FPL Gameweek 1: And so it has begun..
We analyze our picks for FPL 20/21 Gameweek 1
State of things
Last week, we came to the conclusion of using the FDR strategy — Focus on choosing players from teams with a higher chance of victory. With the higher chance of victory, we also gain valuable points in defense, our points should generally be higher across the field and we allow the possibility of significant points from goals scored as well.
After taking a look at the data we accumulated and analyzed, we came to the conclusion that it would be a good starting strategy to pick players from either Everton, Wolves, Tottenham, Leicester, or Chelsea. Of course, we’re all learning along here as we try different methods (especially me), so there are several things we can observe from our results.
Here is also a link to the Gameweek 1 results by fixtures.
For our 15 man squad, we chose:
3 Players from Wolves (Jota, Traoré, Patricio), 3 Players from Everton (Keane, Richarlison, Doucouré), 3 Players from Tottenham (Kane, Alderweireld, Moura), 3 Players from Leicester (Schmeichel, Barnes, Söyüncü)
Since we can only choose a maximum of 3 players from each team and the choosing of these above players left only 1 spot in a FWD Position and 2 spots in DEF, we chose Azpilicueta from Chelsea and Van Dijk, Firmino from Liverpool for the remaining spots.
Pros of our selections:
- 12 of the 15 players we chose played in a winning team. The only loss coming from Tottenham but this was bound to happen as we chose players from both Tottenham and Everton. The result could have gone the other way around. The reason for choosing both was due to the fixtures over the first 6 Gameweeks. We did sacrifice some points here in Gameweek 1 but we may have a better chance over time with our strategy.
- Our defensive players did quite well with the most points accumulated in the team with 22 points vs 5 from midfielders and 6 from attackers. This presents a hypothesis that we generally do not need to have the best attackers to score points. A strong defense is also valuable. Granted, we are dealing with very limited data at this point. This is something to explore.
Cons of our selections:
- Captain points matter so much — we haven’t yet taken this critical point into consideration. For each week captain points are doubled. Taking a look at the best team of the week below, we can clearly see that if we had chosen Salah as our captain, we would have been able to accumulate over 40 points from a single player! That’s 2 points more than our current team total. As the saying goes, When in doubt, choose Salah.
- We also didn’t consider the free transfers allowed per week. In Fantasy Premier League, the rules state that we are allowed one free transfer per week. We had strategized with the first 6 Gameweeks in mind without any transfers.
This scenario allows for more dynamic team selections. Needless to say, FPL has so many things one must consider and it’s easy to miss something. We can start to consider this into our future team selections. One could also argue whether it’s a better option to keep the same team for 6 Gameweeks to see their progression or to switch players often, this is a study worth considering.
This past week, I came across a wonderful lecture by Joshua Bull of the University of Oxford. Joshua is a post-doc who researches in the area of Mathematical Biology at Oxford but in his free time, enjoys playing Fantasy Football. He attempted to tackle FPL with data as well last year and goes into great detail over all the strategies he attempted during the 19/20 season. He also discusses the struggles of trying to create a magic formula for a winning team each Gameweek. Something we just witnessed above with our own team. It was a great listen and I definitely recommend it for anyone interested in this. If you’re reading this, I’m guessing you are. Oh, and it should be mentioned that Joshua also won the 19/20 Fantasy Premier League season. So don’t miss it!
So why these players?
Using the data from FBRef, we can see that there are a large number (Well over 150+) of varied stats we can choose from. It can get quite difficult to analyze each and every stat. So as a base starter, we will choose a few stats that we can consider to be key separators to the rest when it comes to choosing the most effective players.
- npxGxA90 - Non-penalty Expected Goals & Assists per 90. This is another key stat that reveals the most creative and productive player in a team.
- SCA90 - Shot-creating actions per 90. This stat reveals the most creative player in the team. The true playmaker of the team. This stat is valuable when trying to find key players that might otherwise go unnoticed with npxGxA90. These are those players that are next in line to add points after the most effective players.
- PrgDistPass - Progressive Distance Passing. These are the players that have passed the longest and the most towards the opponent’s goal. This could be a useful stat to find the most attacking players, possibly from defense. The effectiveness of this stat is still subjective but we did use it for the past Gameweek.
- KP - Key Passes. This stat helps us find the most effective passers.
- AerialAcc - Aerial Duels Won % - These players have the highest chance of scoring from set-piece plays due to their great heading ability.
- Clr - Clearances. The number of times a player cleared a ball out of their own goal-end. Possibly a valuable stat for defenders in FPL Value.
- PassesCompAcc - Pass Completion Accuracy. These are the most accurate passers in the game. The passes of these players have a higher probability of reaching their team-mates.
- Ast90 - Assists per 90. The most amount of Assists over 90 minutes by any player.
- Gls90 — Goals per 90. The most amount of Goals over 90 minutes by any player.
- SoT - Shots on Target. The players who score higher here are the most clinical shooters.
Using these stats above, let’s take a look at just the players from the teams we chose. We excluded new signings, they have a possibility to start but also a highly unlikely one to not in a new team especially in Gameweek 1.
Defenders
One of the key areas in our team selection was in defense. As we’re able to see in hindsight the number of points we accumulated here. A key criterion to consider was the number of matches each player played last season. So as a base requirement, we are only highlighting those players that have played the most amount of games for their team. Barring injuries, this allows for a solid chance that they will feature in most games this season as well.
Taking a look at the # of clearances a player made last season and combining it with the most capable aerial players gives you an interesting stat. This combined selection allows us to consider defenders that have a higher chance of scoring both from set-pieces while also being strong in the back.
No doubt, Player of the Year Van Dijk stands out. Last season, he scored 178 FPL points and even more the year before! (208)
So it was a no-brainer to add him to our team. We also see that some other solid additions would have been Evans, Saiss, Söyüncü, Keane, and Alderweireld.
Evans would have been a solid choice but he was injured for Gameweek 1. So we added all the other players except Saiss to our list. We will regret not choosing Saiss as he scored 15 points this past Gameweek. Part of it is luck. But looking at this data, It could have been any one of these players. Another hypothesis we could consider is opposition’s defensive ability by pairing these teams against each other. This is an area worth exploring for any of you out there.
Another player you might have noticed in the list is César Azpilicueta. Looking at the key passes players made last season combined with the progressive distance they covered, we see these following players stick out. Considering both Alexander-Arnold (£7.5) and Robertson (£7.0) cost the most among all the defenders, we were left with 2 options, Digne and Azpilicueta. It was a toss-up between the two. However, Chelsea as a team seemed more formidable to Everton (who faced Tottenham in Gameweek 1), so we chose Azpilicueta. In the end, Azpilicueta didn’t start and Digne scored a goal. Cmon Azpi! We failed to consider the Right back competition at Chelsea with Reece James. Azpilicueta played all last season (36 Matches) until an injury forced him out at the end of the season. A parameter that was missed. Albeit another reason for Digne’s omission was the selection of Keane who cost only £5 (£1 less) and did fairly well himself. Alas, we will tinker with the squad and consider more metrics as time progresses.
Mids and Forwards
Take a look at this following spread where we combine the most creative attackers with the most effective attackers. You’ll see some familiar expensive faces. Salah (£12), Mane (£12), Firmino (£9.5) but some valuable cheap options too in Jimenez (£8.5), Barnes (£7), Jota (£6.5), Maddison (£7.0), and Richarlison (£8.0). Instead of going for the most expensive options, we chose a strategy of continuing to choose from our preferred teams. Jota seems like an insanely good value-for-money considering how he ranks in this spread. Unfortunately, he seems to be down the pecking order at Wolves and only came on as a substitute. Salah should have been a no brainer with his 20 points but we also have to take into consideration that if we had picked Mané, we would have scored just 2 points this past Gameweek. So more expensive isn’t always the better option.
Barnes played exceptionally well in his first Gameweek. He had 5 shots overall (1st among all LEI players) and 2 Shots-on-target as well. It was only a pity he couldn’t join in on the fun with Vardy and Castagne.
Also Richarlison did this over the weekend.
Additionally, taking a look at the Pass Completion Accuracy tied into the total Assists per 90, we can see that in addition to the popular choices, Traoré is a good cheap buy and a good addition from Wolves. We’re hoping he produces more over the coming Gameweeks. We also found another cheap but great pick in Lucas Moura who believe-it-or-not compared fairly well to Salah himself on SoT value.
Also finally, you can never go wrong with Harry Kane who produced well in most comparisons we analyzed. The only reason you don’t see him in the previous comparison was due to him being out with injury (Spreads included only players with 30+ starts). Below, we see the list of all players (20+ starts or more) that are both incredibly effective in the air and score the most amount of goals. Interesting to note that Jamie Vardy is almost on par with Harry Kane despite the loss of height. The reason to choose Kane over Vardy was his age. Being younger, Kane seemed a more valuable pick to produce results again. However, Vardy proved us wrong with 2 goals over the weekend.
Note: We didn’t analyze goalkeepers from the 6 teams. Arrizabalaga, Pickford, and Lloris have faced criticism over their shot-stopping abilities over the past season. Which left us with two solid picks in Schmeichel and Patricio and we have chosen both.
Final Thoughts and prep for Gameweek 2
So the above thought-process created the 15 man squad we ended up with. Not a great start but we’re hoping that results will pick up over the next 5 Gameweeks as probabilities go but we’re dealing with uncertain events here when it comes to football. So hopefully with a little bit of luck and math at our side, we will do better!
Here are a few other random thoughts/areas of improving strategy. Tinkering is always fun for us data geeks.
- Player values increase and decrease over Gameweeks based on performance. One strategy is to transfer in a cheap player and hope his value increases. This allows us to purchase a more expensive and valuable player with the added budget.
- We are allowed one free transfer per week. Strategy 1~ Transfer in a new player in form each week. Strategy 2 ~ Wait 2 weeks to transfer two players which will allow for multiple positions to be changed.
- Player transfers ~ If you watched Joshua Law’s talk above, you’ll notice he mentions the dilemmas of player transfers. Do we transfer in a player based on form vs fixture? Do we transfer out an underperforming player or a cheap player?
- Is it better to choose a Triple Captain from Defense or Attack?
- What about Opposition Analysis? Do units matter over selection of single players? Consider Liverpool’s front 3 vs Norwich’s back 4. Should we choose players from the same team increasing opportunity for points with the added risk of losing more together as well?
— — — — — — — — — — — — — — — — — — — — — — — — —
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
