My FPL Experiment: Lessons Learned

After my most successful FPL campaign in 2019-20, one that would have been even better had I not botched my triple captain chip on a GW-38 punt of David Silva’s farewell instead of KDB (approximately a 50 point swing not in my favor), I decided to experiment in 2020-21.

Experiment, in fact, might be too soft a word. Perhaps it’s best to describe the change as a complete and radical overhaul of my decision-making process.

The new process, would be based on data analytics. It would suppress my hunches and instead rely on cold hard math. It would de-emphasize what actually happened while emphasizing what was more probable. This new process took me deep down the data analytics rabbit hole. It made me fully conversant in the language of Statsbomb, Smarterscout, Understat, and The Athletic’s Tom Worville. I relied on algorithms for my transfers.

I became a machine.

I was not prepared for the result.

~

As of the start of GW-30 I am actually on pace to improve my overall points from my successful 2019-20 campaign, but my overall rank is stuck in the 400Ks. I simply can’t gain ground. My green arrows have been far too modest, and I keep backsliding.

On the face of it, my new process has been a failure.

And yet, there have been some very real successes. I’ve consistently transferred in differentials a week or two before the competition got wind of them. I successfully transferred out players just before a cold spell because I saw their underlying metrics were not supporting their actual outputs.

The international break has been a great time to sit down and look at my new process. I want to collect what lessons I’ve learned so I can decide what I must do to right the ship as we head into the last mile before the endgame.

---

Lesson One: The Captaincy

A key failure this term has been my poor captain picks. In approximately 80% of the game-weeks, I picked a suboptimal performer relative to other picks available in my XI.

I already said that my process was quite effective at identifying expected high achievers in a given game-week. However, I did not trust these picks with the arm-band; far too often I went with the popular choice (e.g. Salah despite so-so underlying metrics) instead of a performer whose metrics were far more explosive (e.g. Patrick Bamford, Raphinha, KDB).

Other times, I tried to be more creative, and I’ve had terrible bad luck. For example I captained Ederson in DGW-26 and he blanked twice.

I don’t know the optimal way to pick a captain. I don’t think such a thing exists. In my desperation, I will try something all professional statisticians recommend: throw a dart blindfolded.

Starting in GW-30 I’m picking my captain out of a hat. Serious.

Photo by Laurence Griffiths/Getty Images

---

Lesson Two: Endowment Effect

Gosh have I missed some sitters this term.

Game-week after game-week, I stubbornly stayed loyal to Alex McCarthy (my GW-1 keeper) and found reasons to ignore my compatriot Emi Martinez. Aston Villa’s (and future World Cup 2022 winning) Argentine keeper was always cheap, talented, and hurting my ranking. Yet I was worried about Tom Heaton’s return from injury, and I kept thinking I’d want three Villa outfield players, and I kept believing Hasenhüttl would sort Southampton’s problems.

In other words, I succumbed to the endowment effect: overvaluing what I had relative to something else that could be even more valuable. The endowment effect is a very natural human bias we all suffer from... but I was supposed to become a machine. Ooops.

Emi Martinez, you future World Cup winning prince, I’ve learned my lesson (but given your upcoming fixtures, I’m going with Edouard Mendy instead)...

Photo by Matthew Ashton - AMA/Getty Images

---

Lesson Three: Arrogance

Math, my teacher used to say, is like a superpower. And once armed with a superpower, it is very easy to become complacent and arrogant.

Let’s just say it took me forever to buy Bruno Fernandes. His run-of-play metrics were laughably not supporting his PK-enhanced actual output. In other words, he kept bashing in penalty goals while his expected goals (xG) kept flashing warning signs that it was not sustainable. Comments at the time told me to ignore PK duty at my peril, but did I listen? No.

So I arrogantly downplayed Bruno’s performance and sought statistical comfort by drafting the likes of Ross Barkley, Marcus Rashford and James Rodriguez whose expected goals and assists (and other more complex metrics) were superior. While those three did a job, none of them produced anywhere near the barrage of points that the Portuguese Red Devil did. Eventually I bought him after about ten weeks of punishment.

Photo by Ash Donelon/Manchester United via Getty Images

I similarly was a bit too slow on Everton’s Dominic Calvert-Lewin at the beginning of the season (“DCL is too reliant on headers that others serve up for him”); and I stubbornly held too many duds whose performance never quite matched their expected greatness (thanks for nothing Lucas Digne, Andrew Robertson and Jack Grealish).

---

Lesson Four: Luck

Though none of us want to admit it, luck plays a huge role in FPL. How many of us skilled players did everything right and transferred in Tomas Soucek only to bench him when he exploded for double digits and play him when he blanked?

My luck this season has roughly evened out, as math suggests, and I’ve learned to lay off overthinking tough start-or-bench dilemmas. For every John Stones brace I started, there’s been a Gundogan explosion that I’ve benched. For every Soucek double-digit I benched, there’s been a Michail Antonio goal.

Photo by Visionhaus/Getty Images

Easy fix: No more struggling over tough bench or start problems. I’m just going to flip a coin and let chance decide.

Oh, and few of my -4 hits have paid off. I need to minimize these because on the whole I’ve lost more than I’ve gained from them.

---

Final Lesson: Trust the Process

Perhaps I am too tough on myself. It may seem that my season has been a catastrophic failure.

It hasn’t. My process has led me to some amazing player picks before any of my competition saw them: I was early on Son and bailed at more or less the right time; I was early on Jack Harrison and then Raphinha; I was bullish on Luke Shaw and acted on it; and I trusted John Stones over Cancelo, reaping the rewards. My strikers have been really successful, and I seldom strayed beyond a core of Harry Kane, Jamie Vardy, Patrick Bamford, DCL, and Michail Antonio.

As we head into the last mile then, I am going to trust the process, and I will incorporate some of the lessons above. Pick captains at random; don’t overvalue a player on my roster if the data says I can have someone better; don’t be arrogant; pick Bruno over Sadio Mane); and don’t sweat tough benching decisions.

Just roll with it.

~

Thanks for reading. What lessons did you learn during this crazy FPL season? Do you have a tried-and-true formula that works, or are you always looking to improve your decision-making process? Please share your views, comments and stories below!

~