Wednesday, September 19, 2018

Selecting defenders within the same team

We talked about this earlier in the season, but with a couple more games in the bag, I've updated the viz which attempts to help you choose your defenders. The "All Players" tab should in theory help you do this coming with a blank slate, but the raison d'etre for the viz is to help you choose between players in the same side, once you've decided that, for example, you want a City defender.

The viz can be found linked from the main menu, and here and should be opened separately to best read this piece. A few notes as of GW5 below.

I'm not sure how much value there is with this defense right now, but if you are planning to play the fixtures and take a chance with one of Emery's men, this would suggest it should be Mustafi rather than the 10% owner Bellerin who you should target.

Cook is the man here for now, though if a clear winner emerges from the developing Daniels vs Rico battle then one of those players could topple him based on either a lower price tag (Daniels, 4.4m) or Rico potentially taking set pieces.

However you present the stats, it seems that Alonso can justify his lofty price tag, though I think it's worth noting Rudiger here, who has offered a genuine attacking threat and comes at a discount that will probably reach 1.0m soon. Luiz is a surprise here as he has offered very little at the attacking end, despite his prior reputation.

Crystal Palace 
Coming into the season I had serious doubts about Van Aanholt's ability to justify what I thought was a 1.0m premium over his teammates, but actually turned out to be a 1.5m premium, after the emergence of Wan-Bissaka. In a previous post we estimated that every 1.0m spent needs to earn somewhere around 26 points, and so with a 2.5-3.0 point advantage after 5 gameweeks, we'd project Van Aanholt to end up a little short of that, but not a million miles away. However, if we feel that Wan-Bissaka is going to keep his job then the gap becomes 1.5m and then you're looking at needing something like 35-40 points to justify the added cost, and I just don't think Van Aaanholt is there unless (a) he starts become a legitimate goal threat from direct free kicks or penalties, or (b) you think his playing time is significantly more locked down than Wan-Bissaka.

Keane is an interesting name here as he accumulated those projected attacking points in just three games. He's due to return as soon as this week after his head injury which could make him an interesting name to look at for the GW7-9 run which sees Everton face FUL (H), LEI (A) and CRY (H). Despite the lack of clean sheets and 9 goals conceded, Everton's underlying defensive numbers are actually reasonably good, which could suggest a few defensive returns in the future too.

I know Maguire got his goal but Chilwell continues to offer more attacking potential based on the underlying stats for less money and significantly less ownership. Pereira has received some recent buzz too, but he's only created one extra chance than Chilwell while Chilwell has also offered more of a goal threat than the Portuguese fullback.

17% of managers continue to hold out with van Dijk but even though Robertson's price has risen by 0.2m, I would still encourage people to make that move. Van Dijk has just two attempts on goal through five games, which is admittedly more than Robertson's one, but then of course the Scotsman blows his Dutch colleague out of the water with 10 created chances to 0. At this stage I think one can make an argument that Alexander-Arnold or even Gomez represent better value than Van Dijk, unless you feel that the return of Matip will overly complicate their playing time.

Man City
Similarly, there is really very little reason to stick with Walker over Mendy at this stage, unless we learn that the recent knee issue is a cause for concern. Even then, if you were worried about Mendy, I think Walker to Laporte makes a lot of sense as the Frenchman (rested over the summer) starts to make one of those centre back spots his own in Pep's side.

Man Utd
The worry here is that by the time United get their defense sorted, Shaw will face rotation with Young, Valencia and Dalot. Mourinho has already been discussing the need for multiple full backs and while Shaw has been solid, he's obviously far from rotation proof in this side. At 5.1m he's cheap enough to have benched every now and then though, so he's still probably the pick here.

I don't know if this will be useful information to act on, but prima facie it is remarkable to me that Davies is ahead of Tripper here. The chances created given a slight nod to the England man (7 to 6) but Davies has more total shots and SiB to give him the overall edge, not even counting that these came in fewer minutes. Alas, those words "fewer minutes" are probably the death of Davies' fantasy value as he is already being heavily rotated with Rose, where as we assume Trippier will only be rested as needed, with the less reliable Aurier stepping in. If Davies offered a discount of 0.5m or more, those benchings might be worth absorbing, but for 0.2m I don't think we can justify the risk as things stand.

The underlying data absolutely loves Matt Doherty, whose 8 total shots trail only Alonso, 7 SiB lead all defenders and who places a solid 9th in created chances too. The question here is how often you are going to play your Wolves defender. If he is just an enabler, recruited to warm the bench and deliver 2 points when you have an unexpected rotation issue, then Bennett - at 4.1m - is still worth a thought. However, I think this does a disservice to Wolves, whose 25 SiB conceded are tied for third, behind just Liverpool and City. I don't think we're ready to say Wolves defenders are every week starters, but if you think they might rotate into half your games, then the additional goal and assist threat becomes much more valuable and Doherty becomes an outstanding pickup. After back to back clean sheets, I am amazed his price is holding firm at 4.4m.

Wednesday, September 12, 2018

Selling Salah

Given his exploits last season, and a pretty healthy start to this one, it's hard to believe the thought of selling Salah is even worth exploring. But, international weeks give us time to think through less obvious solutions so let's make a case for it. But first, a few opening notes:
  1. This is a short term measure. Despite Mane's hot start and promising form from the likes of Hazard, I still believe Salah is the best fantasy option for the season. His underlying numbers are excellent, his team looks as good as ever and his playing time is assured (other than the odd rest of course).
  2. This move make more sense if you haven't yet wildcarded, but plan to do so in the next 4-6 weeks, potentially before GW9 during the international break. If you have already played that chip then using three or four transfers in 4 gameweeks to rotate one player (along with a corresponding move) might be tricky (although if you have wildcarded the rest of your team is presumably settled?)
  3. This move arguably makes more sense if you haven't enjoyed the best start of the season and want to shake things up a bit, although I personally try and avoid such thinking at this early stage of the season too much.
The player
We don't need to dwell here, suffice to say that Salah has shown that he is anything but a one season wonder. He leads all midfielders in total shots (19), shots inside the box (12), shots on target (7) and created chances (13). Those totals would place him among the top 2 or 3 forwards too (the majority of which offer very little by way of assist potential). He played at the World Cup, but Egypt of course suffered an early exit so he's also relatively rested (though did face a trip to Egypt over the international break).

The opponents
From GW5-GW8 Liverpool will face Spurs away, Southampton at home, Chelsea away and Man City at home. There's not much argument to be made that this is a tough run. Spurs haven't looked amazing in a couple of their games this year and their 34 SiB conceded is somewhat mediocre, but they've still conceded just 4 in 4, with City and Chelsea both slightly ahead with 3 in 4 (Southampton have also only conceded 4 in 4, though underlying data suggests they have perhaps been a touch fortuitous and aren't a great defense).

In these fixtures last season Liverpool lost 4-1 at Tottenham, beat Southampton 3-0, lost 1-0 at Chelsea and then recorded that memorable 4-3 win over Man City. How to interpet Salah's output during this period is hard to measure. On the one hand, he notched 4 goals and an assist (although one was that fortuitous lob over Ederson after the Brazilian misplayed a pass) - an impressive haul in just 4 games - the kind that earned him his 13.0m price tag. However, Salah's xG was just 1.36 and his xA was 0.35, suggesting he was somewhat fortunate to rack up the points he did. One can note that he exceeded his xG total all season, but 32 goals vs 24 xG is a 33% premium, not the 300% premium we observed in these four games. Admittedly, looking at single game xG isn't really the way it was intended to be used, but I do think it highlights that there remains some question about Salah's output versus these opponents last season.

The alternatives
One of the reasons you can justify Salah's 13.0m price tag is by captaining him most weeks. A 7 point haul for 13.0m doesn't sound great but when you're consistently doubling that to mid-double digit returns, the purchase starts to make much more sense. With that in mind then, whether or not you sell Salah for this short period is in part dependant on who you will give the armband to. It's possible that in three of these fixtures you would likely consider someone else for the captaincy, and thus not having Salah becomes more palatable, and missing out of another captain-worthy option becomes more painful.

A quick reminder of the elite options' fixtures over the next four gameweeks:

  • Kane, Eriksen: LIV, @BHA, @HUD, CAR
  • Hazard: CAR, @WHU, LIV, @SOU
  • Aguero, Sterling et al: FUL, @CAR, BHA, @LIV
  • Lukaku: @WAT, WOL, @WHU, NEW

From this list, I would suggest that all these options are going to offer at least 3 opportunities for captaincy, with the worst game probably being City's trip to Anfield.

Of course, the forwards listed here would require a double transfer to access, but when dealing with such large amounts of cash, I think you'd only make the move if there were specifically two players you really wanted. Who that second player would be is a major question - and possibly subject of its own blog article - but to get an idea of the range of options moving Salah would open up:

  • Salah to Sterling/Hazard frees up 2.0-2.3m which could allow you switch a 2-point bench player like Neves or Kante into a Pedro, Walcott or Richarlison. You could also move a budget defender and jump on the Alonso bandwagon (or any of the other elite defenders)
  • Salah to Eriksen frees up 3.6m which is enough to convert Vardy into Kane or Zaha into Aguero. 
  • Salah to Aguero frees up just 1.7m, although even that can be enough to move up a level among defenders, from either Maguire to Alonso or from Bournemouth to Tottenham.
One final thing we have to consider here is the risk of rotation as we move into the European schedule and bumps and bruises start to take their toll (this might be especially true in a season following a World Cup). The relevant big teams play the below fixtures in all competitions between now and October 20th (the start of GW9, after the international break, and when Liverpool face Huddersfield i.e. the date you probably want Salah back in your side.

Chelsea (Premier League games are underlined)
Cardiff H, PAOK A, West Ham A, Liverpool A, Liverpool H, Vidi H, Southampton A

I'm not typically in the business of guessing lineups, but I think it's probably reasonable to say that, as a team, Chelsea has the best situation here (not necessarily in terms of strength of opponent but specifically as it pertains to rotation threat). PAOK are Chelsea's toughest Europa league opponent so Sarri may well send a full strength team to Greece, but then Chelsea still get the normal two days rest before a trip to West Ham on the following Sunday. With Liverpool to come in the league right on the heels of the Carabao Cup fixture, one suspects we'll see rotation there and then the other Europa League game is against lowly Vidi at the Bridge, where we could again expect to see some bench options. With Hazard also being eased into the season already, I'd be even more confident of Hazard's minutes here.

Man City
Fulham H, Lyon H, Cardiff A, Oxford Utd A, Brighton H, Hoffenheim A, Liverpool H

Guardiola will want to start City's Champions League strongly against Lyon, who are on paper their toughest opponent. Given the squad depth, it wouldn't be surprising to see a rotated team head to Cardiff the following weekend, which is a pretty favourable fixture to miss. I'd perhaps feel slightly more confident in Aguero getting the nod as we've sometimes seen Pep drop his "faster" players against sides who might play 11 behind the ball, in a desire to have as many technically gifted players as possible to work in tight spaces. That would arguably favour the Argentine over Sterling but either are liable to be dropped in this run. We can presume that City will want to be at full strength for the visit of Liverpool, meaning potential lost minutes either during the trip to Hoffenheim or in the previous week's visit of Brighton. One could imagine a scenario where Aguero and co start that Brighton game but are quick to be pulled off if City jump out to a lead (something to be considered when handing out the armband).

Man Utd
Watford A, Young Boys H, Wolves H, Derby H, West Ham A, Valencia H, Newcastle H

With Rashford suspended for the next two Premier League games, Lukaku immediately enjoys almost guaranteed starts against Watford and Wolves, with perhaps Rashford spelling him some minutes in the trip to Young Boys - a must win game for United given the other strong opponents in United's group. Assuming we see a rotated team in the Carabao Cup - as we did at this stage last season in the 4-1 win over Burton - that would again leave Lukaku rested to face West Ham. The only question would then we whether the Belgian would be utilized against both Valencia and Newcastle, which if not would be a huge blow as that's a promising fixture despite Newcastle's solid defensive efforts to date. Given how much he was relied on last season, the pressure on Mourinho and the apparent lack of faith in Rashford, I might say that Lukaku is best placed to play all four games over this period. 

That said, he's also on the weakest team, may therefore have less opportunities and, frankly, just isn't as good a player as the other names on this list so this is a very tough call. Sometimes we avoid making decisions for fear of looking foolish, and of all the players to sell Salah for, Lukaku probably has the biggest risk of making you feel idiotic, but don't let that alone scare you off - there is plenty of evidence to support this move.

Liverpool H, Inter A, Brighton A, Watford H, Huddersfield A, Barcelona H, Cardiff H

Other than this week's visit of Liverpool I was very interested in getting access to this Spurs side, especially with Son returning from international duty. The fixture list does complicate things a little though. Spurs have been handed a really tough Champions League group and face two of their toughest games in the coming month. After presumably playing a full strength side against both Liverpool and Inter, one could see a couple of players getting a rest at Brighton, especially those who enjoyed a busy summer and have started every game so far - including Eriksen and Kane. With the visit of Barcelona then focusing the mind, you wonder if those top assets might then again be given a reduced role in either the proceeding trip to Huddersfield or the subsequent visit of Cardiif. 

Normally I would say that Spurs players have the safest roles given the comparative lack of depth compared to the other elite side but with the emergence of Lucas Moura, the obvious fatigue of the players involved at the World Cup and the complication of their very difficult Champions League group, I am now not so sure. If Kane and co were firing on all cylinders I'd be more willing to overlook this concern but as it is, this group seems like too big of a risk to sell Salah for.

The summary
I think the case for Hazard is probably the strongest, especially if that 2.3m will then allow you to access another key target. Lukaku is tempting as a differentiator (8%) owned but my own personal front line is pretty set - Aguero, Wilson and Mitrovic - and I will probably keep them intact until I wildcard. City's fixtures make Sterling really tempting, even if his likelihood of rotation is higher than some of the other options. Again, in my personal situation I already have three City players (Ederson, Mendy and Aguero) who I'm pretty happy with so it's difficult to fit the England man in. The money saved by switching to Eriksen is pretty tempting - the 4.0m in the bank would allow me to switch the freefalling Jota into Sanchez(!) - but those fixtures are a big concern. One move I didn't discuss above is to do Salah to Mane to lessen the financial blow of the tough fixture list, though that seems like a lot of transfers to commit to the project unless you have a strong idea in how as to how to use the 3.0m saved (I'm not yet convinced that Mane will continue to offer equal output to Salah).

There's a case to be made here, for sure, but I think it only makes in specific circumstances so I wouldn't want to force the issue. Those who do make a move face four weeks of watching games from behind the couch, but isn't that what makes fantasy fun!

Wednesday, September 5, 2018

Comparing like-for-like fixtures and underlying data

As noted in other recent posts, it can be hard to know what data to rely upon in the early gameweeks. We all know that shots inside the box or those on target tend to correlate well to goals, but with only a few of those events happening each week, it can be easy to overreact to unstable data. One data point I am interested in, is how well teams are doing compared to same fixtures from prior season. This data is still fraught with issues - teams might have changed personnel or have injuries or simply suffered a bad game - but we at least have a baseline to compare against which gives a little more context to what we're looking at.

The below visualization shows the Goals (G), Expected Goals (xG), Shots Inside the Box (SiB), Created Chances (CC) and Successful Passes in the Final Third (P3rd) for each team's fixtures for 2018-19 (CY) compared to the same slate of games from the 2017-18 season (PY). We'll highlight a couple of the more interesting observations below the viz:

Of the four games Chelsea have played to date, they drew a blank in three of them in the corresponding fixture last season. Their combined xG over those games was 3.0 - so there's no doubt they were a little unlucky to come up empty - but still, the difference between that streak of zeros and this season's 7 goal haul (with an xG of 4.3) help to underline the potential improvement we are seeing under Sarri. They doubled their SiB in two of those games and have come close to tripling the number of passes they have completed in the final third of the pitch. Looking at the opponents doesn't appear to completely explain this improvement either. Ignoring Huddersfield, against which Chelsea played just about as well as last season, we find Arsenal who one could argue are still finding their feet under Emery but are otherwise largely unchanged from last season, a Newcastle side that is largely unchanged and gave Man City a good game this past weekend, and a Bournemouth side who appear stronger than last season.

The pressure that Sarri's team is going to create with so much possession in the final third is surely going to be too much for some sides, and while one can argue how that translates into wins and losses, in fantasy terms it creates more chances, which typically lead to shots and points over the course of a season. Chelsea's lineup make it somewhat difficult to load up on their players, as you might do with, say, Mendy, De Bruyne and Aguero at City, but this improvement makes Hazard extremely appealing and I think justifies doubling up the Belgian and Kante, or even Pedro (though I fear too much rotation with the latter with Willian also available). If Morata could find some form and develop under this system then he too might play his way into contention before the year ends. We've already discussed how Alonso appears to justify his price tag - at least compared to his Chelsea colleagues - and nothing here seems to contradict that fact.

Liverpool have the worst differential between the goals they've scored in the four games this season versus their haul in the same fixtures last season. Of course, before you immediately start planning for how you're going to divert your 13m Salah funds, consider that the baseline they have fallen from was impossibly high - 13 goals in 4 games - and if you look at their xG they are actually much closer with 9.3 xG for this season compared with 10.4 xG for last. That said, a quick peak at the underlying data does show some drop beyond just unfortunate timing with a significant drop in both SiB and CC. I don't think this is anything close to being a cause for concern but it might be a small factor when considering whether to sell Salah for a few weeks during this tough run of games: if the heights aren't quite what they were last year then it becomes less scary to sell the Egyptian and use your funds elsewhere with a reduced fear of Liverpool massacring sides and earning huge points totals for their players. For what it's worth, I am considering moving Salah and Alexander-Arnold for 4 gameweeks, given the fixtures and the fact that neither have seen a price rise to date that will make buying them back a big issue.

Leicester's stats are interesting, though I'm not sure what they mean from a fantasy perspective. Their xG and SiB are down around 40% from the corresponding fixtures last season, and that has played itself out in the goals column, that has decreased from 6 to 4 (they did of course manager two against Wolves which is excluded from this analysis, so let's not be too harsh too quickly). Their passes completed in the final third have doubled though, which suggests increased attacking possession. I don't see enough Leicester games to conclude that this is an intentional strategic shift or whether it's due to Vardy missing a couple of games (and with him that ability to counter so well), but it's something to watch going forward as it could shift how different Leicester assets are valued in this side. For now though, I want to see a return to form when Vardy is back or else fear this side might offer little fantasy value this year.

Man Utd
This data might prove to be some relief for Man Utd fans, given that the majority of the news about their club seems to suggest the sky is falling. After struggling to generate shots in GW1 and GW2, United doubled the number of SiB against TOT and BUR that they managed last season, and saw their xG rise from 1.6 to 4.0 in those two games. Their completed passes in the final third took an almost unbelievable drop against Leicester and particularly Brighton, to the point I had to go back and check the data but based on what we've since seen against Spurs and Burnley, the outlier wasn't so much those games this season but the corresponding fixtures last season where United were able to dominate possession against weaker sides, albeit not always to great effect as United of course went on to lose that game to Brighton. I'm not exactly sure what to make of this fact. Are United more clinical, luckier, or just playing a different way? I think we need a few more games from this team to figure out what they are and which, if any, of their pricey assets might be able to contribute to our fantasy teams.

It's too early to write teams off, or even be overly "concerned" about their prospects for a long season ahead, but the early data for Brighton is not as promising as one might think given their win over Man Utd and a close loss to Liverpool. Their created chances and shots inside the box have fallen off significantly from last year, and while their xG looks not too bad, it is being propped up by the penalty they earned against United (worth 0.79 xG). Now it's not necessarily fair to just arbitrarily take that penalty away as they did indeed earn it, but just to note that without it, they'd have an xG of 1.8 over three games in which they managed 3.3 xG last year. Again, we're not close to worrying yet or taking away credit for the penalties, but of they were indeed events of luck rather than due to a repeatable skill on Brighton's behalf, this might be of note of caution to those starting to look at the likes of Pascal Gross for fantasy production based on his team's prior season performance.

It's not a spectacular improvement but Everton have improved pretty much across the board in their three fixtures to date (we're excluding the GW1 effort against Wolves in which they also notched a couple of goals, albeit perhaps a bit fortuitously). They haven't played the toughest run of games of course, but the idea behind this analysis is that is allows for like-for-like comparisons and on this basis, the team has clearly developed. With injuries and suspensions impacting Walcott and Richarlison and Tosun and Sigurdsson yet to attract much attention, it's possible that people might start to forget about Everton, but the data here suggests they have the makings of a good mid-level team that might be a source of reasonably price squad players.