Fantasy Premier League 2024/25

Might go totally hardcore next season, contemplating building a points predictor spreadsheet to help me.

Plenty out there already on different websites, all using slightly different formulas and I've looked at them at times. Some are behind paywalls and I'm not paying a thing, screw that. On top of that found myself manually trying to project points one player at a time using stats and the calculator app for bench decisions here and there. That takes ages!

Posted some guy using AI in the last week or two, wouldn't want to do that or have a clue how to, but something to give me shortlists and then make the final decision myself would be the plan.

Definitely a wannabe stats nerd, but without the skills and likely the brains. No idea how to easily grab data, just know that website scraping exists as a concept, and don't have too much experience with spreadsheets, never using them for work or anything like that. Did build one years ago, trying to translate league tables and team form tables into score/results predictions after copy/pasting them with the aim of trying to beat the bookies but it was a slight loser. Was before xG existed in the mainstream, and at a time where I think some people were collecting that data themselves with great success (Brighton's chairman for one).

Don't exactly know what kind of formulas I'd use but do have some vague ideas. Testing? Would probably just wing it and put faith in something completely unproven. No clue how to test.

This might all be pie in the sky as it would take a lot of work and don't 100% know if I can be arsed. Would at least learn some skills and it might even get me back into betting. FPL clean sheets translates to clean sheet betting, FPL goals and assists translate to anytime goalscorer/assist betting etc. Only ever gambled tiny amounts of money anyway, just liked the idea of being right with predictions more than money and the fun of mucking around with numbers. Definitely have an appreciation of that for some reason or other. We'll see.
 
Last edited:
AlexUTD said:
I got 4th in my £365 entry league and 5th in my £255 entry league.

I were in line for way higher payout but then i got like 5-6 red arrows the last GW`s ! just got in the money on both leagues with 1 point thanks to Salah (c)
Click to expand...
You're paying over £600 to enter fantasy footy leagues?! Incredible.

How many are in those 2 leagues and what are 1st places banking?
Has to be the highest payout leagues I've ever seen.

And what sort of points tallies are winning these, down to the last payment spot?
 
Sandikan said:
You're paying over £600 to enter fantasy footy leagues?! Incredible.

How many are in those 2 leagues and what are 1st places banking?
Has to be the highest payout leagues I've ever seen.

And what sort of points tallies are winning these, down to the last payment spot?
Click to expand...
16 in the £365 league and 4 top spots gets money.
24 in the £255 league and top 5 gets money.

winner in both leagues was 3515 (2628 points) OR
 
Some lessons I've learnt that everyone probably knew already..

Don't go mad trying to make your bench as inexpensive as possible in order to get all the money into the starters. It seemed like a great idea at the time but down the road it meant burning transfers I could have banked. That led to a few hits, but more importantly it meant I couldn't really make the changes I needed to when it mattered.

Don't make last minute transfers unless they've been ones you've genuinely been considering all week. It never, ever, ever, ever paid off for me, especially when it meant taking a hit.

Don't wildcard early when you don't need to because you'll start wishing you had it coming up to the midway point and you'll wonder what the hell you were thinking. Everything was looking alright in game week 5, but I thought catching a fixture swing was needed to keep the good scores coming. I got pretty much the same total from 6~14ish as I would have done not playing it, but suffered later on.

Don't pick João Pedro. Ever. Just watch him when he's playing well and enjoy it. He knows when you have him in your team and he punishes you for it.

There's obviously far better lessons all over the thread from people far better than me, but these points are kinda the things I really should have taken an effort to know once I'd committed to a money league. Still, it was good fun and I finished 16th out of 100 knowing nothing. Next year I'm gonna batter the lot of them:devil:
 
AlexUTD said:
16 in the £365 league and 4 top spots gets money.
24 in the £255 league and top 5 gets money.

winner in both leagues was 3515 (2628 points) OR
Click to expand...
The banker of those leagues must earn a decent little wad on the interest :lol:
 
didz said:
Some lessons I've learnt that everyone probably knew already..

Don't go mad trying to make your bench as inexpensive as possible in order to get all the money into the starters. It seemed like a great idea at the time but down the road it meant burning transfers I could have banked. That led to a few hits, but more importantly it meant I couldn't really make the changes I needed to when it mattered.

Don't make last minute transfers unless they've been ones you've genuinely been considering all week. It never, ever, ever, ever paid off for me, especially when it meant taking a hit.

Don't wildcard early when you don't need to because you'll start wishing you had it coming up to the midway point and you'll wonder what the hell you were thinking. Everything was looking alright in game week 5, but I thought catching a fixture swing was needed to keep the good scores coming. I got pretty much the same total from 6~14ish as I would have done not playing it, but suffered later on.

Don't pick João Pedro. Ever. Just watch him when he's playing well and enjoy it. He knows when you have him in your team and he punishes you for it.

There's obviously far better lessons all over the thread from people far better than me, but these points are kinda the things I really should have taken an effort to know once I'd committed to a money league. Still, it was good fun and I finished 16th out of 100 knowing nothing. Next year I'm gonna batter the lot of them:devil:
Click to expand...
Knowing when you need a wildcard is half the battle.
If your lineup is rancid early doors, you might be able to comfortably field 11, but if you don't get on the early risers and identify who will be a steal early, you can find you end up 1-2m behind others fairly quickly.

An early wildcard (first 5 weeks) often got me out of a lot of trouble.

Being able to keep it later of course helps though.


The big problem these days is it's such a big game that there's loads of advice out there, and anyone even vaguely into it starts with a very similar lineup early on, makes very similar transfers and hits a template lineup. Then the chip protocol gets more and more laid out year by year.

The best days were years back when everyone had different ideas week 1.
 
Sandikan said:
Knowing when you need a wildcard is half the battle.
If your lineup is rancid early doors, you might be able to comfortably field 11, but if you don't get on the early risers and identify who will be a steal early, you can find you end up 1-2m behind others fairly quickly.

An early wildcard (first 5 weeks) often got me out of a lot of trouble.

Being able to keep it later of course helps though.


The big problem these days is it's such a big game that there's loads of advice out there, and anyone even vaguely into it starts with a very similar lineup early on, makes very similar transfers and hits a template lineup. Then the chip protocol gets more and more laid out year by year.

The best days were years back when everyone had different ideas week 1.
Click to expand...
I think the wheels really started to fall off just before the halfway mark. I got to OR 9,589 in week 17 and then just got worse and worse, ending only just inside of 300k. Probably panicked somewhere.

Looking at the history, I didn't do well at all when the double GWs started coming in, which makes sense as I had to have the concept of them explained to me twice. Might be I'm just more clever on the drink than when I'm quitting :lol:

Unfortunately I'll never know that simpler time. Never took an interest in FPL before and we'll all only get more information and thusly more homogenised I'd guess. I suppose the really good players just avoid that stuff anyway. So do the really bad ones, probably, but that's the gamble.
 
Thoughts on the new additions?

Ass man chip was decent but too complicated. I would prefer if they adjusted it so you pick a manager and it doesn't take from your budget or team placements, lasts only one week but keep the points system as is to encourage players to not just go to Slot.

The 5 transfers was the biggest issue. It just rewarded those who had a good start and chasers were unlikely to ever catch up. Maybe I'm being cynical after my lowest rank but this was a big issue for me Vs. a rival. With that said, dunno how you fix it.
 
didz said:
I think the wheels really started to fall off just before the halfway mark. I got to OR 9,589 in week 17 and then just got worse and worse, ending only just inside of 300k. Probably panicked somewhere.

Looking at the history, I didn't do well at all when the double GWs started coming in, which makes sense as I had to have the concept of them explained to me twice. Might be I'm just more clever on the drink than when I'm quitting :lol:
Click to expand...

Think me and you were on a similar trajectory at that time.
fc3XA6m.png


Was on Slot AM in 24 & 25 with the 2 double gameweeks and still lost significant ground. That whole period was a mess.

Saka got injured in 17 and I brought Bruno in. He gets sent off next game vs Wolves so I bring in Son. Son misses a penalty at home to Wolves and there's a leak he's being benched for the match after that so I have to bin him off immediately.

Meanwhile I also bought Solanke in 18 for Cunha, think that was when Cunha's fixtures were turning bad. Solanke had been doing bits until then, but then scores 1 goal in 5 weeks as Spurs slump and then he got injured.

Also had Jackson who was doing great early season but suddenly stopped scoring in in 17. 0 goals in his next 8 before injury. Had him that whole time before finally selling in 26.

Hated those few weeks.

Made a triple transfer transfer in 26 who all immediately fired and ended up getting back on track somehow. 11 out of 13 green arrows to end the season.
 
Last edited:
Glad this season is over, by far the worst season in 20 years of playing this game. Have only once finished higher than 500,000 and that was last year when I was just over 600,000. This year I finished 3.7 million and I improved about 500,000 places with a good last week. I have no idea what went wrong, I thought I had decent players every week, but every move I made backfired.
 
Chipper said:
Was on Slot AM in 24 & 25 with the 2 double gameweeks and still lost significant ground. That whole period was a mess.
Click to expand...

That period with Slot AM (along with no Salah until end of Nov or so) were the only two real big mistakes I made. Dropped to 4M at one point and finished 250k and top of most of my MLs. Happy with that but feel like I could've got 100k if that Slot AM decision went my way.
 
Finished with 2483 points, my highest since I started playing in 2007-08 season. My second best overall rank of 172,732. Tailed off in the last month of the season as had comfortably won my ML.
 
Chipper said:
Might go totally hardcore next season, contemplating building a points predictor spreadsheet to help me.

Plenty out there already on different websites, all using slightly different formulas and I've looked at them at times. Some are behind paywalls and I'm not paying a thing, screw that. On top of that found myself manually trying to project points one player at a time using stats and the calculator app for bench decisions here and there. That takes ages!

Posted some guy using AI in the last week or two, wouldn't want to do that or have a clue how to, but something to give me shortlists and then make the final decision myself would be the plan.

Definitely a wannabe stats nerd, but without the skills and likely the brains. No idea how to easily grab data, just know that website scraping exists as a concept, and don't have too much experience with spreadsheets, never using them for work or anything like that. Did build one years ago, trying to translate league tables and team form tables into score/results predictions after copy/pasting them with the aim of trying to beat the bookies but it was a slight loser. Was before xG existed in the mainstream, and at a time where I think some people were collecting that data themselves with great success (Brighton's chairman for one).

Don't exactly know what kind of formulas I'd use but do have some vague ideas. Testing? Would probably just wing it and put faith in something completely unproven. No clue how to test.

This might all be pie in the sky as it would take a lot of work and don't 100% know if I can be arsed. Would at least learn some skills and it might even get me back into betting. FPL clean sheets translates to clean sheet betting, FPL goals and assists translate to anytime goalscorer/assist betting etc. Only ever gambled tiny amounts of money anyway, just liked the idea of being right with predictions more than money and the fun of mucking around with numbers. Definitely have an appreciation of that for some reason or other. We'll see.
Click to expand...
If you enjoy playing FPL, Review is well worth the money. You will get a huge amount of value out of tinkering with the tool. Seems expensive until you realise how many hours you use it in a month for 3 Euros.
 
Dirty Schwein said:
Thoughts on the new additions?

Ass man chip was decent but too complicated. I would prefer if they adjusted it so you pick a manager and it doesn't take from your budget or team placements, lasts only one week but keep the points system as is to encourage players to not just go to Slot.

The 5 transfers was the biggest issue. It just rewarded those who had a good start and chasers were unlikely to ever catch up. Maybe I'm being cynical after my lowest rank but this was a big issue for me Vs. a rival. With that said, dunno how you fix it.
Click to expand...

Ass Man chip is ridiculously OP. Don't like it at all. Also I don't think chips should be lasting 3 weeks either.

Have no problem with the transfer carry over rule (even though I was never really able to take advantage of it). It rewards patient players which I keep trying to be but failing.

Finished on exactly 2400 points (573k). Really poor season for me. Gutted about that 1 point which cost me second place and cash but made so many mistakes this season only got myself to blame. Biggest one was removing Palmer for Saka the week he went mental against Brighton and scored 4. That cost me first place effectively.
 
CassiusClaymore said:
Ass Man chip is ridiculously OP. Don't like it at all. Also I don't think chips should be lasting 3 weeks either.

Have no problem with the transfer carry over rule (even though I was never really able to take advantage of it). It rewards patient players which I keep trying to be but failing.

Finished on exactly 2400 points (573k). Really poor season for me. Gutted about that 1 point which cost me second place and cash but made so many mistakes this season only got myself to blame. Biggest one was removing Palmer for Saka the week he went mental against Brighton and scored 4. That cost me first place effectively.
Click to expand...
Ha, I finished on 2,399 which is a 579.770 rank. Would've gotten a decent amount of money and 3th place in my ML had I just captained Salah or Bowen in this last GW instead of fecking Evanilson, but oh well.

I echo the shit season sentiment, never really felt like I made the right choices and was catching up a lot. Had a clear strategy from GW30 onwards with my WC but then Gabriel immediately getting injured, me picking Gordon instead of Murphy, and my Palace players shitting the bed kinda ruined everything :lol: Salah playing out of his mind with my entire ML on him with TC while I already did Haaland in GW2 also didn't help.

Ultimately you just also kinda need to get lucky that the sensible picks on your chips do have a nice week in that particular moment and my BB and FH didn't play out this season either. Onto the next one.
 
CassiusClaymore said:
Ass Man chip is ridiculously OP. Don't like it at all. Also I don't think chips should be lasting 3 weeks either.

Have no problem with the transfer carry over rule (even though I was never really able to take advantage of it). It rewards patient players which I keep trying to be but failing.

Finished on exactly 2400 points (573k). Really poor season for me. Gutted about that 1 point which cost me second place and cash but made so many mistakes this season only got myself to blame. Biggest one was removing Palmer for Saka the week he went mental against Brighton and scored 4. That cost me first place effectively.
Click to expand...
Yeah definitely that chip was too powerful. I wonder if they do something different every year for mystery chip? But if they bring it back, they need to simplify it and make it less powerful for sure.

Ah yeah we were similar in the end, I finished on 2408 (519k), and missed out winning a ML by 2 points. Would have won it had Martinez not got that red or if that Rogers goal stood.

I was typing my errors/bad luck but there were just too many :lol:

This was my lowest ever finish since I started playing. Have to do much better next year.
 
TMDaines said:
If you enjoy playing FPL, Review is well worth the money. You will get a huge amount of value out of tinkering with the tool. Seems expensive until you realise how many hours you use it in a month for 3 Euros.
Click to expand...

Looked at the free points predictor they have on there sometimes. Limited to 3 weeks only, does the paid one have longer term forecasts? I'm guessing it does. Can't be making transfer decisions purely based on the next 3 games.

Slightly tweaked, improved forumula too perhaps?

There was a guy on reddit who was tracking these, assessing them as if they had a FH every week with only £100m to play with. By the time a lot of actual players had team value of 102-106m they had to stick to the 100m for the purposes of this.


I liked Sort It Out Si as it was looking weeks ahead but they completely stopped updating it around February or March. Suppose my aim was/is to try and build a free replacement of that.

Apparently I beat FPL Review according the first set of results from this page on FPL Optimized. Think you've mentioned that site recently. https://fploptimized.com/highlights.html
 
Last edited:
Started work on that points predictor thing I was talking about knocking up for next season.

It can/will only ever be spreadsheet based I think, copying and pasting data in. Read a tutorial on web scraping and it's beyond my comprehension, same with APIs. Too stupid for all that. Can't see me using it for every player either or I'd have copy in every player's data with each game they play, a monumental task - just going to make it for those players I was already looking at.

First step is working on something to predict goals or xG of games, not sure which should be my target but I've gone xG first, and only when that's sorted to look at players. In the very early stages of the first draft of that. Got past experience of football betting so I'm using an old formula I used to price up matches.

Decided to compare the results my spreadsheet was spitting out to the pre match score projections as posted up by RobtFPL on X where he posts the bookies spread betting median point from SpreadEx. Looking alright so far, but I'm only 4 gameweeks/ 40 matches in, going from most recent first. Tiny sample size and end of season games can be unpredictable anyway.

Initial small sample was similar to the bookies so that's kind of reassuring, and in fact is performing slightly better. Bookies were overestimating home teams by 0.33 a game for these 4 GWs/40 matches, away team by 0.13 compared to the xG that occurred, not goals. For me it's been 0.09 and 0.03 and my trend line is fitting better on scatter plots. Should really start looking at goals too. Not bad so far though, perhaps something to work with.

This is going to be a whole summer long project, doing a bit here and there if seeing it through. Won't be posting about how it's going because let's face it nobody really cares. Just felt a bit giddy about getting started. Having 'fun' learning/relearning to work my way around some pretty basic spreadsheet functions.
 
Last edited:
Chipper said:
Started work on that points predictor thing I was talking about knocking up for next season.

It can/will only ever be spreadsheet based I think, copying and pasting data in. Read a tutorial on web scraping and it's beyond my comprehension, same with APIs. Too stupid for all that. Can't see me using it for every player either or I'd have copy in every player's data with each game they play, a monumental task - just going to make it for those players I was already looking at.

First step is working on something to predict goals or xG of games, not sure which should be my target but I've gone xG first, and only when that's sorted to look at players. In the very early stages of the first draft of that. Got past experience of football betting so I'm using an old formula I used to price up matches.

Decided to compare the results my spreadsheet was spitting out to the pre match score projections as posted up by RobtFPL on X where he posts the bookies spread betting median point from SpreadEx. Looking alright so far, but I'm only 4 gameweeks/ 40 matches in, going from most recent first. Tiny sample size and end of season games can be unpredictable anyway.

Initial small sample was similar to the bookies so that's kind of reassuring, and in fact is performing slightly better. Bookies were overestimating home teams by 0.33 a game for these 4 GWs/40 matches, away team by 0.13 compared to the xG that occurred, not goals. For me it's been 0.09 and 0.03 and my trend line is fitting better on scatter plots. Should really start looking at goals too. Not bad so far though, perhaps something to work with.

This is going to be a whole summer long project, doing a bit here and there if seeing it through. Won't be posting about how it's going because let's face it nobody really cares. Just felt a bit giddy about getting started. Having 'fun' learning/relearning to work my way around some pretty basic spreadsheet functions.
Click to expand...
Post away mate, I find it interesting and also nice seeing the progress on something someone is passionate about.

With that said, I will still ignore data driven FPL moves and follow my own method... YOLO Fam!
 
Dirty Schwein said:
Post away mate, I find it interesting and also nice seeing the progress on something someone is passionate about.

With that said, I will still ignore data driven FPL moves and follow my own method... YOLO Fam!
Click to expand...

Alright, I will post! A lot of the data moves are the same anyway when it comes down to it. You're aware who the good players and teams are, aware of the fixtures, aware of who is in and out of form. The data side of things just attempts to quantify it. The only times they'll truly differ is when someone knows they're taking a big gamble on a player that they just had a good feeling about.

When it comes to data they can sometimes be slow to react to changes in form, that's one downside of it. Cole Palmer was still at or near the top of all the algorithm picks at the end of this past season for example when most people had righfully dropped interest in having him in their team. So even if using data an FPL player or a gambler should never become a slave to it, got to learn when to ignore it in cases like that.

The ultimate data model would in theory adjust faster, eliminating players from being considered if or when their form drops off like Palmer's did but that's a problem too. You could make it very sensitive to recent form rather than overall class, but then it risks recommending too many players that simply had a 2 or 3 week hot streak who are normally not that good. It would be chaos, and it would be like how people sometimes kneejerk players into their team in the same way when they see he scored a random brace.

Finding that sweet spot between form and class is always difficult when using a more data driven approach or less of one.

If going the data route do you look at the whole season's performance, or focus on last 12, 8 or 4 games? Some other number of games? Some sort of hybrid with weighting towards recent games, but how much do you weight in favour of those recent games? Finding that would be the holy grail, and I don't think anyone ever has or even can. Using intuition or feel as to when to dump a player, or be convinced he's someone worth buying always needs to be part of the process too I think.
 
Last edited:
Still very early on this whole process, on the step of trying to put together something to predict how many goals each team will score in a game before mapping the players onto it all. Obviously Newcastle are less likely to score a bucketful away at Arsenal than they are at home to Southampton so that in turn would impact Isak's expected FPL points for the two matches.

Essentailly running with the first part of the idea on this article, from a betting blog but it's transferable to FPL too. It's a well known method that goes back a long time for getting half decent approximations of how a game is likely to go. Difference is I'm using xG instead of goals as it's proven it's more reliable as a predictor of future performance. There have many studies showing that, here's one.

Going to play around with how far back I go, the first article I linked to used the full season and average goals for the season as the base multiplier. Will try out a few different ones but the most I'm going to go back is 12 home or 12 away games.

This video kind of talks about the method too.


Seeing as I've no idea about importing tables from FBref (doesn't seem to work for me?) I'm copying and pasting a lot, but trying to design it where I'll just have to put in the latest week's scores each week once it's all set up for the team side of things.

Overall spreadsheet design a bit of a problem because I don't know what would be best. Right now I'm keeping every team's background data on their own sheet and will try to pull it all together in one place in the end. Just populating all that before starting all the real calcualtions and it looks like this.
MOEw13P.png


Teams looking reasonably consistent across 4, 8 or 12 games which is good. That's Bournemouth's last 12 home and away from 23/24 and I'll start to test against the 24/25 season. Should mean safeish assumptions can be made if things hold up in the same way.

Start of the new season always going to be a problem as teams become better or worse due to transfers, plus there's the 3 newly promoted sides to take into consideration when trying to use a previous season's results to predict a new season's results. Will have to cross that bridge when I start all that which is the next job. Think I'll just use the relegated teams as a proxy for the promoted ones until there's a reason not to. Luckily the last 2 seasons have seen the 3 promoted sides all go back down with similar poor records but who knows if that will be same? Not likely to want to pick many or any players from the promoted sides anyway until they've shown what they can do. The one possible exception being a bench fodder starting defender who is cheap.
 
Last edited:
Really gone to work on this points predictor thing even though I'm stuck on phase 1 - predicting xG of matches. Again, players will be mapped on later.

Now have 9 different systems/methologies I'll be testing, comparing them to the bookies odds and then to the actual xG that happened. Ready to go now.

If I'm close to the bookies it's fine, doesn't matter so much if we're both off as the idea of me being able to better model football matches than they can is a bit fanciful. Bookies only usually have their odds up for 1 match, sometimes 2 but I need to be able to look further in advance as best I can to pick which teams I want to transfer players in from, targeting those high scoring attacks/low conceding projected defences.

Anyway. it's looking like this - the various xG predictions of Arsenal v Wolves from the opening game of last season across the top.

uczeT8O.png


The bookies saw it like this:
FdEdVxw.png


Actual score was 2-0, xG only 1.2 vs 0.5.

Looks like I'm going to be close to what the odds are saying. The 2 methologies with a 4 in their name, "Last 4" and "4 plus" seem off already as they only use 4 games-worth of data so are likely to be very volatile. They're throwing the ones called "merge" off too as that's just the average of the previous 3. The ones with a 12 & 8 in their name looking decent as that's how many games they look at. Might do averages of just those as well which adds 2 or 3 more methologies.

Now to do the actual testing across a full season to find out for sure where I am with all this.
 
Last edited:
Tested the different score prediction models against last season's results. Might assess thew figures in other ways but I think I've got my winner for determining fixture difficulty/which teams I'm going to target when next season starts. Players come next.

No idea if it's ultimately good, but it's the best one of this batch. Used MAE (Mean Average Error) to determine which one I want to go with and I feel like I'm off but it was better than bookies predictions I was comparing it to. Guess it's hard to nail it down on a match-by-match basis while still looking alright overall.

uutUFBJ.png


Across the 760 team xG's for the season bookies estimate off by an average of 0.595, my best method by 0.584.

Season-long stats:

Average actual home and away goals = 1.51-1.42
Average xG (Opta - FBref) = 1.49-1.34
Bookies pre-match odds (middle point of spread on SpreadEx) =1.76-1.43
My best model (Last 12) = 1.62-1.34
 
Now started the preliminary work on mapping player stats onto this points predictor I was working on. First thing was to come up with a method. Going with something like this.

gDChjQ1.png


Column C was non-penalty xG per 90 last season from FbRef.

Column E is an estimate of the team's starting XI. Column F is those player's xG per 90 again with it all added together at the bottom (1.59).

Column G is everything normalised to equal 1.00. You just divide their xG per 90 by the 1.59 in this case. That spits out the estimated xG for each 1.00 xG the team accumulates.

At the top of Column H (1.75) is the team's estimated xG for any given game. This will change depending on how they've been playing/who they're playing against. The work I was doing the other week was to try and produce that number.

The rest of column H is therefore each player's estimated xG should the team register the amount of xG that is shown at the top. It's the 1.75 (or wahtever that number is) multiplied by column G. If Arsenal register 1.75 xG, the estimate is that Saka will get 0.34 of that etc. That's only open play, got a method that should be able to tag penalties on later up my sleeve but that's not for now. Then I need to translate projected xG to projected FPL goal points and do the same for assists blah blah blah.

Need to work on the inputs before really getting to grips with all this because garbage in = garbage out. If I put Tomiyasu in the starting XI it would feck everything up, as I think he had one chance in his few minutes on the pitch which translated to insane xG per 90 of 2.37. Obviously he's not going to maintain that. Could just ignore cases like Tomiyasu, not like I'm going to want him in my squad anyway as he doesn't get enough minutes.

In general still going to want to test different inputs. Whether xG per 90 last season should be the starting point, or whether I should be looking at longer (full career?) or shorter (last 10 games?) timeframes. Going to need a way to adjust things as players may improve or decline/go in and out of form as the season progresses. This is going to be the hard bit.
 
Last edited:
You must log in or register to reply here.