Monday, August 13, 2012

Are four gloves better than two?

One of the first ideas I wrote about on the blog was the idea of utilizing two budget 'keepers to deliver the same value as one elite player. As a quick reminder, the basic premise of the strategy is that:

  • A player's opponent has a strong influence on his weekly success and thus if you select players whose fixtures 'gel' well together, you can maximise the number of games played against weaker opponents.
  • If you select an elite keeper, you still need to spend at least 4.0m on your second keeper who will likely never see the light of day. Hence, to enjoy the right to play Joe Hart every week will cost you at least 11.0m where as playing a pair of average players in the best circumstances will set you back 10.0m or even less.
It's tricky to predict how teams will perform this year using only prior year statistics, but by and large we can get a decent idea of what to expect as teams rarely change dramatically on a yearly basis. The calculation for the promoted teams is more complicated, as previously discussed, but is again based on averages and forecast. 

With the caveats out of the way, let's look at the individual candidates and whose fixtures fit well together for the first 10 games of the season (forecasting much beyond that period will only compound our uncertainty). I've set the maximum price at 5.0m and only included players who will (or at least should) open the season as their team's starter. Our candidates are therefore as follows:
  • Al-Habsi WIG 5.0m
  • Begovic STO 4.5m - Pulis has said this is an open competition but the quotes read of a manager trying to motivate his players in the preseason and I believe this is Begovic's job to lose.
  • Davis SOU 4.5m
  • Federici RDG 4.5m
  • Foster WBA 5.0m
  • Given AVL 5.0m
  • Green QPR 4.5m
  • Jaakelainen WHM 4.5m
  • Mignolet SUN 5.0m
  • Ruddy NOR 4.5m
  • Schwarzer FUL 5.0m
These 11 players give us 55 combinations that will each generate an average goals per game conceded, calculated by taking the best fixture for the pair for each week and taking an average. These combinations are summarised below:

It's worth noting here the number represents the average forecast goals per game for each fixture rather than purely a measure of strength of schedule. I opted for this approach as otherwise, taking an extreme example, one could be led to the conclusion that you would prefer to have Jasskelainen playing opponents who score 1.2 goals per game that Joe Hart playing opponents who score 1.3. We need to account for not just the opponent but the relative strength of the 'keeper and his defense too.

All these combinations cost between 9.0m and 10.0m (there are no 4.0m starters this season) with the majority of the best combos coming in at 10.0m. How does this compare to the elite starters? Well, for the first time since I've done this analysis the situation isn't clear.

If we first look at Joe Hart, we can see the scenario which has played out in prior years. Pairing Joe Hart with a 4.0m 'keeper (who remember won't start) will cost you 11.0m and your average forecast goals per game conceded will be 1.02 over the first 10 weeks of the season. So you save 0.11 goals per game conceded but you spend an extra 1.0-2.0m and also risk getting zeros if Joe Hart gets injured. If you want a legitimate sub to pair with Hart you need to cough up an extra 0.5m, thus leaving you a total of 1.5-2.5m short for the rest of your team.

I base my forecasts on a pre-season target of 2000 points and assume I'll use 80m on my starters and 20m on the bench, which means I need 53 points per week at a P$ of 0.658. This all translates to the simple question of which scenario do you think 1.5m is more likely to earn you the ~38 points needed to reach the 2000 mark: playing Joe Hart every week versus a combo of 5.0m options, or upgrading a Villa defender to a United defender, or turning Danny Graham into Demba Ba. I have always felt that the later scenarios will, on average, generate more points and have thus always gone with the rotation strategy.

Notice the past tense above. The question mark this year arises due to David deGea' price tag of just 6.0m. Pairing him with a 4.5m option like Begovic would give you an average goals per game conceded of just 1.01 and only come at a 0.5-1.5m premium over the above combos. Looking specifically at the two best combos above (SUN-WBA and SUN-WIG), the DeGea/Begovic combo will save you 0.12 goals per game (5 goals over a season) at a cost of 0.5m. That 0.5m should generate around 12 points which equates to three clean sheets, so the question is whether the 5 goals saved will equal those extra zeros. That's very close to call and probably too much so for us to reach a conclusion here.

We also need to consider a couple of other points, though these two split their support for each strategy. On the one hand, United could improve defensively from prior year thanks to the return of Vidic and Ferdinand, thus the data may be underestimating de Gea's value. On the other hand though, a budget team may well emerge as a very good unit, much like Swansea last year, and those using the rotation strategy will be best placed to exploit them.

This is a very close call and therefore one that probably won't dictate your season. I see some upside in United's defense though not to the 22 goals conceded 2007-08 days and I'm confident de Gea will solidify his place in the side and dispel any rumours of Lindegaard taking over. He is the only 'keeper I would consider costing over 5.0m and will only pencil him in if I'm 99% sure he has the #1 jersey and United have a fit back line. If you think any of the 9.0m combos are currently being underrated then it's likely they will deliver better value for money than any elite 'keeper, but if you're current plan is to go with a 10.0m combo, I think the extra 0.5m to get De Gea is worth a strong look.

I have been neglecting comments to date as I've been scrambling to get some posts out before Saturday's deadline, but if you post them below or @plfantasy I will go through all previous posts and do a user questions post soon.


Kalix said...

Hey Chris,

I too do this calculation every year, and came to similar conclusion that it's much closer this year than in previous seasons. As a result I dug deeper, and found that ignoring the 5.5m is a mistake.

Tot(5.5m) + Stoke/Reading(4.5m) rotate to give a score of 1.05 (should be similar to your based on my numbers for the pairs you've listed).

Tot + Stoke can be had for 10m, with a gpg of 1.05. Only 0.04 worse than DaGae.

vanilla said...

Why not Tim Krul instead?

Gummi said...

I've come to the same conclusion and will be going for De Gea, most likely.

Do you think the extra 0.5 (for a De Gea substitute) for a playing goalkeeper is worth it? My instinct says save 0.5 to spend elsewhere.

Anonymous said...

In my opinion, if we're considering a combo of two 5m goalies I can't see why we aren't considering 5,5+4,5m combos? If it's possible I would like to see similar projections for combos of Vorm/Krul/Friedel/Howard/Szczesny and Ruddy/Green/Federici/Begovic/Davis/Jaaskelainen.

Kalix said...


It looks like your team ID thingy has changed for the new season. Your link to the Pure Juice teamsheet now points to someone elses team :)

Chris said...

Why not use a 5.5M Keeper and their 4M backup? If one gets hurt, the other will play. I was thinking of a Vorm/Tremmel combo.

Big MB said...

One more thing to think about if picking degea over a keeper from "minor" club is that picking degea will limit you to two outfield united players

Gummi said...

@Big MB: That's a good point. However, picking De Gea would mean that I wouldn't be looking at a United defender.

Well, that, or skipping a midfield option to double down on clean sheets.

NovaHammer said...

With Chelsea playing twice this 1st week I see PC as a strong gamble for a week or two then selling to return to a lower season investment ratio as suggested....thoughts?

Nibbling Ferret said...

@Novahammer: That was my thinking as well - have splashed some cash and put Jaaskelainen in as the reserve keeper (as he has the best of the 4.5m keeper fixtures in gameweek 3)

Even considering having PC as captain for the first gameweek...

namedisplay said...

MU has an ugly Liv, Tot, New, Che, Ars in a six week period through Sep and Oct. I would delay the De Gea strategy until after that.

Sean D said...

Great analysis as per usual.

Guy Rendell said...

If you’re still tinkering with your team, make sure you read these top tips first:

Any comments on my ‘ten commandments’ and team are welcome…