To see less ads Register or Login ----- Daily Fantasy Sports games 18+

FPL's Likely Champion - What Should We Learn?

A Fantasy Football forum for news on fantasy football games run by the Premierleague (FPL).
User avatar
raoul
Dumbledore
Posts: 5031
Joined: 13 Oct 2005, 18:32
FS Record: Tied for OR 1st worldwide in FPL in GW5 in 21/22 (only to then finish 1m), and ranked 1st in TFF late in season 21/22 (eventual finish 95th). Won an old game called Football Fantastic. Best Fanteam finish 117th and cashed £150. Best FPL finish 2.4K.

Re: FPL's Likely Champion - What Should We Learn?

Post by raoul »

Finisher1 wrote: 13 May 2019, 13:29
raoul wrote: 13 May 2019, 13:23 Assuming you want to go for it, picking 3 defenders from one team has a high ceiling. But it stops you picking others and so reduces likelihood of hitting other Clean Sheets.

If defending a good OR, cover makes more sense.
No, always go for the highest expected points.

Let's say team A has 60% likelihood for a clean sheet, while team B has 55% likelihood and team C has 50% likelihood. Then your best defence would include three defenders from team A. Not one defender from each team, no. Three defenders from team A.
But that ignores a whole host of other variables, such as:

- you now have 3 defenders from that team for the next game, limiting your options
- one match badly predicted kills 3 of your players. Pick one CS out of 3, and 3 players from 3 teams outscore 3 players from one team.

Every decision looks clever if it turns out right - need to factor in if it turns out wrong as well.

Finisher1
Dumbledore
Posts: 7159
Joined: 05 Mar 2013, 10:10

Re: FPL's Likely Champion - What Should We Learn?

Post by Finisher1 »

raoul wrote: 14 May 2019, 13:55 - you now have 3 defenders from that team for the next game, limiting your options
Yes, but that's a different discussion. We were only talking about the concept of maximizing points and ignoring 'differentials' and 'covers'. So, in isolation, a defence that includes three defenders from a team that is 60% likely to keep a clean sheet is a better defence than a defence including defenders from teams that are 60%, 55% and 50% likely to keep a clean sheet.

raoul wrote: 14 May 2019, 13:55 - one match badly predicted kills 3 of your players. Pick one CS out of 3, and 3 players from 3 teams outscore 3 players from one team.
Well, if you have three defenders from the best defence, you will win more often than lose. If you have three defenders from a defence that is 60% likely to keep a clean sheet, then you are more likely to get three clean sheets than no clean sheets at all. Some gameweeks you lose big time, but another gameweeks you win big time. The point is you win more often than lose. So, in a long run you win.

raoul wrote: 14 May 2019, 13:55Every decision looks clever if it turns out right - need to factor in if it turns out wrong as well.
You can't factor it in, not with 'differentials' or 'covers' anyway. Good decisions turn out right more often than poor decisions, hence in a long run you get most points by always maximizing your expected points.


...

Trust me raoul, this is how elite managers think. People always pay so much time (sometimes even money) trying to find out how elite managers think, yet the answer is always very obvious just in front of everyone. Elite managers simply play their own game, maximize their expected points and ignore 'differentials' and 'covers'. This is how it is.

User avatar
Sutter Kane
Dumbledore
Posts: 7522
Joined: 05 Aug 2010, 12:13
FS Record: Unknown.

Re: FPL's Likely Champion - What Should We Learn?

Post by Sutter Kane »

Yes expected probabilities over a long stretch and all that, should get you more points because the overall points will tend towards a mean potentially involving big losses and big wins that even out. It's not safer by any means, it's just statistically better. The problem lies in having a limited number of these similar decisions/gameweeks, perhaps not enough for the stats to be borne out. And we often remove players on bad form, just before they seemingly hit form again. :lol:

I like the idea of 3 defenders from the same team (if I don't want attackers) if it's over a long stretch of decent fixtures. If it's one or two fixtures, whilst it may be statistically the right thing to do, your options for future rotation are hampered by having the same team's players. So from an FPL logistical point of view, I'd likely swerve it because you'd want to remove at least two of them at the same time. If it's the last gameweek, too many variables come into play inc money leagues.

User avatar
Sutter Kane
Dumbledore
Posts: 7522
Joined: 05 Aug 2010, 12:13
FS Record: Unknown.

Re: FPL's Likely Champion - What Should We Learn?

Post by Sutter Kane »

This also takes judgment in points probabilities (do we go bookies odds?). Is anyone good enough to separate a 55% from a 60%? If they are, then perhaps that's why they win!

User avatar
Stemania
FISO Jedi Knight
Posts: 20448
Joined: 27 Aug 2006, 11:54
Location: On the Iron Throne of xG, the seat of The Crown Prince of the Stat Perverts. Or if not, in the STC!
FS Record: Best: TFF 321st. FPL 129th. FFS Career HoF Peak 2nd (Live 1st). Ability since lost.

Re: FPL's Likely Champion - What Should We Learn?

Post by Stemania »

I suppose the point is that the example is meant to be illustrative, but even with that in mind it does seem a very rare situation to me. Defenders (in addition to CSs) obviously have different attacking profiles and different prices - so there's so much more to consider if two teams CS prospects are relatively similar than just CSs. Then you've got SK's point on actually being able to accurately separate two teams with similar CS odds, the fact we'd always be looking at a run of fixtures in practice rather than just one game, the fact that transfers are limited and we might want to maintain flexibility etc.

So, whilst I agree in principle with the idea of picking three defs from the same team in the example above (and wholeheartedly with the whole maximising point expectation philosophy), in practice I think this situation would never actually come up. I do think the idea of 'playing your own game' and maximising point expectation are one of the few things that appear common among those that have been successful (winners and consistent players alike), but I wouldn't even be confident to say that it's an overriding factor. It seems to me there's far more to it than just a general approach the likes of Ville, Fabio, Mark Southerns, Jay in his hayday, etc might utilise - player identification is a massive separate category for example, and for me the day-to-day micromanagement and ability to spin plates is absolutely huge; plus planning, patience, and to balance flexibility with eggs-in-one player/pricepoint bracket and many other factors. Imho, they're all very good at some of these. For example, I would tentatively describe Ville as (usually) a very patient expert at micromanagement, whereas I'd be more inclined to say player identification might be one of Mark's strengths. :D

I guess my personal view (from looking at these guys for a while) is that actually none of the top players necessarily excel at everything. Whilst I'm a fully signed-up member of the maximising point expectation society that comes across as a common factor, there's nothing stopping any one of us being better than them at a given aspect - and a rarely emphasised point is that a huge number of FPL players on forum's/site are really excellent managers. On any one-off decision/player evaluation, any poster is capable of making a way better decision than any given well-known names and past winners, but somehow if you aggregate it all together, the top guys relentlessly seem to make slightly fewer bad ones than us mere mortals. :)




Finisher1
Dumbledore
Posts: 7159
Joined: 05 Mar 2013, 10:10

Re: FPL's Likely Champion - What Should We Learn?

Post by Finisher1 »

Stemania wrote: 14 May 2019, 16:51 I suppose the point is that the example is meant to be illustrative, but even with that in mind it does seem a very rare situation to me. Defenders (in addition to CSs) obviously have different attacking profiles and different prices - so there's so much more to consider if two teams CS prospects are relatively similar than just CSs.

...

So, whilst I agree in principle with the idea of picking three defs from the same team in the example above (and wholeheartedly with the whole maximising point expectation philosophy), in practice I think this situation would never actually come up.
Yes, it was an illustrative example and CS likelihood was just an easy way to describe the law of numbers in this case. Of course it's only in isolation, ignoring all other potential aspects of this game. It was simply meant as a response to raoul's idea about picking a defender with an inferior CS likelihood just for the sake of 'cover'. I'm sure you understood this and agree with me :)

User avatar
blahblah
FISO Viscount
Posts: 108503
Joined: 13 Oct 2005, 18:46
Location: .. he thinks that he knows something which he doesn't, whereas I am quite concious of my ignorance.

Re: FPL's Likely Champion - What Should We Learn?

Post by blahblah »

Finisher1 wrote: 14 May 2019, 18:41
Stemania wrote: 14 May 2019, 16:51 I suppose the point is that the example is meant to be illustrative, but even with that in mind it does seem a very rare situation to me. Defenders (in addition to CSs) obviously have different attacking profiles and different prices - so there's so much more to consider if two teams CS prospects are relatively similar than just CSs.

...

So, whilst I agree in principle with the idea of picking three defs from the same team in the example above (and wholeheartedly with the whole maximising point expectation philosophy), in practice I think this situation would never actually come up.
Yes, it was an illustrative example and CS likelihood was just an easy way to describe the law of numbers in this case. Of course it's only in isolation, ignoring all other potential aspects of this game. It was simply meant as a response to raoul's idea about picking a defender with an inferior CS likelihood just for the sake of 'cover'. I'm sure you understood this and agree with me :)
I'm on the side of theoretically yes, but actually very unlikely unless another Chelsea Jose mk I or Graham's Arsenal turn up ditto Queen Kev's Toon.

I'm fully expecting to start next season with 2 Defs plus one from Lpool and Citeh next season.

User avatar
raoul
Dumbledore
Posts: 5031
Joined: 13 Oct 2005, 18:32
FS Record: Tied for OR 1st worldwide in FPL in GW5 in 21/22 (only to then finish 1m), and ranked 1st in TFF late in season 21/22 (eventual finish 95th). Won an old game called Football Fantastic. Best Fanteam finish 117th and cashed £150. Best FPL finish 2.4K.

Re: FPL's Likely Champion - What Should We Learn?

Post by raoul »

Finisher1 wrote: 14 May 2019, 18:41
Stemania wrote: 14 May 2019, 16:51 I suppose the point is that the example is meant to be illustrative, but even with that in mind it does seem a very rare situation to me. Defenders (in addition to CSs) obviously have different attacking profiles and different prices - so there's so much more to consider if two teams CS prospects are relatively similar than just CSs.

...

So, whilst I agree in principle with the idea of picking three defs from the same team in the example above (and wholeheartedly with the whole maximising point expectation philosophy), in practice I think this situation would never actually come up.
Yes, it was an illustrative example and CS likelihood was just an easy way to describe the law of numbers in this case. Of course it's only in isolation, ignoring all other potential aspects of this game. It was simply meant as a response to raoul's idea about picking a defender with an inferior CS likelihood just for the sake of 'cover'. I'm sure you understood this and agree with me :)
It wasn't my suggestion to get someone inferior just as cover!!

I did not propose this idea. I was looking at the 3 defenders from one team proposal and putting forward my opinion.

If you put 3 defenders from one team into your squad, and they are from a team that is predicted clean sheets, then of course you can score a lot of points. That is obvious and it would be madness to argue against it.

However it assumes many things:

- such a team can be predicted
- it is the only team with a clear CS advantage
- the opportunity cost of not being able to own anyone else from that team does not outweigh the advantage of the 3 defenders
- the opportunity cost of the likely money tied up (I presume this will be a top 6 side) in those 3 premium defenders does not hurt elsewhere

Also, with the % you suggested, I would have thought the better proposal is to have 3 defenders each from both the 60% and 55% teams rather than just focussing on the 60% team and play a back 5.

The problem with he expected value argument is that you cannot look at a single proposal in isolation. All possibilities need to be looked at, with all their EV knock on effects, in order to find the best solution.

To be clear, I am absolutely NOT against the idea of 3 defenders from a single team as a principle. What I am against is the proposal that it is a cut and dried proven best solution, when there are so many variables being completely ignored from the argument.

(and as far as I can see, Stemania (an elite player by anyone's criteria) did not even double up defenders much, let alone triple them - other strategies are available that work!!)

Finisher1
Dumbledore
Posts: 7159
Joined: 05 Mar 2013, 10:10

Re: FPL's Likely Champion - What Should We Learn?

Post by Finisher1 »

raoul wrote: 15 May 2019, 10:13 However it assumes many things:

- such a team can be predicted
- it is the only team with a clear CS advantage
- the opportunity cost of not being able to own anyone else from that team does not outweigh the advantage of the 3 defenders
- the opportunity cost of the likely money tied up (I presume this will be a top 6 side) in those 3 premium defenders does not hurt elsewhere

Also, with the % you suggested, I would have thought the better proposal is to have 3 defenders each from both the 60% and 55% teams rather than just focussing on the 60% team and play a back 5.

The problem with he expected value argument is that you cannot look at a single proposal in isolation. All possibilities need to be looked at, with all their EV knock on effects, in order to find the best solution.

To be clear, I am absolutely NOT against the idea of 3 defenders from a single team as a principle. What I am against is the proposal that it is a cut and dried proven best solution, when there are so many variables being completely ignored from the argument.

(and as far as I can see, Stemania (an elite player by anyone's criteria) did not even double up defenders much, let alone triple them - other strategies are available that work!!)
Look, the whole CS likelihood thing was just an illustrative example that I used to demonstrate why 'coverage' is a non-factor in our calculation.

To be very very clear:

"I think team X is the best defence in the league but I already have two defenders from that team and I don't want to have a triple-up because if they concede a goal then my gameweek is ruined" = NON-SENSE!

"I think team X is the best defence in the league but I already have two defenders from that team and I want to have an attacker from that team or the third defender is too expensive or the third defender doesn't have any attacking threat or whatever or whatever or whatever or whatever" = VALID FACTORS!

So is this clear enough now?

User avatar
raoul
Dumbledore
Posts: 5031
Joined: 13 Oct 2005, 18:32
FS Record: Tied for OR 1st worldwide in FPL in GW5 in 21/22 (only to then finish 1m), and ranked 1st in TFF late in season 21/22 (eventual finish 95th). Won an old game called Football Fantastic. Best Fanteam finish 117th and cashed £150. Best FPL finish 2.4K.

Re: FPL's Likely Champion - What Should We Learn?

Post by raoul »

Finisher1 wrote: 15 May 2019, 11:04
raoul wrote: 15 May 2019, 10:13 However it assumes many things:

- such a team can be predicted
- it is the only team with a clear CS advantage
- the opportunity cost of not being able to own anyone else from that team does not outweigh the advantage of the 3 defenders
- the opportunity cost of the likely money tied up (I presume this will be a top 6 side) in those 3 premium defenders does not hurt elsewhere

Also, with the % you suggested, I would have thought the better proposal is to have 3 defenders each from both the 60% and 55% teams rather than just focussing on the 60% team and play a back 5.

The problem with he expected value argument is that you cannot look at a single proposal in isolation. All possibilities need to be looked at, with all their EV knock on effects, in order to find the best solution.

To be clear, I am absolutely NOT against the idea of 3 defenders from a single team as a principle. What I am against is the proposal that it is a cut and dried proven best solution, when there are so many variables being completely ignored from the argument.

(and as far as I can see, Stemania (an elite player by anyone's criteria) did not even double up defenders much, let alone triple them - other strategies are available that work!!)
Look, the whole CS likelihood thing was just an illustrative example that I used to demonstrate why 'coverage' is a non-factor in our calculation.

To be very very clear:

"I think team X is the best defence in the league but I already have two defenders from that team and I don't want to have a triple-up because if they concede a goal then my gameweek is ruined" = NON-SENSE!

"I think team X is the best defence in the league but I already have two defenders from that team and I want to have an attacker from that team or the third defender is too expensive or the third defender doesn't have any attacking threat or whatever or whatever or whatever or whatever" = VALID FACTORS!

So is this clear enough now?
crystal - we are in agreement.

Finisher1
Dumbledore
Posts: 7159
Joined: 05 Mar 2013, 10:10

Re: FPL's Likely Champion - What Should We Learn?

Post by Finisher1 »

raoul wrote: 15 May 2019, 11:17 crystal - we are in agreement.
Perfect :)

User avatar
Sutter Kane
Dumbledore
Posts: 7522
Joined: 05 Aug 2010, 12:13
FS Record: Unknown.

Re: FPL's Likely Champion - What Should We Learn?

Post by Sutter Kane »

People use coverage because we are afraid of a team's potential. In the absence of a conclusive percentage 'play' either way, many are afraid of what might happen even though on average it won't. E.g. Getting BSilva to cover Man C because they could crush a team and he nicks an assist/goal. In reality if the game was played out many times which we don't/won't get the advantage of, BSilva is probably not the player we'd choose, but it's the fear. Finisher, perhaps we need more fear - :D I can't speak for you but my lack of fear cost me Aguero's hatties this season, extremely costly, especially the one in GW25. Playing safer means you are keeping luck swings within boundaries - I played without fear in GW35 so didn't use my BB on my 3 Brighton defenders as I wasn't afraid of them - that destroyed me too because most others did. I try and play pure, without concentrating on ownership but ownership often exists for a reason. Deciding when to go for your own hunch is difficult when there are so many options/variables in play. As mentioned above, juggling all those is key to FPL. Perhaps you require less juggling of permutations with safer play as you don't need to worry about as much.

User avatar
Sutter Kane
Dumbledore
Posts: 7522
Joined: 05 Aug 2010, 12:13
FS Record: Unknown.

Re: FPL's Likely Champion - What Should We Learn?

Post by Sutter Kane »

On the Brighton GW35 point, if we played that scenario out many times, I'm convinced I'd have been on the right end of my decision to BB a week later (though I wasn't expecting Salah 19 points, lack of fear again). But we don't have time to play out enough scenarios. Two or three of these decisions will not converge towards the mean 99% of the time and those decisions to go with your own views instead of covering yourself can be catastrophic. Or they can be outstanding, bur rarely will they be smack on the mean we expect.

Another example, I keep mentioning them but Trent and Laporte especially the latter. It was a calculated risk going without Laporte even though most others had him. Backfired spectacularly.

Finisher1
Dumbledore
Posts: 7159
Joined: 05 Mar 2013, 10:10

Re: FPL's Likely Champion - What Should We Learn?

Post by Finisher1 »

Sutter Kane wrote: 15 May 2019, 11:20 People use coverage because we are afraid of a team's potential. In the absence of a conclusive percentage 'play' either way, many are afraid of what might happen even though on average it won't. E.g. Getting BSilva to cover Man C because they could crush a team and he nicks an assist/goal. In reality if the game was played out many times which we don't/won't get the advantage of, BSilva is probably not the player we'd choose, but it's the fear. Finisher, perhaps we need more fear - :D I can't speak for you but my lack of fear cost me Aguero's hatties this season, extremely costly, especially the one in GW25. Playing safer means you are keeping luck swings within boundaries - I played without fear in GW35 so didn't use my BB on my 3 Brighton defenders as I wasn't afraid of them - that destroyed me too because most others did. I try and play pure, without concentrating on ownership but ownership often exists for a reason. Deciding when to go for your own hunch is difficult when there are so many options/variables in play. As mentioned above, juggling all those is key to FPL. Perhaps you require less juggling of permutations with safer play as you don't need to worry about as much.
No, we both are very good managers so most often we are correct with our assessments.

User avatar
Sutter Kane
Dumbledore
Posts: 7522
Joined: 05 Aug 2010, 12:13
FS Record: Unknown.

Re: FPL's Likely Champion - What Should We Learn?

Post by Sutter Kane »

Let's assume that's true, though I have my doubts because of this point (and these are just arbitrary examples, we've got a few of these):

Either one is good at judging points expectation so I was right to dodge Aguero in GW25 as I thought he'd play once (but of course his first game hatty changed everything). Also I was ok with my decision to avoid Laporte. Those two single decisions cost so many points.

Or regardless of my judgment, I should have played safe in the major decisions and picked my battles elsewhere where the chances of a heavy defeat are minimal. I'd be confident of winning a high % of those because there are so many more of them.

Derailing this thread a little. So what can I learn from the champion and well done to him for literally crushing everyone! Well, pick better captains!

User avatar
Bobby Fetta
Grumpy Old Man
Posts: 1093
Joined: 21 Dec 2017, 21:57

Re: FPL's Likely Champion - What Should We Learn?

Post by Bobby Fetta »

Only just noticed this topic. Thanks to everyone but Ruth in particular for the very interesting analysis (to me at least :roll: ).

I've used the analysis tool I've been using on dod's blog (full explanation on the methodology there) to look at Adam Levy's team. Unsurprisingly his numbers are pretty extraordinary compared to the other teams I've looked at. In summary:
- He owned the big 12 key players (identified by dod) a lot
- But even so he managed to generally exceed their season average points per game scores - i.e. he owned them at the right time
- He also managed to vastly exceed season average scores for much of the rest of his squad

Some other things which I noticed as different to other successful teams I looked at and that I think are important to his success:
- he owned Fraser and Wilson for quite a long time and got good returns from them (I think these two were worthy of being considered key players actually)
- the TAA and Robertson double up in defence for Liverpool effectively for two thirds of the season
- He had 84 and 81 appearances from Liverpool and Man City players respectively. It may seem obvious to focus on the two best teams in the league but these number are unusually high.
- He only owned Salah for 16 GWs, freeing up budget for the rest of his squad
- His average score for defenders (5.8) is incredible
- Laporte :evil:
You do not have the required permissions to view the files attached to this post.

User avatar
Turd Ferguson
FISOhead
Posts: 943
Joined: 16 Aug 2015, 03:32

Re: FPL's Likely Champion - What Should We Learn?

Post by Turd Ferguson »

Bobby Fetta wrote: 15 May 2019, 21:20 - His average score for defenders (5.8) is incredible
Insane. In other words, he averaged a clean sheet every appearance from his defenders. The Liverpool boys really changed the landscape of the season.

View Latest: 1 Day View Your posts
Post Reply

Return to “Fantasy PremierLeague.com (FPL)”