Thursday, August 30, 2018

Looking at early conversion rates (to justify your transfers or crush your dreams)

There seems to be an emerging narrative in the early week of the season that fantasy teams are starting to converge and a “template” has emerged for people to follow. The strong start of the elite full backs, the apparent “must own” status of pricey Salah and Aguero and a couple of budget midfield enablers (Neves and Kante) make it seem like this is the case. If we look at the prior year ownership data for GW3 though, we see a broadly similar pattern of ownership:

Overall we had 14 players with 20%+ ownership where as this season we have 17, with one clear leader over 50% and a handful more over 30%. This season I think we have a couple more budget options with the likes of Wan-Bissaka, Richarlison, Neves, Kante and to an extent Zaha and Mkhitaryan, where as last year we saw just Hernandez, Hegazi and Rooney. I wonder if the the presence of these budget options here might well be what is leading folks to conclude that teams are starting to all look the same. Differentiation is always an important consideration and so while the question as to whether or not teams are less diverse is an interesting question, but the point I wanted to dwell on here was another.

If we look back at that list of early favourites from last season and see where their ownership percentage was a few gameweeks later, we see some interesting results:

By GW10, 12 of the 14 most-owned players had either decreased in ownership or stayed approximately the same. The likes of Rooney, David Luiz, Mane, Hernandez and Alli had all seen their ownership percentage drop by over 10% with only Salah and Kane increasing their numbers significantly. By the halfway point of the season five players had lost 20 percentage points or more. On the flip side, players like David Silva, Jesus, Sane, Otamendi, Doucoure, GroƟ, Azipilicueta and Richarlison all had low ownership numbers at this point last season but by GW10 were all close to 20% owned.

So all that is a long road to get to our destination of concluding that we probably don’t know as much as we think we do, and those players who look to be essential now might be discarded in a few weeks and new options will surely emerge. I’ve noted before that I’m not super comfortable suggesting that Player X is a “good” pickup because he has good shot totals as this data is so noisy after just a couple of games. Schurrle, for example, seems to be getting some buzz given his high shot totals but a quick peak at the underlying data shows a mediocre xG, backed up by watching last week’s game – he’s clearly a promising option with a good pedigree and involvement in that team, but his mandate appeared to be to shoot on sight and thus his lofty shot totals are probably not a great data point to rely on alone. To try and tease some of these over (and under) achievers out, we can plot players’ conversion rates for this season (CY) and compare to the prior year data (PY) as well as the rate at which their teammates are converting chances (Tm). If we see some of these “hot” players with conversion rates that are well above normal then that might pour some cold water on our enthusiasm (or vice versa for players whose rates are inexplicably low).

Mkhitaryan (20% ownership)
Clearly his G/SiB rate is higher than expected, but not so much as to scream regression. His SiB/PBT rate is right around where we'd expect, suggesting he's earning his shots at a reliable rate, and of course he still offers some value from shots outside the box given his ability. The rate at which he's creating chances coupled with a slightly lower than expected conversion rate suggests his assist potential could even be a little understated. Arsenal have a good fixture list over the next seven weeks but there aren't too many other options in this team you'd feel comfortable backing. You typically like your mid-price options to offer a bit more differentiation but 20% isn't too high and the pick just makes sense right now.

Wilson and Fraser
As I started to write this, I very much expected both Wilson and Fraser to have the kind of stats that pointed towards a nosedive in the coming weeks. I'm a fan of Wilson and Fraser has really impressed, but for a reasonable but non-explosive Bournemouth side, I just didn't imagine there would be enough there to support their points hauls. However, Wilson actually has a slightly lower G/SiB conversion rate and while he is earning shots at a slightly higher rate based on his PBT, that could at least be in part down to the way this side is playing. His assist threat looks overstated but for a 7 million forward, you don't need many assists so long as the goals keep flowing. 14% (and climbing) is a reasonable number of managers, most of whom are presumably quite active as he surely wasn't a huge GW1 play, but with many teams looking to install two mid-level forwards, there's good reason to suggest Wilson can be one of those players.

Again, Fraser appears to have some hints around the edges that he's overachieved a little, but really nothing to worry about. His set piece ability will mean that he will always have the potential to create more chances than open play numbers would suggest and with a decent stable of teammates, it's not unreasonable to think they will be converted a decent rate.

The Spanish full back is good value for his goal and is arguably as likely as anyone to lead all defenders in goals by the end of the season. His assist haul however is significantly overblown and we'd expect some serious regression there. This means that to justify that premium price tag, you need reasonable confidence that he's also going to earn defensive points. Chelsea only have the one clean sheet from the first three games, although they are second only to Liverpool in terms of shots conceded inside the box, despite two away games (albeit to somewhat weak opposition) and the visit of Arsenal. If you buy the red flag about assist potential and aren't totally convinced about Chelsea's defensive potential then I'm not sure Alonso justifies the 0.4m more than Mendy, or the 0.6m over Robertson (both of whom I would expect to notch plenty of assists and clean sheets but perhaps not the goals). That said, there is a pretty strong case for owning all three of that group and I therefore wouldn't argue against feeling the need to bring Alonso in while his price continues to rise.

The numbers here suggest we pour a bit of cold water on Zaha, although given his price and guaranteed playing time, I don't think we need to panic. However, with a G/SiB more than double last year, we'd expect him to slow down a bit and when you look down that list of teammates, you can see why. Without too much other attacking quality in the side, teams should be able to focus on Zaha and as the season goes on and the tactics become more known, it'll be a hard lift for Zaha to continue his form. Of course, Townsend is a decent player, the attacking full back pair of Wan-Bissaka and van Aanholt look promising and new arrival Meyer comes with a good reputation, so it's possible this Palace side continue to develop as the season progresses. I'd say if you own him or sign him now then the move still make sense, but with brutal fixtures starting in GW10 and continuing for the most part through GW18, time is certainly of the essence.

Richarlison and Walcott
The Everton duo have 5 goals between them and a combined ownership now approaching 40% (with a shift from the Brazilian to the English man due to the former's red card). However, Richarlison has scored 3 goals from 4 SiB and Walcott has 2 goals from his 3 SiB; both rates which look unsustainable for anyone, nevermind two players whose profile pictures weren't exactly shortlisted when defining "clinical" in the dictionary. Now, 4 SiB in two and a half games is a totally reasonable haul for Richarlison and his penalty box touches and pass numbers suggest he will continue to earn those high value shots. The numbers don't, however, speak of a "must own" player and with his absence in GW4-6 followed by trips to Man Utd, Chelsea and Liverpool in GW10-14, I wouldn't rule out selling him and circling back after those games are in the books.

Walcott too has some reasonable underlying data and his game isn't necessarily going to be well reflected by stats looking at total touches, but his shot totals are pretty low and he's offering little assist threat, making him a fine but not overly exciting option at 6.7m. With home games against WHU, HUD and FUL in the next 4 gameweeks, he makes sense as a short term rental, or for anyone convinced they need to switch back to Richarlison immediately post-suspension, but I don't know if this move is certain enough to justify the 500,000+ transfers we've already seen in the gameweek. Of course, there aren't too many options in the 6-8m who look totally convincing right now, so the move certainly isn't crazy either.

Speaking of mid-level midfielders, I am surprised Maddison's ownership numbers are so low given the chatter around the Leicester man, although not at all surprising based on his underlying data. A single SiB and two CC don't suggest a breakout in the near future, even if he has impressed to the eyeball test in that role off of Vardy and more recently Iheanacho.

Mane and Salah
One current line of thinking is that Mane can provide sufficient Liverpool cover to free you of the need to own Salah, giving you over 3 million to spend elsewhere in your team without losing too much ground. Through three gameweeks that strategy has obviously proven wise with Mane not only matching but actually surpassing Salah in both goals and fantasy points. This data suggests the next set of gameweeks might not go the same way though. While Mane has been significantly more clinical than last season - converting his SiB at double last year's rate - Salah has actually been less clinical than we might expect, although I think it's reasonable to expect some regression from last season's lofty heights. The fact that Mane is second only to Salah in total shots sounds promising but when you realise that Salah has literally doubled Mane's total efforts and out shot him 11 to 8 inside the box, it starts to suggest there really does remain a gap here, and it's probably not a question as to whether Salah starts to outscore his teammate, but by how much, and more importantly, whether than gap justifies 3 million pounds. I think there's a pretty strong case to be made for owning both Mane and Salah, especially given Firmino not looking quite as sharp as last season, but given the 60% combined ownership numbers for Robertson, Van Dijk and Alexander-Arnold, that means you could have 28 million or so invested in a single team that is about to face TOT, CHE and MCI in the next five gameweeks. I could definitely see a scenario where I own both these players in the future but right now that looks tough to justify and given Salah's threat, that Mane ownership number starts to look a bit big to me.

Despite Mendy having 6 totals shots (compared to Alonso's 7), I don't think it's controversial to suggest that the Chelsea man would be strong favourite to finish the season with more goals. Assist wise, both players have benefited from unsustainably high conversion rates of the created chances, though Mendy's case is not as extreme as Alonso who has literally more assists than created chances. I think it's fair to conclude that Mendy offers just about the best assist threat among defenders though, given his advanced role in the best team and so while the lack of reliable goal production is a concern, I do think the assists will continue to come, and I'd expect City to have the best defense again, which together with a lack of too many options at his position in the City squad, conspire to make Mendy justifiable at any price up to probably close to 7.0m.

Not much to add on Aguero here, other than to note that while he already leads the league in goals, he's just doing what he does. Nothing suggests he's overachieving or outperforming his stats by a material amount, and the only thing likely to slow him down is rotation (which could come as soon as this week) or injury.

All this data should be taken with a pinch of salt given the small sample, and that applies even more to Trippier who missed a game, but I just want to note here that while the England man can probably fairly be put in that elite fullback group with Mendy and Alonso when it comes to assist potential (5 CC already in just 166 minutes of action), I don't think his goal threat is that much higher, if at all, just because of his free kick ability. He's managed just two totals shots in his two games, and only four touches in the penalty area, compared to double digit efforts from Alonso and Mendy. The free kicks are certainly a nice bonus, but with Eriksen, Alli and even players like Lamela and Dier in the side, I don't know if that's quite the gold mine that memories of Russia might suggest for the England fullback.

I think most managers buying Pereyra are savvy enough to know that his current pace is obviously going to slow down, the question just being by how much. WIth that in mind then, I think this data is actually quite positive. A 38% G/SiB is high, but not laughably so and it's fair to argue he's been good value for at least one or two goals based on the underlying data. His 8 shots have all come inside the box giving him the exact shot profile as the significanlt pricier duo of Alli and Mane. At 6.2m he's just about cheap enough to bench every second or third week, although the upcoming fixtures of TOT, MUN and at ARS look perhaps too daunting to justify pulling the trigger now. If you haven't already jumped in, then it might now be worth waiting out that run of games given the presumed slow down in price rise after the early gameweek rush.

I imagine a lot of teams had Arnautovic in them at some point during the preseason given his reasonable price tag, ability and success last year in a more advanced role, and he opened the season with a solid but not sky-high ownership number that surprisingly hasn't changed too much. I guess his injury status is weighing against his back-to-back goalscoring games, offering a bit of time for prospective buyers to consider their options. He's earning chances at a higher rate than last year, though that could partly be role-related, but nothing else is really a cause for concern here and once the injury situation is cleared up I would definitely have the Austrian shortlist The only significant arguments against the West Ham man are the overall uncertainty around his team - the Hammers have managed just 16 SiB in three games, the 4th lowest mark in the league - and the decent array of mid-price forwards that have emerged (Wilson, Mitrovic, Zaha, Tosun) as competition.

I'm going to leave it there for now, having only focused on those with high ownership numbers, but plan a follow up post where we look at players with low ownership numbers whose underlying data suggest some position regression in their future.

No comments: