FPL Gameweek 10: Where do we go from here? Wildcard awaits.
State of things
Nine Gameweeks have come and gone by. An unusual Premier League year with the pandemic and many wild results later, the title race is still wide open. If you’re new here, first of all, welcome. We’re attempting to leverage the use of data available to understand if we can generate more points than the average player playing FPL.
We’ve had some mixed results. We’ve done well in some weeks and not so well in other weeks. The data never lies so we have not cracked the case yet and we may still not. However, The aim is to continue to explore and see if we can beat the monster that is Fantasy Premier League. Given that I have little experience before this year on playing FPL the proper way, we’ve done pretty well. Also taking into fact that several of the players we chose either got injured, banned or struck with Covid-19.
Gameweek 9/10
As you can see, we scored fairly similar to our previous weeks. We were again quite close to the average. But we can assume that there isn’t a significant jump in points with our current strategy. Actually, we got quite lucky given the fact that the player we bought in got injured (Romain Saïss) and several others failed to perform. But as always with us this season, Harry Kane saves the day. Bringing Heung-Min Son back was also a good idea. Ben Chilwell and Timo Werner continue to be valuable contributors to the team.
Admittedly though, doing okay wasn’t why we started this experiment. My early secret lofty expectations of smashing this thing out of the water with data have been brought down to earth. We really have to start asking the important questions. But how do we even start asking them if we don’t know which to ask in the first place? More on this later.
For now, let’s further analyze how we did over the past weeks.
We had two good Gameweeks in GW2 and GW5. The rest have been quite average. We lost a lot of Bench points over 5 Gameweeks. Our highest Gamerank was in Gameweek 5 when we reached the 1.9m position. Despite our terrible form, we are still at a respectable 2.5m rank over 7.5m players which puts us in around the 66th percentile. Not bad, but not exactly what we were hoping for. It would be a sign of laziness if I was okay with that, not just to myself but to you as well, my readers whose time I value a lot. Considering that we’re already a 3rd of the way through the season, it maybe too late to climb to the top unless we got everything perfect from now on till the last Gamweek. However, that shouldn’t stop us from trying. Since we’ve reached this predicament, I’m going to activate a wildcard as my first attempt to rectify things. But herein lies our first challenge..
The General Conundrum
What quantifies as a valuable stat? This is the thing I’ve been pondering over in my head for the past few weeks. Our hope is that it has to be somewhere in the data and we just have to discover it. Unless the Premier League with it’s unpredictable scenarios only depends on luck as the main factor for success.
Let’s ignore some crazy scenarios such as season-ending injuries or someone getting affected with Covid-19. What are the key factors that we are left with? One area is injuries - Do we look at when a player could be most prone to injury? Even then, does past injury always transpire into a future injury? Or how about when a player may actually score or keep a clean sheet - How do we know ahead of time that James Ward-Prowse will score two freekick goals against West Ham? Or How do we know when a goalkeeper is going to save a penalty? How do we know when an underperforming team beats a well-performing one? Or what about players who discover their form suddenly out of the blue moon? There are so many difficult questions here. What data do we even look at in this case? Is it even available to the public?
Even if we had all the data in the world and the tactics straight from the manager, we still are not sure what would happen on the pitch in real-time. From all these scenarios, we have to select 15 players… Sounds like a mammoth task. Yet, here we are and we must try to find an answer somehow to at least a few of these questions. We must either create valid data, look at things in the past data in a way that we haven’t before or change our strategy. There is also danger in not being patient enough but considering that we haven’t seen improvements over 10 Gamweeks, its definitely time for a change.
There is also the thing to consider about prices. FPL states that the value you get out of a player depends on his price rise or fall.
For example, if you sell an in-form player, let’s say Harry Kane, and he was worth £10 when you got him and now is worth £10.5. If you sell him, he will have a value of £10.2 (You only get £0.1 for each £0.2 increase in value). However, if you accidentally sell him and you had wished to keep him (even in a wildcard scenario), now he would cost £10.5 to buy back. That is a loss of £0.3 because of an error on our part!
On the other hand, if his value had gone down to £9.8, your selling value for him would remain £9.8. The profits only matter in the case of price rises. There is also the case of when to buy and sell. Price rises go up and down throughout the week so the earlier you sell, the better chances to make a profit. This is an easy error many managers make. Thankfully, I was made aware of this recently. Hopefully, you knew about it too.
First Steps
We’re only allowed two wildcards during the entire season so implementing our first one now would mean that we can’t change the team anymore until January 2021 (realistically around March 2021). So this is a big decision in who to bring in and who to keep.
Switching the Transfer Pattern
For the first 10 Gameweeks, we did not consider any major transfer strategies. Most transfers we completed until now were reactionary based on bans, injuries or sickness. The players we bought in however were strictly based on best form - the pattern of transfers being one every week.
However, this time around, we’re going to try an alternate one. We’re going to change the pattern of transfers conducted from once every week to once every two weeks allowing for two free transfers. Yes, we will have to slower changes but from the past Gameweeks, I’ve noticed that this allows for further flexibility of being able to swap out multiple positions on the pitch. This could be a valuable change to implement. After all, we won’t know unless we try.
Looking at the schedule ahead
Another area to consider change is deciding which teams to pick players from. In the first week, we chose that the initial part of the season would be tested using the strategy of simply choosing the best players in teams with the easiest schedules. We assigned each team a strength based on their potential to do well. This was a safe strategy to start with and it has resulted in some valuable and safe but average results.
First, let’s take a look at what it would mean if we used that same strategy for the next 7 Gameweeks. Data purists would kill me for saying this but for this analysis, I used Excel to see what schedule lay ahead of us and which teams would benefit the most. Seemed like the best tool to get some quick work done.
If you look at the table above, you’ll see some random numbers assigned to each team over the past Gameweeks. On the left-hand side, we have the initial strength we had assigned to each team when the season began. In the middle, we have the strengths of each of their opponents in display (We assigned a +1 value to each team if they were playing at their Home ground). On the right, we see some numbers assigned on each Gameweek. These numbers are correlated to each team’s results vs. said Opponent.
These are based on a few parameters.
Clean Sheet | Each Goal scored | (+1) Point.
Each Goal conceded | (-1) Point.
A 0–0 Draw (1 Point) | 1–1 Draw (0.5 Point) | 2–2 Draw (1 Point) | 3–3 Draw (1.5 Points).
The reason a 0–0 Draw is more valuable than a 1–1 Draw is in the instance of points scored by the entire Defence and Goalkeeper as opposed to just one goal-scorer and assister in the 1–1 Draw.
For instance, Arsenal beat Fulham in GW1 3–0. Arsenal get 4 points. Fulham lose 3 points.
West Ham drew with Chelsea 3–3 in GW3. Both teams get 1.5 points each.
Southampton beat Aston Villa 4–3 in GW7. Southampton get 1 point. Aston Villa lose 1 point.
Based on these Parameters, we do a few more calculations and we then take a look at the next 8 Gameweeks (GW11–18) for each team based on it’s strength alone and the opponents it may face. Let’s take a look at 4 interesting spreads.
Additionally, let’s also take a look at the list of the teams that scored the most goals up until now. We skip Aston Villa because of their fluke 7–2 win and injury to Ross Barkley. TEAMS TO CONSIDER — Chelsea, Liverpool, Tottenham, Everton, Leicester City, Southampton, West Ham and Manchester United.
We also consider the teams that have had the most stingy of defenses. TEAMS TO CONSIDER — Tottenham, Wolves, West Ham, Manchester City, Chelsea, Arsenal, Aston Villa and Leicester City.
Finally, let’s also consider the teams with the most adequate of goalkeepers. TEAMS TO CONSIDER — Chelsea, West Ham, Southampton, Aston Villa, Tottenham, Leeds United & Wolves.
So we’ve taken a look at several key data points. Now let’s tally them up to see which teams appear the most in these lists. We would also want to consider some teams specifically only for defensive purposes and some just for offensive purposes.
|6|Aston Villa|
|5|Liverpool|Manchester City|West Ham|Arsenal|Tottenham|
|4|Chelsea|
|3|Southampton|Leicester|
|2|Manchester United|Wolves|
|1|Everton|Leeds United|
I’m quite surprised to see West Ham and Southampton up so high in the rankings. Aston Villa are a little skewed and not valuable as much as these other two teams because of injury to Ross Barkley and their 7–2 win over Liverpool. But this is valuable information for us. Last time, we chose players from the best teams alone with the easiest schedules. This time though, we’re hoping that with all this additional scenarios taken into consideration, we can make some better picks.
Another thing to consider too is not to go ahead and just choose the best players from these teams but to work with the schedule to decide when to bring some in. We will limit our scope to the next 8 Gameweeks. Within these 8 Gameweeks, we can think about bringing some players in later if their respective teams have a more difficult opponent at first.
Who to keep and who to let go
Looking at our squad again, we’re faced with the crucial decision of which players are worth keeping on and which players have underperformed. Let’s take a look at the data.
Above us, we have two metrics combined. The Total points that a player has accumulated so far and his total points over the past 4 Gameweeks. The latter being included as a marker for form. It is pretty clear that keeping Heung Min-Son and Harry Kane is a no-brainer. We don’t even need data to tell us that. Ben Chilwell has brought in the most points over the past 4 Gameweeks followed by Diogo Jota. However, we can’t be sure yet if it is worth keeping them as we need to compare these results with that of the schedule. César Azpilicueta and Çaglar Söyüncü are definite sells with their poor return for value.
Another valuable overview is when we compare the average output for our players vs. the players that had brought FPL managers the most value over the past Gameweeks. We can see that we have 3 of those players with the highest returns of the league already in Chilwell, Son and Kane. This is great news that we don’t have to go and purchase such expensive players again at the moment at a higher price. Interestingly, Michael Keane, Daniel Podence and Harvey Barnes produce much less output when compared to the best players. This was not apparent to us during the past Gameweeks as one can easily get carried away with their good performance over one Gameweek and forget about their lack of output over the following Gameweeks. This proves to be a case of why we should have pulled the trigger on these players sooner but just couldn’t see it without the data.
We also have a dilemma when it comes to selling and buying. Most managers sell their cheaper or underperforming players. Selling higher value players may not seem ideal but in actuality, that’s where one can make the most profit for a future transfer in a Gameweek or two. It will also be valuable to buy the players that have the most potential for a price rise in the coming Gameweeks. Given that our team value has plummeted to £99.6, we need all the help we can get in this area.
To recap, we have decided the following so far.
KEEP: Harry Kane | Heung-Min Son | Ben Chilwell
SELL: César Azpilicueta | Çaglar Söyüncü
Final Decisions
Before sealing the fate of the rest of the players, let’s take a look at the schedule opportunities we mentioned before and see when might be good idea to bring in which players.
I have decided to apply a combination of both the best teams to choose from and the schedule. Based on this strategy, we need to have a split within the 15 man roster for diversification.
One error we faced over the past few Gameweeks was in losses - Multiple players from the same team losing causing us massive drops in points. While the past strategy was valuable to gain points as well, we have to be more strict with our decision making, we can only apply it to the teams that have a combination of the best form and results.
(Putting this here again for easier readability)
|6|Aston Villa|
|5|Liverpool|Manchester City|West Ham|Arsenal|Tottenham|
|4|Chelsea|
|3|Southampton|Leicester|
|2|Manchester United|Wolves|
|1|Everton|Leeds United|
When we take a look at both the best options for us in terms of teams and schedule - there are different ways we could approach this.
Approach #1- Two players each from the top 3 teams. One player from each of the rest of the teams.
Approach #2 - Split the schedule into 4 categories. Initial Team Selection/Gameweek 12 Selection/Gameweek 14 Selection/Gameweek 16 Selection. Pick the best players for each category based on opponents.
What we can do is combine these two approaches for a more robust team.
Initial Team Selection - We can see that Leicester, Southampton and Tottenham have an easier start in the next two Gameweeks.
Gameweek 12 Selection - Multiple teams with easier schedules open up such as Aston Villa, Manchester City, West Ham and Manchester United.
Gameweek 14 Selection - Better schedules for Liverpool, West Ham and Arsenal.
Gameweek 16 Selection - Better schedules for Liverpool, West Ham and Arsenal. Similar to the week before so it would be more valuable to have players from these teams.
Earlier, we also mentioned that we want to split our roster into defensive and offensive choices. We want our defense to be more from the most compact of teams and the offense to be from the ones with the most goals.
Defensive Advantage - Tottenham, Manchester City, Chelsea, West Ham & Arsenal.
Offensive Advantage- Chelsea, Liverpool, Tottenham, Southampton and Leicester City.
Taking out Kane, Son and Chilwell (Given that we still see Tottenham and Chelsea as valuable teams), we have 12 remaining spots.
- Diogo Jota and Timo Werner remain valuable picks as well but let’s evaluate all our options before we bring them back in as well.
- Taking out 12 players leaves us with £72.6, an average of £6.05 given to us to play with for our Initial Team Selection. It’s not much but if we can find some value picks that align with both the data and the schedule, we could make it work.
Goalkeeping
First of all, we have some great keepers but given the fact that Kasper Schmeichel hasn’t performed well so far and that the next few Gameweeks look like tough matchups for Rui Patricio, we have decided to not keep either of them. When we take a look at the keepers that have kept atleast one clean sheet and combine that with their overall clean sheet to save percentage, one name sticks out from the rest. Edouard Mendy. The Senegalese shot-stopper has been head and shoulders (literally) above the rest after his recent move from Rennes. Add the fact that he is £0.3 cheaper than both our current keepers, this is a definite buy.
However, Chelsea don’t seem to have an easy schedule when compared to some other teams. Not that they still won’t do well in defense for they have covered that as well. To account for this, we want to bring in someone else with an easier schedule but still an excellent output.
Lukas Fabianski seems to fit this bill. He too is £0.4 cheaper than our current goalkeeper lineup and currently leads FPL in all goalkeepers with 53 points.
BUY: Edouard Mendy (£5.2) | Lukas Fabianksi (£5.1) | Remaining £62.3
Defense
Our defense has cost us a lot of points this season. From sudden drops to 2019 regulars such as Toby Alderweireld, César Azpilicueta to injuries for Virgil Van Dijk and Çaglar Söyüncü, it hasn’t been an easy road. We need a massive shift in defense with a more consistent output.
We have Ben Chilwell. The question is who to add next. Keep in mind that for all these selections, they will be based on both schedule and best team players available.
Here we have the players that take on the most opposing players from the defensive end of the field, but also who rush back to their area when under threat to make great blocks. We’ve often seen players like Walker-Peters and Ayling rush forward.
An interesting pick Gabriel Dos Santos (45 pts.), Héctor Bellerín (43 pts.) or Vladmir Coufal (33 pts.)
Given that Dos Santos is £0.1 cheaper and in better form, we’ll choose him. Coufal has an easier schedule and he is ridiculously cheap at £4.6. He goes into the team.
Above, we have the players that have made the most clearances in addition to the most number of goals. Apart from all the attackers displayed, we have a few defenders as well. The ones that stick out are Dos Santos, Ezri Konsa, Jaanik Vestergaard and Kurt Zouma. Additionally, Southampton have an easier schedule. Kurt Zouma is the more attractive option with 3 goals compared to 2 of Jannik Vestergaard. However, he is more expensive and we have a player from Chelsea already in Ben Chilwell. We want to diversify unlike with our previous strategy. Ezri Konsa has two goals as well and is underrated at £4.7 albeit he is in a weaker team.
BUY: Gabriel Dos Santos (£5.1) | Vladmir Coufal (£4.6) | Ezri Konsa(£4.7) | Jaanik Vestergaard (£4.8) | Remaining £43.1
Midfield
If we look at the most active players on the field from all our current teams alone, we get the following bunch. I was unsure if I wanted to keep Diogo Jota around but he continues to be in great form and in a great team. For £6.9, he is definitely worth keeping. This spread shows far more expensive players such as Salah and Kane. Jack Grealish sticks out with the most amount of shot-creating actions in this group. Aston Villa are off this weekend but they do have some easy games coming up. He is a bit expensive at £7.7 but I do think it’s worth the money. Villa are a risky team to pick but Grealish is worth keeping as a sub atleast when one of the other players have a tougher opponent.
It is tempting to choose Mohammed Salah but given that we have Jota and the fact that they do have a difficult three games coming up in addition to Liverpool’s injury worries, Jota alone maybe sufficient for the moment. Hopefully, we won’t come to regret that decision.
As we said before, Southampton have an easier schedule. This is going to be a risk but I do think it will be quite valuable to have a player who can score two free-kicks in a game. Enter James Ward-Prowse.
Ward-Prowse is not only great at scoring free-kicks. He is also the main corner-kick taker for the Saints.
This gives us more reassurance to select him. While I was here, I noticed another steal of a pick in addition to Jack Grealish. Jarrod Bowen has been a stellar player for West Ham and it was interesting to see that he too was the main corner-kick taker. From this data and given that it is valuable to consider players from West Ham, we will choose him as well.
KEEP: Diogo Jota ( £6.7)
BUY: Jack Grealish (£7.7) | James Ward-Prowse ( £6.1) | Jarrod Bowen(£6.3) | Remaining (£16.3)
Forwards
We’ve already decided to keep Harry Kane which left us with two choices. We could keep Timo Werner and Patrick Bamford and leave it unchanged. Timo Werner has been a valuable addition to the team. Patrick Bamford has dipped a little in form but Leeds are quite an unpredictable team.
However, when we take a look at the schedule again, it does not favor Chelsea or Leeds. Leicester on the other hand have quite an easy schedule.
If you’ve followed along in this series, you would know of my bias for Jamie Vardy. Especially a Vardy and Kane pairing. Vardy is an excellent player and given that Leicester have two easy games before their schedule becomes tougher, it’s worth to see this pairing happen. As we’ve seen in the charts before, Jamie Vardy is a consistent performer given that he’s healthy and Leicester are performing reasonably well. For that reason alone, I have chosen Jamie Vardy.
However, choosing Jamie Vardy leaves us with having to give up Timo Werner and just £6.0 left. There are very few valuable forwards for that price besides Patrick Bamford who also happen to be exactly worth £6.0.
Fate must have it that we go this pairing for this Gameweek. I’m excited to see how this team performs and the tinkering we’ll attempt over the upcoming Gameweeks. I hope you are too.
Last but not the least, here’s our final wildcard team for Gameweek 11.
Let the games begin.
Thank you for taking your time to read through this. If you’d like to follow along in this series, here is a timeline of the FPL data analytics project.
1. Introduction
2. Gameweek 1
3. Gameweek 2
4. Gameweek 3
5. Gameweek 4
6. Gameweek 5–8
— — — — — — — — — — — — — — — — — — — — — — — — —
And there we go! Hope you enjoyed reading this piece as much as I did writing it. Feel free to write down any thoughts or suggestions. I’m always working on improving my analysis and my articles as time goes on so I appreciate all comments! Collaborating is fun so if you’ve got any interesting projects in mind, feel free to reach out to me personally. If you would like to follow me to keep up with updates in this series, follow me here or — — @__tomthomas
— — — — — — — — — — — — — — — — — — — — — — — — —
