What makes that a slightly hard question to answer is the question of which of those is actually the best goalie in FPL terms.

In theory, the test we should consider is how much difference is there likely to be between 4.5 goalies vs how much benefit can we gain from having a nailed on home fixture each week. If we do that, we see that the worst 4.5 keeper scored ~1.6 points per game more than the worst. Given that, it would seem quality totally trumps home advantage.

Except here's the kicker: Hahneman was the second best last season, Juski was the second worst. Do we believe that will repeat this season? Do we even believe last season was anything other than a fluke?

However, even given that, I'm not sure home advantage means much at all for keepers. The chances of two random keepers both playing away in a given week is 25%, so there will only be 9.5 gameweeks where having a guarenteed home game is an advantage. Of those, you will still pick the away team a few times due to opposition strength; maybe you lose one to injury. That leaves maybe 5 or 6 gameweeks of advantage over a season, or about 1.5 points. Take the better keeper, IMO.

I just don't know who that is



This I have a strong opinion on - doubling up a goalie and a defender doesn't matter. Some weeks you'll double up well, some weeks you'll double up poorly, but what's important is that over the course of the season you average as many points as possible. the only time you'll be playing both defender and goalie is if they have a good opponent, in which case they'll score well over the course of a season. Better to alternate between 4 and 12 points than to get 7 points every week, and by having a goalie and a defender you still have full choice.

That's different from having two defenders from the same club, which restricts your ability to rotate.

Tell us who to pick already!!!

The people want answers...

This is an excellent thread...something that has really helped me clarify my keeper situation. I've been so hung up trying to maximise H/A cheap keepers. Whilst it's not iron clad proof that Xmas' theory will work this season, it has given me the push I needed to just pick two strong cheapies and be done with it.

Great stuff.

It is a very good thread. One thing not mentioned is if home advantage matters more to struggling teams than good ones. When I analysed it a while back I found top team got 60% of their points (league, not fantasy) points, bottom teams 70%. Can't recall if this was on a decent sample though.

I had a quick look, and one season's data isn't enough to give a definitive answer. Here's a graph of FPL home advantage vs team rank (100% means a team scores exactly the the at home as away, 110% means they score 10% more) for one year.



Note there is very little correlation. Overall, there is a bigger home advantage for the lower teams in terms of a percentage, but a bigger home advantage in terms of points for higher teams - since they are scoring higher in general, a 16% home advantege for the better teams is larger than the 24% home advantage for the smaller teams.

Great discussion. The only thing I would caution is although you state what's important is that you average as many points as possible over the course of the season, whilst that is obviously correct and entirely logical, it often is totally impractical as the season unfolds. As we saw last season, severe weather can often play havoc with the fixtures, not to mention numerous injuries, and consequently your ability to mechanically rotate over the course of a season is severely impaired. If the game was to pick a squad of 15 players over the course of a season and never be able to make transfers I would select players exactly as you suggest, but it isn't.

I'm not quite sure what point you are making - can you clarify? How does weather/injuries affect the choice between doubling up players or not? Or are you making a different point?

You can't double up if they're not playing! What you're suggesting is a very long term approach over the course of a whole season, I'm just saying I can't disagree with the logic but I think you'll find it harder to maintain that discipline and implement over 38 weeks, as the season unfolds you will find other factors influencing your decision making process of who to play, factors that are often out of your control.

I don't really get why you think that. I haven't layed out a strategy at any point.

When I say you need to make the decisions that will benefit you over the long term, I don't mean you have to stick to the same one over the entire season. I just mean that you shouldn't worry about short term variations. Some people say "don't double up, because you risk getting nothing" - I'm saying that if you consistantly make the decision with the highest expectation (and that will be a different decision every week), then you'll be better off. It doesn't matter if one week you get 2 points and the next you get 12; that's still better than getting 6 points in both weeks.

You get transfers, so you can change the players at any point. But all the factors we discussed affect you transfer choices: do you bring in the player with 4 home games in the next 5, or the player with slightly better opponents? If you have a struggling goalie and a struggling attacker, which is you prioity to get out first?

^ my new favourite poster

so is that last year's readings? I'd say it's not a lack of correlation but rather a picture of which teams did better at home: Villa, stoke, bolton, wigan, west ham and hull... I know for a fact the west ham stat was because we had such a torrid time away from home and was more due to tactical weakness and awful squad mentality that made away victories so difficult, I guess the same sort of thing with wigan and hull.

Also stoke and bolton make sense, interesting stuff, but villa - who knew?

Ahh hang on, that's not rank by league table, it's rank by my ranking system which is based on FPL points. Rank x isn't necessarily the same team for attack and defence - you can be good at one and bad at the other.

Obviously there's a close correlation between FPL points and league rank, but it's not exact, so don't read the table like that.

I guess what I'm not conveying clearly then is the point that the statistics you are basing your findings on are an average over an entire season, so in order to capture the benefits of your findings you would need to hold them over a long period in order to get that average, and in practice you'll find most people don't, not to mention that although the statistical sample is over 38 games, it's also just 1 season, and I wouldn't get too carried away with findings on a statistical sample of 1.

It's like trading the market, if you told someone to invest in stocks with certain earnings criteria because on average they'd do better over a 1 year period than selecting them randomly, whilst it might be true, in practice people will find a way to screw it up because their emotions get the better of them and the volatility inbetween will shake them out or make them buy at the top and sell at the bottom. So instead of getting that 2 and 12 you get a 2 and then another 2, instead of an average of 6 over the whole period.

Great post. But one thing is missing from this thread. A list of those teams who are actually paired as opposites on the home/away front.

That's probably very true. But if there's one thing I'm not pushing in this thread, it's emotional decisions!

doh, didn't realize that was how they were ordered. still shows just how much home advantage depends on the style of play of the team - would imagine that it'll be the likes of stoke and blackburn this season

At most you can rotate 2 defenders (FPL's 3 defenders rule).

Furthermore, rotatees need to have identical rock-bottom prices, otherwise you are wasting unnecessary funds having money sitting on the bench (Wyld's Rule of Optimum Formations).

Rotation of defenders therefore pays off only if the price of your backline looks something like 4.0 4.0 4.0 4.0 6.0
AND you play 3-5-2 or 3-4-3.

As each M you spend needs to yield around .5 points each week, with your calculated home advantage of 31%, your maximum theoretical gain over the course of the season from "defender rotation" is (4+4)*.31*.5*38 = 47 points.

Useful.

I was wondering where you were.

This is a great thread.

t

That's just the home advantage though; there's also the opposition strength advantage. The two aren't cumulative , as they often offset, but it's bigger than just home advantage.

Also, 0.5pts per million spent is a relatively meaningless stat - as I said in another thread, if every player in FPL was worth £10m more and every manager had an extra £150m to spend, the game would be identical but the basic VFM calc would throw up totally different advice. In truth, the £4m you spend on base defenders is a sunk cost, and it bears no relation to value. The correct way of looking at these things is to look at incremental costs over base, compared to incremental points over a base rate player.

Your formula assumes that a 4.5m player is worth 2.25 points, but clearly there are options to get far more than that. Players at Wolves, Stoke, Blackburn, Birmingham etc average in the high twos or low threes over the course of an entire season.



I agree that rotating expensive player on a regular basis is a bad idea (you should really transfer them out if they have a series of bad fixtures); but the results apply to irregular rotation as well. As we only get one transfer a week for free, there's going to be times this season when your cheap reserve is at home to Blackpool, and you £6m is away somewhere. Them the question is "how difficult does that fixture need to be before I'm better off sitting the expensive player?" By your 0.5/pts per million figure (which I don't necessarily agree with), a couple of million difference is worth a point a week; but the difference between United away and Blackpool at home is likely to be 2+ points, so at the extremes the opposition strength becomes the dominating factor.

Again, I agree that we should generally try to avoid the situation by arranging our transfers accordingly, but often you don't have that choice, whether due to injuries or an isolated poor fixture amongst a series of favourable ones. For example, I have Dawson on my bench tomorrow - that is a wasted £1.5m for week one. But it allows me to take advantage of the favourable 5 or 6 fixtures than follow without using a transfer. Acknowledging the fact that there will be rotation is important, and you should therefore plan to take advantage of it, even if you aren't going for a full on platoon.

This is the nub of it, and I don't think that statement is true.

*duplicate post*

I've said this before, but it's the first time I've seen anyone agree with me.

Wyld - how can it NOT be true?

The players would not be £10m, more they would be 50% more

It's SO true. The conclusion we should draw from that is that the way FPL calculates value (points/price) is waaaaaay off. The FPL value isn't by all means an absolute stat, you can't measure the value of a player without comparing him with different priced players and players of different positions. It's much more complicated than it sounds and all the math in the world won't save you from dealing with unexpected injuries, fixture changes, form shifts etc.

That's because your logic is faulty. When you are standing alone at the busstop, chances are it's because the bus has already left.

*duplicate post*

Actually we aren't alone - stood next to us (amongst others) is every single economist who has ever lived, who all appreciate what a sunk cost is.

(It's a long queue. I really should cycle rather than wait, but I can't be bothered).

Free feel to point out the hole in the logic.

But it's really not a sunk cost at all is it? None of us have had to pay anything to play this game, we don't have any money on the line, the only investment we've made is time and effort, which admittedly could lead to the sunk cost fallacy emerging when we blindly carry on to justify having made all the effort in the first place, but personally I think to attach sunk cost to FPL in the first place is a stretch.

And besides I wouldn't want to align myself with economists, as a trader I can tell you they just blindly extrapolate current trends for the forseeable future which is why every time they speak they adjust their forecasts and hardly any of them have ever predicted a turn.

Grow your own dope - plant an economist.

Obviously we are talking about the in game money, and clearly it's possible for a sunk cost to exist with hypothetical money.