About Shom

Shom Biswas is a writer from India. @Spinstripe

FPL Cognitive Biases: Framing Bias (or) Sakho AND PVA, are you kidding me?

So here is the thing. Most of you know that the 5.0M defender I am most comfortable having in my team, is Sakho (bonus magnet. Uncle Roy’s teams definitely will get ~11-12 CSes). And perhaps most of you have figured out that the 5.5M defender I am most comfortable having in my team is PVA (CSes, as discussed, plus almost guaranteed 5G+5A). Both are guaranteed starters.
So why don’t I have both of them in my team?
Because they are both from CPL, silly. What if both of them concede a freak goal? Both of them would get their CS busted. That is conventional wisdom.

That my friends, is framing bias.

What if both of them get a freak clean sheet? because I will be playing both of them in most games – no? Yes. 5 defenders is my almost-default formation, unless the one true bandwagon MID/FW arrives.
And in the (almost guaranteed) no CS games — MCY away, LIV away — I will still play PVA, because he can get a goal, and I will play the 12th player (Cairney) in place of Sakho, dammit.

I will get Sakho soon.


The Manchester City Line up (from an FPL PoV)

Since I am a bit of a Pep fan, though not one of Man City; and being a formations / strategy nerd, I thought of providing my take on what City’s formations would look like – with specific focus on how it affects fantasy football –  PS: I am not doing one for Arsenal, because my personal biases on players often impact my comments…


Most often 4-3-3. The alternate is 3-5-2,  which will be deployed much less often.

Center Forwards:

Aguero first choice, Jesus second choice. Sterling third choice. I do not see both of Aguero and Jesus playing together often, even in the 3-5-2. There, I suspect Sterling to reprise his role for England as the second striker. Also, here’s a judgment call – last season we saw that when Aguero is fit, he almost always plays. Jesus plays when Aguero is injured. I think it will be a bit more distributed this season i.e. even if Aguero is not injured, the distribution of playing minutes would be about 60:40. Jesus is far too good already to not get a lot of minutes.

CF: Aguero (Jesus) (– Sterling)


Wing forwards in a 4-3-3, and full backs (which should be discussed together, I think)

It seems from the pre-season and GW1 (where my team was quite comfortably beaten) that most of the width will be provided by the full backs than wingers (more Barcelona, less Munich). Interesting to note though that the backups for Mendy (Delph), and Walker (Danilo) are better as inverted fullbacks, than flying attacking fullbacks. I assume therefore that if the first choice are injured , then the formation becomes more Munich for that wing, with the winger providing the width – similar to last season on the left, where the width was provided by Sane, a traditional winger – and not Delph the fullback.

Thus, with Mendy and Walker playing in a 4-3-3, the wing forwards would both be cutting inside (and that was the formation we saw at the Emirates). Sterling on the left, Mahrez on the right. Tentatively I suggest that their first backups are Jesus and Sane respectively. (thus my emerging hypothesis is that Mahrez is the first choice right wing attacker,and Sane is his primary backup. It could as well be Sane first choice and Mahrez as backup.

Thus : in a 4-3-3,

RB: Walker (Danilo) (– Stones)

LB: Mendy (Delph) (– Zinchenko)

RW: Mahrez (Sane) (— Bernardo)

LW: Sterling (Jesus) (— Brahim Diaz)



This is simple. And stays consistent in a 4-3-3 and 3-5-2

Deepest midfielder: Fernandinho (Gundogan) (– Delph) (others–Zinchenko, Gomes)

RCM: KDB (Gundogan) (– Foden)

LCM: Bernardo Silva (David Silva, or the other way around) (– Foden)


Central Defenders:

Very difficult to say. I assume that in normal circumstances, Kompany plays key games. But he will be heavily rotated even when fit. I think Guardiola’s ideal formation is with Laporte as LCB and Stones as RCB, though I do not think they are ready to play together consistently yet, and Guardiola will pair one with Kompany and Otamendi. As for Otamendi, even with all last season’s heroics, I somehow see him as 4th choice, and not a natural fit for Guardiola. He will play enough too, but will not have the security of starts as he had last season. Two of Tosin, Mangala, Denayer and the new guy will be sent on loan I think.


Tactics: Defender Targets for the season


It seems that Defenders are the opposite of the goalkeepers, in that the graph has a more steeper slope than any other position. My understanding from that is that the marginal benefit of having the best higher priced defender over the best lower priced defender is the most in comparison to the price. And remember, this was the season where Burnley pulled out trees, and BHA and NEW did well too. I would guess this feature would be more prominent in other seasons.

Simple Language: The best-performing high-priced defenders are the greatest value.

Tactics: Goalkeeper Targets for the season

I did some number-crunching from last year’s data

https://public.tableau.com/profile/mike … ublish=yes
is the tableau graph I used. Thank you MoSe for sharing

The comparison – levelling exercise that I had in hand was last year’s top 10 and top 10k approximations that RuthNZ made on the 2016-17 players, and the rough estimate guidelines I used for this year.
I only considered players that had played at least 10 matches, (a match is considered at least 30-min appearance), and that had at least a 7.5% ownership.
There are not too many data points obviously, but the point is – I accept that this is not an exact science, and my objective was to get a trend, and a trend-line, to help me decide.

Will not do a detailed story around the numbers here. Here is a bit of synthesized data – make of this what you will. Some basic hypotheses are provided alongwith.

This is the graph for goalkeepers, and the resultant ppg calculation as per the trendline


My big finding? Goalkeepers have the most ppm in comparison to any other position. However, the slope of the graph is less steep than any other position (expect for forward which is about the same. My understanding from that is that the marginal benefit of having the best higher priced goalkeeper over the best lower priced goalkeeper is the least in comparison to the price.

Simple language: if your team permits – the best cheap goalkeeper (s) is the way to go.



Is Salah essential? Yes at (say) 10.5M, no at 13M

There’s a question I am trying to get my head wrapped around. At what PPG is Salah worth getting in the team? I thought there is a mathematical solution to this.

Caveat: You know I do not have Salah in my team. But I have tried to keep the bias out of the equation.

So above are the numbers. How to interpret them?

Part A – If Salah has a similar season to 17-18

On the left
Assume Salah will get 8.4 ppm
The standard 9.5M midfielder (Mane) will get 6.28 ppm as per the gradient of last year.— (NOTE: this is not Mane’s number, but the gradient of all the popularly owned midfielders, and where it coincides with 9.5M).
Also, since Salah will not be a captaincy option, I would assume the captain would be the next highest ppm-getter for the season i.e. Aguero at 7.30.
Now, if you have the control group to compare Salah with as (Mane + Aguero as captain),
You will have 3.5million (no having Salah) giving you 3.22 ppm as your control case.

On the right
While drawing the gradients as we had done below, we have arrived at the ppm gain per 0.5m addition in price. We see that the highest gradient is for the defender, and lowest for the goalkeeper.

Now even for the best case -i.e. we utilize the 3.5m into the best defenders and take advantage of the incremental benefit of the extra money, we would still be about 33 points in arrears. more realistically, if you spread it around (0.5 GK; 1M DF; 1M MID, 1M FW), you are in arrears by ~40 points.

Inference: If Salah captures last season’s form, you have to have him. He is essential. Nothing else you do can top not having him, even at 13M (i.e. unless there is another player who is giving similar returns as Salah)

But if Salah would do a little worse than last season, then? How much worse can he do if you compare him against the most realistic control case (same as what we tried last time

Part B – If Salah does slightly worse than 17-18

On the left
Experiment has given me the number of 280 points. i.e. 7.8 ppm for a 36-game season. This is the tipping point below which having Salah does not remain a realistic option.

If Salah will get 7.8 ppm
The standard 9.5M midfielder (Mane) will again get 6.28 ppm as per the gradient of last year.
The next highest ppm-getter for the season i.e. Aguero will be the captain at 7.30.
Now, if you have the control group to compare Salah with as (Mane + Aguero as captain),
You will have 3.5million (no having Salah) giving you 1.98 ppm as your control case.

On the right
Now even for the worst case -i.e. we utilize the 3.5m into the two worst gradients – 1.5M for GK, and 2M for striker; and take the least advantage of the incremental benefit of the extra money, we just about make 1.98 ppm, par. More realistically, if you spread it around (0.5 GK; 1M DF; 1M MID, 1M FW), Salah will have to get 282 points to just make par.

Inference: Salah will have to get at least 280 (7.8ppm for a 36 gameweek season) points to be a realistic option at the 13M price. Salah’s score can fall only by about 7.5% from his lofty standards of last year for him to not be viable anymore.

PS: even with my horrible captaincy targets (7.0 ppm target), Salah will still need to get 272 (~7.6ppm) to make par.


FPL 18-19: Strategic | Operational | Tactical

The concept that I keep close to my heart while playing FPL is, breaking the play into the Strategic, Operational, and Tactical dimensions.

Strategic: This dimension tries to answer How well a team is performing in comparison to the target points it is supposed to reach during various stages of the game –  to eventually get a great overall score. The Strategic dimension will also answer the question of how well a player is performing in comparison to the expectations emanating out of his price point. This is a static dimension – the changes are brought about at the end of the season.
Operational: This dimension tries to answer questions of how optimal is the team structure to achieve the target score, as discussed above. This is not as static as one thinks it is – because there would be players who overperform their target scores by a great deal – who would need to be brought into the team (these are what are often referred to as essentials). This dimension also tells how well a manager has been able to shuffle his/her pack to achieve the most out of his/her team — namely fit the best juxtaposition of over performing players in comparison to price point, and players that simply get the most points – to get the overall highest points the team can get).

Tactical:  This dimension answers questions of how well the players are doing in terms of scoring points – how well can we predict the amount of points they could be getting in the future, and how well we can understand if a player needs to be transferred in / out – is it a true bandwagon? is it just a bad patch? This is where your xA, xG, fixture-based plans take shape.

—- All three dimensions, put together, are the considerations we have in hand to make a great season. We can and should use all three dimensions

Fantasy Premier League 18-19: Intro

This year, like for ten of the last eleven years, I will be playing the Fantasy premier league https://fantasy.premierleague.com/

I have been a serious FPL player, and have been a regular at FISO, one of the oldest FPL forums. This year, along with playing, I would like to pen down my thoughts, off and on, in this blog. Some of these are a result of discussions with other members at FISO’s FPL group, and I do not take sole credit for all the ideas posted here – excellent posters such as Mav3rick, Stemania, Ruth_NZ, Aldershot Rejects, No Way Jose, blahblah, Smurphy’s Paw, Sutter Kane, MoSe and so many others have contributed much to these ideas.

I call myself the once and future excellent FPL Player. My best rank has been 212 (in the season ’11-’12). I also have got 4 more top 10Ks ; and 2 more top 1%s (within the toop 15K). last year, my rank was around the 6K mark. My ID for this year is 89493.

A couple of introductory statements:

  1. Some of these thoughts are not for the absolute beginner FPL player. There are concepts, terms etc which would require a bit of getting used to. But if you, as a reader, find some use of these – find this website and find some use to it, please ask for directions. I will be happy to help.
  2. I am not keen on advertising this set of posts too much. This is mostly a journal of the year to come. Of course there would be a twitter feed ( @baganboyFPL ), but if you would really want to engage with me and other like-minded folks, the best place is perhaps FISO.
  3. I had done this kind of an exercise for the first 30-odd gameweeks of last season 0 the details of which are at http://www.fiso.co.uk/forum/viewtopic.php?f=99&t=125426  . This is another good place to keep in touch with me.