
The issue with this approach is that players are not actioned simultaneously but one after the other. You can only realise which one is the cheapest once all the players are auctioned…but at that point is too late for you go back and bid. This is why you need a fair value for each player in your roster. Going back to my previous example, a top striker fair value according to my model is 300 credits: you should bid any name in the first cluster of strikers in your list (the top eight names) up to 300; if a name is bidded above this level, you are out of the auction. Once you win a name, that role is filled and you can only bid for players in other clusters.
Here the fair value for each cluster of roles:
| Goalkeeper 1 | 77 | Midfielder 1 | 135 | |
| Goalkeeper 2 | 1 | Midfielder 2 | 70 | |
| Goalkeeper 3 | 1 | Midfielder 3 | 32 | |
| Midfielder 4 | 18 | |||
| Defender 1 | 37 | Midfielder 5 | 12 | |
| Defender 2 | 18 | Midfielder 6 | 1 | |
| Defender 3 | 12 | Midfielder 7 | 1 | |
| Defender 4 | 5 | Midfielder 8 | 1 | |
| Defender 5 | 1 | |||
| Defender 6 | 1 | Striker 1 | 300 | |
| Defender 7 | 1 | Striker 2 | 150 | |
| Defender 8 | 1 | Striker 3 | 97 | |
| Striker 4 | 26 | |||
| Striker 5 | 1 | |||
| Striker 6 | 1 |
Each cluster is formed by eight players. This means that the top 8 defenders will be in the first cluster, from 9 to 16 in the second and so on. The fair value of each cluster is based on historical performances. The trick in determining those values is that you do not have to concentrate only within relative values in each role, but also decide when/how your third midfielder will contribute more than your third striker and therefore you should bid more; each week you can play with different formations, set of roles: the most common ones are 3-4-3 (i.e. 3 defenders, 4 midfielders and 3 strikers) or 4-3-3.
There are a lot of variances to the standard set of rules of the game. The most commonly used is that if your team has 4 or 5 defenders on the field, you can achieve a bonus based on their performance. Another common mistake that I observe frequently is that participant decide BEFORE the auction they want to achieve this bonus and therefore pay a premium for defenders. This is a suboptimal strategy: you should aim at the bonus only based on what the other participant decides to do; if defender prices are high, you will only field 3 and forget about the bonus, if they are low, you will go for it.
If you already have filled the top midfielder spot, you can bid other players in the first midfielder cluster but only up to the fair value of your second midfielder, i.e. 70 (most likely you will not get any). If you get a name that you had in the third cluster at the price of the fourth, you can move that name in your fourth spot and continue to bid for players in the third cluster: this is the way to achieve an over-par team.
There is value in flexibility. You have a set number of spots in your roster to fill and once they are gone, you cannot bid anymore for a player in that role. If other participants already have eight defenders, your competition for the remaining ones is lower: there is a balancing act between jumping on a ‘deal’, a player that is cheap compared to his fair value, and waiting for an even better deal in the future.
During the auction, you will build a ‘reserve’ of credits, made by the difference between fair value and the actual price you paid for each player. You can splurge this reserve on your favourite names, bidding them over their fair value; just remember to first build your reserve and then overspend, do not do it the other way around. This is also the best way to build a war chest for the ‘mid season’ auction if your league has one.
Conclusion
Given that I use this method since years, and I developed it, I am sure I forgot to mention relevant details in the above tutorial; let me know in the comment section below if you have any questions. If you developed your fair value method and want to share, I am curious as well! I hope my post will bring you success next year 😉
What I am reading now:
Follow me on Twitter @nprotasoni