FISO

zipnolan wrote: ↑26 Jul 2022, 15:10 Incidentally (bit of a digression), I used that FFH rate-your-team tool yesterday. It's all over the place. I watched a video with "Bigmanbakar" getting a 97% rating for his squad. I then put exactly the same squad into the tool (just changing the order I put in the players) and it came back with 88%! I mean, I'm sure I'm a worse manager than he is, but I am not that bad at data entry....

How weird! Was that is three premium draft I saw on Twitter? If so I find it hard to see it getting rated 97%. Far too unbalanced. I used to have access to the Hub mainly to use the fixture ticker on their app as it was easy to access on the go. I no longer have it.

97% of what?

Neath boy wrote: ↑26 Jul 2022, 15:21

zipnolan wrote: ↑26 Jul 2022, 15:10 Incidentally (bit of a digression), I used that FFH rate-your-team tool yesterday. It's all over the place. I watched a video with "Bigmanbakar" getting a 97% rating for his squad. I then put exactly the same squad into the tool (just changing the order I put in the players) and it came back with 88%! I mean, I'm sure I'm a worse manager than he is, but I am not that bad at data entry....

How weird! Was that is three premium draft I saw on Twitter? If so I find it hard to see it getting rated 97%. Far too unbalanced. I used to have access to the Hub mainly to use the fixture ticker on their app as it was easy to access on the go. I no longer have it.

It was this: not a threemium.

https://www.youtube.com/watch?v=XE_ngHQ9r24

As for 97% of what - who knows? Of perfection, presumably. Apparently not to do with GW1 as some teams are projected high in points, but relatively low in percentage.

I like stats like that

I've never done any linear optimisation, but it would be interesting to see.
Maximise expected points over first 5 GWs.
Constraints: n = 11, GK = 1, DEF between [3,5], MID between [2,5], FWD between [1,3], price <= 83 (assuming you want 17 for the bench).

Anyone want to give it a crack? I tried it on Excel but I hate that software with a passion (it kept recommending me a team of 11 Alexander-Arnolds )

We tried it in the STC preseason maybe 3 years ago. There's an excel file with a solver to basically pack the boxes.

They usually just end up throwing out the things you'd expect though - lots of defenders, cheap strikers.

Mav3rick wrote: ↑29 Jul 2022, 08:56 We tried it in the STC preseason maybe 3 years ago. There's an excel file with a solver to basically pack the boxes.

They usually just end up throwing out the things you'd expect though - lots of defenders, cheap strikers.

Yeah. This is basically the trend with any FPL analysis - 5-4-1 with premium defence and a budget striker.
Does it pay off though? How often do the top 1000 end up with that kind of lineup?

My guess would be rarely as budget busting enablers always emerge (Bowen last year for example) plus players get injured and you can't replace them with like for like value.

I think top 1k is more dependent on captaincy luck if I'm honest.

Mav3rick wrote: ↑29 Jul 2022, 12:05 My guess would be rarely as budget busting enablers always emerge (Bowen last year for example) plus players get injured and you can't replace them with like for like value.

I think top 1k is more dependent on captaincy luck if I'm honest.

Yes captaincy luck and timing. The latter is more possible (to be lucky) if you have more transfer budget to play with in attacking positions - 5 heavy hitters at the back is less flexible in this regard, which is why teams often move away from huge money at the back...at some point. (i.e. hopping between Podence and Bailey probably won't cut the mustard)

Perhaps this season, with the wing back power, and their prices, things are different and we should ignore Neco and spend BIG BIG BIG!

Charade wrote: ↑29 Jul 2022, 10:28

Mav3rick wrote: ↑29 Jul 2022, 08:56 We tried it in the STC preseason maybe 3 years ago. There's an excel file with a solver to basically pack the boxes.

They usually just end up throwing out the things you'd expect though - lots of defenders, cheap strikers.

Yeah. This is basically the trend with any FPL analysis - 5-4-1 with premium defence and a budget striker.
Does it pay off though? How often do the top 1000 end up with that kind of lineup?

I do think that its very few people that can follow a long term plan when theres nothing ”on stake”. People are more willing to take risks. E.g. Move away from original plan because of fomo, good ideas…
Especially with 5 at the back, looks more like a grinding stratergy then getting much points in one week.
What is it called.. Skin in the game? Tell someone to bet 15 000 on and if your not following the plan you Will loose .

Who knows, maybe none good fpl player been true to that tactic for a whole season - and if they were they would won?

I bet the top 1000 players in season 21/22 had more money in defence then top 1000 players in season 17/18. So then its still the trend to put more money in defence?

blahblah wrote: ↑23 Jul 2022, 16:37 Yes, my base prices need 1 adding, or a similar thang

1+(Price-Position Base Price) or lower all the Base Prices by 0.5 etc

There are pages on this from probably more years ago than I want to remember..

Personally, I don't see the 2 vs 1 Appearance Point as long as it is App not Minutes divided by 90 as there has to be a final stage of guessing whether they go Higher or Lower next season. Arguably this could be used during the latter half of a season, if the Higher or Lower can be applied....

I think the last paragraph is very similar to the Base price argument, ie you have to pay 4.5m for a Defender, so what does the extra get you; and your point that they all get 2 points so what else do they add?

Goodness, I vaguely remember getting embroiled in such an exercise a few years back.

I do think there is gold to be found in these hills. What randomnly occurs to me now is you should calculate Real Value being (Points * Base Price) / Price

Then do the linear optimisation based on that Real Value.

snout wrote: ↑02 Aug 2022, 22:25

blahblah wrote: ↑23 Jul 2022, 16:37 Yes, my base prices need 1 adding, or a similar thang

1+(Price-Position Base Price) or lower all the Base Prices by 0.5 etc

There are pages on this from probably more years ago than I want to remember..

Personally, I don't see the 2 vs 1 Appearance Point as long as it is App not Minutes divided by 90 as there has to be a final stage of guessing whether they go Higher or Lower next season. Arguably this could be used during the latter half of a season, if the Higher or Lower can be applied....

I think the last paragraph is very similar to the Base price argument, ie you have to pay 4.5m for a Defender, so what does the extra get you; and your point that they all get 2 points so what else do they add?

Goodness, I vaguely remember getting embroiled in such an exercise a few years back.

I do think there is gold to be found in these hills. What randomnly occurs to me now is you should calculate Real Value being (Points * Base Price) / Price

Then do the linear optimisation based on that Real Value.

Yep, I'm sure you were there

I think the thread made slightly more sense than whether the plane took off

If someone digs it out, please post something there so I can see it on the Latest/ My Posts

zipnolan wrote: ↑26 Jul 2022, 15:32

Neath boy wrote: ↑26 Jul 2022, 15:21

zipnolan wrote: ↑26 Jul 2022, 15:10 Incidentally (bit of a digression), I used that FFH rate-your-team tool yesterday. It's all over the place. I watched a video with "Bigmanbakar" getting a 97% rating for his squad. I then put exactly the same squad into the tool (just changing the order I put in the players) and it came back with 88%! I mean, I'm sure I'm a worse manager than he is, but I am not that bad at data entry....

How weird! Was that is three premium draft I saw on Twitter? If so I find it hard to see it getting rated 97%. Far too unbalanced. I used to have access to the Hub mainly to use the fixture ticker on their app as it was easy to access on the go. I no longer have it.

It was this: not a threemium.

https://www.youtube.com/watch?v=XE_ngHQ9r24

As for 97% of what - who knows? Of perfection, presumably. Apparently not to do with GW1 as some teams are projected high in points, but relatively low in percentage.

Sorry, only just spotted this, think I can answer these queries.

The issue with the BigManBakar team, and with any other team you see rated on the FFH youtube channel, is not down to your data entry skills, it is one of timing. Because the prediction metrics are constantly being fine-tuned, as more data emerges from pre-season, the percentage ratings of any one team will change over time. What FFH did (which was a bit naughty IMO) was to pre-record that video with Bakar but only put it out as a stream over a week later. So his team for example IIRC included both Chilwell and James, whereas by the time the video was streamed, Chelsea’s pre-season form meant few were still on that double-up. But of course he got the high % rating because at the time it was recorded, the predicted points from those two were a lot higher than they were when it was streamed.

The team percentage rating is a reflection of the points predicted for this team over the coming weeks (they don’t say exactly how many), with a weighting for the earlier weeks. Since the launch, because that caused some confusion, they now show two percentage ratings - one for the team, as above, and one (GW%) that is based purely on the predicted points from that team in the current gameweek.

FFS have something similar, though not as slick, where you can enter your team to see how many points it predicts you will get in the coming weeks.

Thanks, eventually I came to the same conclusion, though the variance was enormous and Bakar will surely not be going with that team any more. I imagine he's published a dozen or so different "team reveals" since. But it's what sells papers, eh? Discontent providers....

I do find the fact that you can tab through each GW on the tool and see how each week is rated. The entropic decline of an FPL squad is cruel to behold.