Over analysis of FPL defence from last season

[Mav3rick]

WARNING: This is a long post and only of interest to stats perverts. If you are not interested in over-analysis of your FPL team (and specifically the defence) then I suggest you skip this thread! I don't think that there is any new knowledge here, so you may find that you read the whole thing and find it a waste of your time. Consider yourself warned.

I wanted to do a little bit of analysis on the defensive strategies that could have been utilised in FPL over the last season and how they stacked up against eachother. Note that the conclusions drawn here are not a guide for next season since everything could be different if there are one or two keenly priced defenders, a late transfer for a cheap defender or a great fixture list for some respectable defensive teams. I guess you could call this more of a list of my own thoughts on my FPL defence last season.

There are lots of ways to structure your defence in FPL so to simplify things enough that we can actually analyse anything, we need to set a few restrictions and assumptions before we begin. Starting at the base of the formation, we have the goalkeeper selection.

There is usually an argument somewhere along the line that you can select a goalkeeper from say Chelsea as a CRD and thus save yourself some money. While this is true, I still believe that the overall best use of funds is in cheap keepers.

I can illustrate this by looking at the keeper points gained by Chelsea (159@6.0m) compared to the average of the best three options at <=5.0 (Fabianski, Heaton and Forster 148@5.0). Now, take the extra 1.0 spent on the keeper and add it to one of your defenders and it’s the difference between Terry and Azpilcueta (40 point gain), Terry and Fonte (35 point gain), Fonte and the best 4.5 defenders (30 point gain) or some combination of extra attacking points from investing the extra cash in the front seven. So you see that the roughly 10 point gain is easily cancelled out by the 1.0 invested elsewhere. This looks to be because cheap keepers more closely match their premium rivals, making up for a lack of clean sheets with save points and bonus points.

I’m going to assume that most FPL managers agree with me and would be picking a cheap keeper combination and ignore keepers when thinking of the ideal defensive strategies. I am also going to assume that nobody really thinks the extreme formations are a good idea, so both 5 cheap defenders or 3 premium defenders are ignored in the discussion.

Also, I will assume that everyone will pick at least one premium defender (6.0-7.0) to build their defence around. So we are left with 4 defenders to budget and allocate.

Given the 4 spaces and the requirement to play at least two more defenders, we are left with the following potential combinations to analyse:

A. Two rotating pairs of cheap defenders, either based on fixture strength or H/A rotation (e.g. Newcastle/Sunderland). Cost 18.0.
B. One premium defender + 3 cheap defenders. Cost 18.5-19.0
C. One “value pick” + 3 cheap defenders. Cost 17.0-18.0.

It’s my belief that (3) is the optimal solution and that is indeed how I started last season, with Debuchy as my “value” pick and starting with Wisdom, Hutton and Duff (plus 0.5 in the bank for an upgrade when a 4.5 pick took my eye). Debuchy lasted until GW4 when he became Clyne (this is before Debuchy got injured but enough time to see that the Saints full backs were doing better than expected). I retained this structure with various combinations of cheap defenders filling the 3 spaces and combinations of Chambers, Bertrand and Clyne filling the “value” spot and Terry, Baines and Ivanovic filling the premium spot over the season.

If I’d just left the value and premium spots alone, I’d have got 177 and 142 for Terry and Clyne respectively, which is possibly a lesson for me to just leave those two positions alone.

For the purposes of further comparison, we need to construct some profiles for defenders of different values and strategies. We can start with the “value” pick being somewhere around 5.0-5.5 and scoring about 140. Two seasons ago we had Coleman (who went to 180), last year it was Clyne/Bertrand, if there’s someone in that range next year they need to be identified and selected as early as possible.

The premium pick may as well be profiled as costing 6.5-7.0 and is a 170 point player.

Then we need to collate some data for rotation pairs – H/A pairs are easy to work with because we can see the H/A scores over at ATFPL:

FPL Stats from 201415

Combinations would have scored like so (note I’ve had to extrapolate and round off some of these scores due to lack of minutes etc of players):

Stoke/Swansea - 110 points
Villa/WBA – 100 points
New/Sun – 100 points
Bur/Hul – 110 points

Those numbers are REALLY low (average 105 points) and compare awfully to a Clyne/4.0 combination. I think it proves one thing that is already likely self-evident, which is that you can’t just rely on teams being at home, they have to be decent defensively and have nice fixtures or you’re just wasting money. There is some reason for increased expectations from the New/Sun pairing next season though, and I would keep an eye on that as a potential good pairing if the prices are right.

Favourable fixture strength combinations are harder to work with because they change through the season. You need to be on top of the fixtures and be planning your transfers to maintain a healthy run of games. For that reason it is hard to put a points metric on that position. To come up with some sort of number for this I will refer to the suggested combinations from FPL Wizard and then extrapolate the PPG (up to the Jan wildcard) and then convert to a 38 game total points score for comparison with the numbers for straight H/A rotation pairs:

Villa (Hutton)/Stoke (Pieters) – 115 points
Villa (Hutton)/Burnley (Shackell) – 125 points
Leicester (Morgan)/Swansea (Taylor) – 145 points
Newcastle (Coloccini) /WBA (Wisdom) – 115 points

(Note that I’ve taken a few liberties with the actual rotation and the numbers should be viewed as best case).

Average out those pairings gives you 125 points, which is much more respectable than the 105 for the H/A pairings, especially when remembering that the price of the rotation pairs (H/A vs Fixture Strength) are the same.

Lastly, we need to consider the profile of the 1-from-3 cheap defenders strategy. In the STC thread we discussed the options in detail before the start of last season. In general it came down to WBA (Wisdom), Villa (Hutton) and one 4.5 defender who could have been from Stoke, Newcastle, Swansea (as indicated by the FPL Wizard app) or from Burnley as I went for in the end.

Taking my own season, the one-from-three rotation up to GW21 yielded an average of 3.68ppg, which is 140 points over the whole season (this included players such as Wisdom, Hutton, Duff, Bartley and Yun at different times). This does seem high to me under normal circumstances (it may be normal, but I’d like to quantify it with the fact that Wisdom and Hutton’s fixtures lined up very well, and next season there may be no such opportunity).

So, now we can summarise the defender profiles as follows, with rounding done for ease of maths (plus these aren’t accurate figures so rounding removes any implied preciseness):

Premium: 175 points @ 6.5-7.0m
Value: 140 points @ 5-5.5m
H/A Pair: 105 points @ 4.5x2
Fixture Strength Pair: 125 @ 4.5x2
One-From-Three: 140 @4.5x1 + 4.0x2

Now we can finally start slotting those numbers into the templates A, B and C above:

A. Two rotating pairs of cheap defenders, either based on fixture strength or H/A rotation (e.g. Newcastle/Sunderland). Cost 18.0.
4x defenders at 4.5 each, using fixture strength as it’s better than H/A rotation – 125 points from each pair, so 250 points in total from your four defenders at a cost of 18.0.

B. One premium defender + 3 cheap defenders. Cost 18.5
The premium pick is 175, 3 cheap defenders is 140 so 315 points at a cost of 18.5.

C. One “value pick” + 3 cheap defenders. Cost 17.0-18.0 (depending on your “value” pick).
The value pick is 140, the one from three is also 140, so 280 at 17.0.

Conclusion (finally!)

On the numbers produced, to me we can see that B and C are better than A, on last season’s start prices anyway. In B, the extra 1.5m invested has yielded an additional 35 points. The rule of thumb is 1ppg for every 1.5m invested so the numbers are pretty much identical, although C is slightly ahead on that basis. In theory I would be happy to find a 1.5 upgrade in midfield/attack that can make up a 35 point gap.

What I do think is obvious is that both B and C are superior to A due to the extra boost that the 1-from-3 has achieved. Whether or not that is a sustainable boost I’m not sure because the fixtures and available players really did line up well last season. The first thing I will look at though will be if there are three defenders for a combined cost of 12.5 at most (and 12.0 preferably) that can match Wisdom, Hutton and Duff’s performance over the first 21GWs of last season.

If there is no such boost available, then it really becomes a decision between value pick plus cheapies or fixture strength based rotations (if option B lost 30 points for the same cost, then it would be obvious that the extra 1.5 invested elsewhere could make up the remaining difference). Since there are unlikely to be two obvious 5.0 candidates next year (although Man Utd defence, I’m looking hopefully your way for one of them) it seems that the might be some middle ground that we need to look for, which is essentially a value pick plus three cheapies in some combination, with an extra 0.5 going in as necessary.

Anyway, nothing can be decided on until we see the prices for next season and any late transfers since the entire makeup of your defence can swap based on a cheap defender moving clubs, but I hope this very long post has been useful to someone.

[Stemania]

Lovely jubbly - I may be nursing a stat semi. Tbh I think we were probably a bit fortunate with the number of good 4.0m defenders this year. In fact, if we knew beforehand probably then Option D: Two value picks and two 4.0, 4.0; would likely have been optimum this year, but as you say having two such reliable 5.0 'value' defenders of the quality of the Southampton two is pretty unusual so couldn't really be planned for in advance.

[Taff Murray]

My head hurts.

Great work though with a lot of food for many thoughts between now and the start.

[Billy Bongo]

I'm looking at two rotating home away pairs in defence
Swans Lei Williams Morgan 4.5s and club caps
Newcastle Sunderland Collo Oshea both 4.5 and club captains ( I think anyway for Oshea )

[Mav3rick]

I'm looking at two rotating home away pairs in defence
Swans Lei Williams Morgan 4.5s and club caps
Newcastle Sunderland Collo Oshea both 4.5 and club captains ( I think anyway for Oshea )

I think Williams is 5.0. To be honest unless we find some 4.0s later on my initial thought is to lineup with a starting every game 5.0 and a 3-way at 4.5;

premium, 5.0, 4.5, 4.5, 4.5

Last season I started 6.5, 5.5, 4.0x3 - if I start with Mertesacker in my premium slot this year then 5.5, 5.0, 3x4.5 is the same budget...

[Billy Bongo]

Yes my thoughts as well

But Cech and Cahill at 12m is enticing

[Notned]

I'm looking at two rotating home away pairs in defence
Swans Lei Williams Morgan 4.5s and club caps
Newcastle Sunderland Collo Oshea both 4.5 and club captains ( I think anyway for Oshea )

I think Williams is 5.0. To be honest unless we find some 4.0s later on my initial thought is to lineup with a starting every game 5.0 and a 3-way at 4.5;

premium, 5.0, 4.5, 4.5, 4.5

Last season I started 6.5, 5.5, 4.0x3 - if I start with Mertesacker in my premium slot this year then 5.5, 5.0, 3x4.5 is the same budget...

Find myself agreeing with you a lot of late, but I am thinking along the same lines. Only difference is I prefer Kosc over Mert for the extra .5m, which will be covered by potentailly having 8.5m keepers.

To Billy- of the Sunderland 4.5m picks, I would actually favour Coates I think. Quickly became a favourite of theirs in the last 6 games last season, and I'm sure I read he was well near the top of the rankings for goal attempts for a defender, over the same period...

[Mav3rick]

But Cech and Cahill at 12m is enticing

I need to think about Cahill more (not in that way) and the possibilities of the super premium Terry/Ivanovic.