Thursday, March 28, 2013

Gameweek 31 Preview

Clean Sheet Rankings
Attacking Rankings
Individual Rankings

Note: Individual rankings are based on an historical average data model only and no account is taken of expected playing time or rotation risk. Where a players is confirmed as injured or suspended I endeavor to remove them for the below listings but assumptions regarding rotation are not included in the below analysis.

Thursday, March 21, 2013

Player share of SiB

The new Tableau just launched and they've added treemaps, a form a visualisation I've been meaning to add to the site for some time. Now that Tableau carry this format, my existing data is already ready to go, so below is the first of likely many treemaps, here showing players' share of shots inside the box, split between games at home and away. Player share is given on a percentage basis based on the minutes they have played so you don't need to adjust for playing time. However, standard caveats on small samples should be noted, hence I've excluded all players with less than 900 total minutes played:


Saturday, March 16, 2013

Late season form for mid table teams

You don't hear it so much these days - or at least I don't, having left the world of 24/7 Sky Sports News to one in which my weekly football dose is received in 90 minute spells with little or no nonsense chat - but anytime a mid table team would lose towards the end of the season, we always heard that they had one eye on their upcoming holiday and had mentally checked out by late March. If this is indeed the case then the fantasy impact would be huge as (a) we would obviously wish to avoid such players and (b) playing someone like Swansea would suddenly become a potentially easier fixture than facing a team battling to stay in the league.

The below assessment is very simplistic and should not be taken as an attempt to definitively answer this question. It's merely a first attempt to shed a little bit of light on this topic and see if there's something here to investigate. The analysis contains a number of limitations, not limited to:
  • I have not adjusted for strength of schedule, so while I'm comparing a team's form through 30 weeks versus the final eight weeks, if teams happen to play the best eight teams in the league to end of the season and lose most of them, the analysis would flag this as an exception where as in reality such results were pretty much expected. 
  • I have not factored in anything surrounding managerial changes or player transfers so if you believe that January signing or interim manager turned a team around, this wouldn't necessarily be explained below.
  • I have not gone into why teams are winning or losing, only the end results. As we know, results can be misleading when judging how well a team is playing, and mere randomness would account for some fluctuation in points, even a team continued to generate and concede the exact same number of chances every week.
With those caveats in mind, let's take a look at the data. We have plotted the points per game of teams through the first 30 weeks of the season (PPG30) against their points per game through the final eight games (PPG8). I like this cutoff point as (a) it's the stage of the season we're at now and (b) by the 30 game mark you have a nice group of teams who are very likely safe but have little chance of making a run at a European spot (this year this is probably everyone between West Brom and Norwich). Our simple hypothesis would be that if a team averaged 1.8 PPG through the first 30 games, they should continue to do so through the final eight weeks (again, ignoring quality of opponent for a minute). What we would deem an interesting result is whether certain teams (say, those in the top six) continue to perform a ready rate, while others (say, those in mid table security) see a noticeable decline in results. For this reason, we have colour coded the chart to highlight each team's position at the 30 game mark (simply split between, top six (T6), mid table (mid) and the bottom six (B6):

The only obvious conclusion from the above chart is admittedly that nothing is obvious, but if you bear with me for a second, there are a couple of points to note:
  1. We've seen some pretty incredible changes in fortunes for teams at almost every position with teams like 06-07 Man United going from a 2.5 PPG average to 1.75 over the final 8 games, while 06-07 West Ham and 11-12 Wigan went from a 0.8 PPG average to finishing with championship form (2.25).
  2. While we have seen differences every season, there is clearly a strong trend here (r-squared = 0.88) so our starting point for teams should still be that we expect business as usual, rather than working on the assumption that the QPR and Readings of the world are all set for late season surges.
Let's pull some of the key trends from the data to see if anything else jumps out:

On the face of it, the above data is quite promising with regards to identifying a genuine trend of late season declines/surges. We see more decline among top six teams and more improvement among the bottom six teams, though this is to be expected as those teams have further to fall/rise and - at least at the very top - we will have some seasons where the best teams are resting players once the league is effectively over if they are involved in other competitions. The key though are those middle teams, who have shown a much higher propensity for declining (47/80) rather than improving (33/80), which seems to fit with the popular narrative. 

Less exciting is the above chart which shows that the majority of mid table teams (66%) improving or declining by just 0.5 GPG either way, though again, this was always likely given that they have less scope to rise or fall. The fact that we have three teams who declined by a full point per game all within the mid table category, is however a plus.

So, it looks like we have something here, namely that mid table teams do appear more likely to decline than improve, presumably with the teams at the bottom of the league vulturing their points more often that not. Let's dig a bit deeper though and see how they improve/decline in terms of goals scored/conceded.


There are a few questions coming out of this data which might be worth looking at in a future post, but for now let's stick with those mid table teams. We can see that they lost ground both in terms of goals scored and goals conceded, though the decline on the defensive side was more pronounced with the average decline being close to half a goal per game conceded. This again fits with expectations as it's easier to stay focused on the attacking side of the pitch, while mundane marking from corners or tracking a midfield run might seem less important when the end result isn't overly important.

So what does this all mean? Well, that's hard to say on such a high level but I think there's enough here to at least give us pause for thought before buying someone from the Swanseas and Fulhams of the world (particularly on the defensive side of the pitch). Saying that though, it should be noted that even given the above, teams in the bottom six still conceded, on average, 0.2 GPG more than the mid table teams over the end of season run in, so replacing those mid table defenders will be tricky. Considering that the offensive declines have been lower across the board, it might be advisable to try and put less emphasis on defense, trying to grab a couple of premium players to start every week then spot starting one of the cheaper options where you can. History would suggest that a four, or even a five, man defense would be inadvisable during the late stages of the season. 

Wednesday, March 13, 2013

Dousing the fire, fanning the flames: Gameweek 29

A slight change of pace with this week's fanning the flames piece due to a couple of observations over the last few weeks and months. First, I felt that old graphic didn't really show enough information that might lead us to fan said flame (or, more likely douse those hyped up fires). Second, looking at the data on a single week basis was always going to lead to a large number of 'variances' due to the ridiculously small samples involved. I've therefore currently included three gameweeks of data, but have built the model to allow for anywhere between one and five to be included so this may again be tweaked in the future. Finally, this is a piece which should probably run every two or three weeks to avoid too much repetition, so this new format lends itself more to that time frame.

First then, let's look at the new graphic and a couple of lines to illuminate the way the data is intended to be read:



The G, xG, A, xA, P and xP simply set out the expected goals, assists and points for the selected player over the defined period (in this case three weeks). The only point to note is that the expected values are based on the actual shot totals accrued, rather than the model which uses historic averages. The key reason for differences therefore will be players converting chances at a higher than expected rate, a telltale sign that we way wish to temper our excitement after a big game or two.

The charts are split into two categories: trends and full season data. The former show the players' performance over the last x weeks (again, this time we're using three) and the latter shows the conversion rates for the full season. The key point here of course is to illustrate unusually high conversion rates to suggest that a player's uptick in "form" may be unsustainable or that it is a genuine product of taking more quality shots etc.

SiB% Percentage of a player's shots taken inside the box
SoT% Percentage of a player's shots that hit the target (a rate that limited research suggests is a sustainable skills rather than goals per SoT which appears to be heavily luck driven)
CC% Percentage of player's created chances converted into assists

Without further delay then, on with the analysis:

Gabriel Agbonlahor
It's not that Agbonlahor is totally without appeal, it's just that for 0.3m you can upgrade to the legitimately excellent Benteke, who is still held by an inexplicably low 11% of managers. As we can see from his data here, his 'success' has been more attributable to higher conversion rates, with his goal coming on just two SoT, which then of course earned him a dubious three bonus points which are still massively correlated to goals.
Status: Sell

Gareth Bale
I try not to dwell on the elite players in these pieces as you all know that Bale is one of, if not the best midfielder in the league right now, but just to emphasise that point: a quick scroll through each of the highlighted players will show that only three players boasted an xP of 10 or more, with Bale topping them all with a stunning 16. This score was supported by 20 (twenty!) shots and 12 created chances, so while it's inevitable that he can't quite continue at this rate for the rest of the season, there is more than enough here to justify any price rises or the large ownership numbers. The only knock on Bale is the lack of double gameweeks on his horizon, which puts him at a slight disadvantage to some of his premium peers.
Status: Buy

Dimitar Berbatov
Berbatov appears to have lost some of his luster of late, seeing him lose 170k followers over the past few weeks. That will likely change after back-to-back scoring games and there's enough evidence in the data to suggest his success may be sustainable. You don't love this week's trip to White Hart Lane but otherwise Fulham enjoy a decent run for the remainder of the year and Berbatov is well placed to take advantage of those fixtures. There's some concern about the rate at which he's hit the target of late, but the Bulgarian continues to take quality shots and is helped out by his penalty duties to supplement his points haul. He isn't as flashy as the likes of Benteke or Lukaku (who admittedly, I hype all the time) but his ownership numbers are now considerably below that pair (along with other mid level favourites like Lambert, Fletcher and Sturridge) and thus he deserves some attention.
Status: Buy

Yohan Cabaye
I love Cabaye and given my complaints in the past about a lack of central midfielders in the league who can genuinely dominate a game for 90 minutes, I'm delighted to see him emerge as a force at Newcastle. The shot totals are great too but that SiB% is worrying and needless to say he won't continue to hit the target at a 50% clip is he continues to shoot from long distance (even if he can shoot with the accuracy shown this week with his delightful free kick). His goalscoring threat is, of course, supplemented by a very good assist potential, though for one reason or another his teammates haven't helped him to date in that department (one assist from 36 created chances). The upcoming trip to Wigan is harder than it sounds and given the follow up date with Man City at the Etihad, there's strong reason to hold off here for a couple of weeks. After that though Cabaye is an interesting pickup at 6.6m if you aren't convinced by Sissoko's prospects in this side.
Status: Consider investment after GW31.

Santi Cazorla
Cazorla is another once-loved option who lost 150k owners in a matter of weeks before bouncing back with a couple of good performances in GW26-27. The data above (only based on two games given Arsenal's off week in GW29) shows nothing out of the ordinary other than a slightly higher tendency to take shots inside the box (and of course of a note of caution that his two goals came against a weak Villa side). Arsenal's next four gameweeks are pretty sensational on paper, and even extended to the end of the year look pretty solid, meaning that the likes of Cazorla could easily be slotted into your lineup and relied upon from here on if you so desire.
Status: Buy in a vacuum, though he's still behind Bale in the model

Papiss Cisse
I was set to write that Newcastle have really turned a corner of late, adding 12 goals over their last 6 games, but a closer look at the data makes me a touch nervous. Over that period that have created, on average, 25% less SiB than other teams did against those same opponents, with the only positive being the +47% performance against Chelsea (without that game they average -39%). There have been plenty of long range efforts, but while I sung the praises of Cabaye above, I'm not willing to heavily back a team who are relying on lottery-shots for success, especially when it comes to judging Cisse. The model sees Cisse as more of a Benteke/Berbatov/Fletcher type player, and I don't see anything to really justify the premium needed to upgrade to him over that trio (among others). One positive is his low ownership number, though given that required premium, I'd personally prefer to hold the more popular option this time, and try and differentiate elsewhere, where better value can be found.
Status: Sell based on that price tag

Jack Collison
It appears that Nolan, Noble and Cole are all short term absentees so the increase in playing time for Collison will likely be short lived.
Status: Sell

Stewart Downing
Downing is a potentially perfect Moneyball type player, as the very mention of his name sends scores of fans into giggling hysterics, particularly if accompanied by any kind of praise. His apparent security in this very promising Liverpool side is certainly a plus, as is the rate he's been taking close range shots (which has clearly improved of late too, whether by chance or design is less clear). The problem is still a lack of volume of efforts though, and his goal this week was obviously a touch fortunate (though in his defense, he did have to be in a good position and then finish it off). At 5.7m he's possibly a touch overpriced as he really belongs in the same group as the Puncheon/Pilkington/Maloney group, but if you have a small amount of cash lying around and are currently invested in someone whose form or playing time has suffered, I don't think it's crazy to take a look at the oft-mocked winger. The next four games look solid (though three are on the road) and a lack of double gameweek potential does hurt him over several of his peers, but his 1% ownership is a factor and anybody who gets to play with Sturridge and Suarez every week with little competition for playing time looks okay to me.
Status: Cautious buy, though I'd still say better upside can be found in his price bracket.

Marouane Fellaini
The trends for Fellaini show a fantasy player in decline, though in fairness he is coming down from a frankly insane platform where we was racking up elite SiB totals on a weekly basis for a mid level price tag. The fixtures don't look too promising for Everton, at least for the next three weeks, after which they get the double gameweek (though with half of that coming at the Emirates I wouldn't be forecasting huge totals for Fellaini and co that week either). A look at the midfield points standings shows that no one ahead of Fellaini can be had for close to his price tag, and that probably holds up for projections too, though with the tougher fixtures and his declining stats, I'd venture he's probably closer to the 6-7.0m group now than his is to the truly elite options above him.
Status: Still ownable but he shouldn't be targeted until GW33 for the double gameweek.

Junior Hoilett
Hoilett is a useful player, though as a Canadian resident I am obliged to offer a comment that he really needs to give up on his chances of playing for England and come to the great white north. Moving on from that though, Hoilett looks like a classic regression candidate, having hit the target with every one of his shots these past couple of weeks, while getting his three created chances converted into two goals. While I like Remy quite a bit, I think we can conclude that QPR aren't ready to shatter conversion records this year, so I'll pass on Hoilett until we see quite a lot more from the Canadian.
Status: Sell

Shinji Kagawa
Three shots, three on target and three goals (extremely well taken I might add). However, the fact that we only have one game of data here illustrates the chief concern here: playing time. I love the idea of Kagawa, particularly if deployed further up field, but nothing to date suggests that situation is sustainable.
Status: Sell

Arouna Kone
Kone's numbers look good and haven't really fluctuated over the past run of games so there's good reason to buy what the Ivorian has been selling of late. There's a lot to like with Wigan's fixtures, and aside from the trip to the Etihad, you probably feel good starting Kone in every game from here on in. The fact that Wigan absolutely have to try and win every game is a further bonus as the traditional late season slumps may start to kick in for some of the mid table teams (on another note, has anyone ever actually researched this? Do mid table teams tend to 'slack off' towards the end of the year or is that just a convenient narrative to explain natural slumps in form? Sounds like a project to me). Anyway, at 6.5m he's right in that group with a few of the players mentioned above (and below) and his upcoming fixtures and low ownership number may well propel him towards the top of that class.
Status: Buy

Rickie Lambert
There's really nothing to be said about Lambert that hasn't already featured in these very electronic pages. His consistency has been excellent for a mid level option, with 13 games of 5 or more points and 2 points or more every week bar one. He lacks the upside of some of the flashier picks (as evidence by just three double digit games in 29 appearances) but he makes up for it with his penalty duties which help flatten his production and allow him the potential to contribute against any team. Though the results have taken a turn for the worse, Southampton are still creating plenty of chances and scoring goals at a decent clip, and thus Lambert looks well placed to continue his useful fantasy form. The fixtures alone wouldn't necessarily turn me off, but together with his slightly higher price tag, I'd suggest that Lambert is probably towards the bottom of the best mid-level options, but he remains ownable and moving him on isn't a priority if your team is struggling.
Status: Hold

Romelu Lukaku
Another excellent mid level forward, Lukaku has been simply sensational this year and his star would be even brighter had he not been held back through a lack of playing time. For context, the list of players with a better goals per 90 minute rate than Lukaku starts and ends with Javier Hernandez (who has only played for 690 minutes). Lukaku's rate of 0.86 goals every 90 minutes tops Suarez (0.79), van Persie (0.77) and Rooney (0.66) while obliterating his mid-level peers in Sturridge (0.54), Benteke (0.50) and Lambert (0.46).    Just imagine where we'd be if Steve Clark had read the Moneyball piece over at Fantasy Football Scout. As it is, this isn't a great time to invest, with a couple of tricky away trips (Stoke and West Ham) followed by Arsenal at home and then a week off. GW34-37 look great though, particularly if sweetened by a double gameweek, so though I wouldn't necessarily be looking to pull the trigger now, he's certainly someone to keep in mind.
Status: Monitor, and wait for better fixtures

Leon Osman
I like Osman quite a bit and would give him a look were Everton a team that was steam rolling opponents, as they were for a while in the early stages of the season. Now, however, while it's unfair to say they're in poor form, they are certainly more inconsistent than before and thus the fringe players like Osman can sometimes get left out in the cold for weeks on end in terms of points production. I'm still okay owning the alpha dog (Fellaini) in this team but Osman isn't cheap enough to offer value as a bit part player in an increasingly inconsistent team.
Status: Sell

Steve Pienaar
Pienaar is essentially in the same situation as Osman, though he does bring a better assist threat. Still, his price puts him in an awkward spot between those who can be relied on more readily and those who come at a substantial discount. That, plus the aforementioned struggles and upcoming fixture list make the South African a sell too.
Status: Sell

Loic Remy
I had heard about Remy before he came to the Premier League, but had never seen him play until his scoring debut against West Ham. I was impressed then and nothing has countered those initial feelings in the subsequent weeks. His SoT% over the last few games is obviously too high but his 50% rate for the year looks fine and while his G/SoT again looks unsustainable, he's clearly doing enough to offer some value at 5.9m, just be careful you're not assuming he's going to score in three of every four starts. Despite the hype around Redknapp, I would offer a word of caution that under his guidance the team is averaging 37% less SiB that opponent adjusted league average (-7% under Hughes) and this has actually declined of late with a -48% rate over the last 10 games. They may be more "committed", "gritty" or "harder to beat" but in fantasy we generally care about volume of chances and only the truly special players can overcome a lack of chances to become an elite fantasy option. Whether or not Remy is there is questionable but the encouraging point is that while QPR have struggled to create quality chances, a large percentage of them (42%) are falling at Remy's feet and thus he can still succeed despite his teams limitations. The fixtures look good enough and Remy comes at a nice discount, suggesting he too deserves a look in this crowded mid level group.
Status: Buy

Morgan Schneiderlin
Schneiderlin's shot totals look balanced but only in the same way that the return on investment earned on sticking your money under the mattress is consistent every year. Simply put, he isn't a fantasy option.
Status: Sell

Stephane Sessegnon
Sessegnon was one of the players I was interested in coming into the year and while it feels like he never really got going this campaign, his totals will actually end being okay. I'm not crazy about this Sunderland team's consistency and I do fear that they could be one of those teams who simply drift through the rest of the season if the results suggest they are relatively safe in the next couple of weeks. The next six games look pretty tough after this week's visit of Norwich and for 7.1m Sessegnon is too expensive to be spot starting sporadically.
Status: Sell

Gylfi Sigurdsson
Good player, bad situation. Lennon and Dempsey are both said to be close to a return, with Adebayor also rated as 75% to play this week to further cloud the playing time at White Hart Lane. I like this Spurs team and really want another way to access it outside of Bale but Villas Boas hasn't shown too much faith in Sigurdsson when everyone is fit and I'm not willing to bet 7.8m that he changes his mind at this point in the year.
Status: Sell

David Silva
Oh David, what happened? Once hailed as the best fantasy midfielder in the league, his shot totals are now so meager that even his ultra-elite assist potential can only rescue him to the point of justifying a price tag about 2.0m less than his own. I've been waiting for City to turn things around all year, and while they've shown flashes at times, the inability to get their best players on the pitch and in the key positions has cost them and anyone who invested in their stars. The upcoming games don't look great and with plenty of other elite midfield options, the great Spaniard simply isn't an option anymore.
Status: Sell

Luis Suarez
Suarez's trend numbers look a touch worrying, but remember that they are based on just two games worth of data given Liverpool's lack of fixture in GW27. Needless to say he hasn't justified his insane points return over those couple of weeks (it would be impossible to do so based on an average model like mine) but there is really nothing in the data or the narrative to cause any concern here. The only slight knock are the fixtures which see Liverpool face three away games in the next four (all to teams fighting to stay in the league) compounded by a lack of a double gameweek enjoyed by some of his peers. Still, right now you're getting van Persie production for a 20% discount so you can't have everything.
Status: Buy

Carlos Tevez
Meh. Like Silva, it's hard to totally write Tevez off as these players are so talented that they can make a humble blogger look rather foolish, but the odds just aren't in Tevez's favour with his recent 'form' looking to be as much attributable to a higher G/SoT conversion rate than some sustainable skill improvement or tactical revelation. Throw in a continued lack of guaranteed playing time and you have a player whose production is closer to the pack behind him rather than the elite pair ahead of him and thus he's another City prospect I'm out on for now.
Status: Sell

Andreas Weimann
Weimann has done really well when he's played and he actually rated atop the model projections for last week but I filtered him out due to the lack of sample size of his production. He's creating a lot of chances to supplement a healthy shot total, while his ratios have remained solid and consistent throughout his starts. With Bent close to a return and N'Zogbia/Agbonlahor also competing for finite time alongside Benteke, I'm a bit nervous about how often Weimann will play, but for 5.0m he's certainly worth a look and would make a nice third forward for those needing to free up cash to use elsewhere. The playing time risk might make me hesitate about using a transfer on him (for fear you'll need to sell again very soon) but if you're lucky enough to have the wildcard in hand, Weimann is a nice option to supplement your elite picks.
Status: Cautious buy in the right situation

Yaya Toure
The good news is that he's probably offering close to the same value as teammate Silva (with a slightly lower assist threat) but I'm not sure that's even enough to justify his price tag. Yaya seems to be used in an advanced role more sparingly this season and given City's upcoming games I just can't get excited about him anymore.
Status: Sell

Sunday, March 10, 2013

Lineup Lessons Gameweek 29

Aston Villa
Guzan, Lowton, Clark, Vlaar, Baker, Weimann, Bannan, Westwood, Sylla, Agbonlahor, Benteke
Subs: Given, N'Zogbia, Holman, Bowery, Dawkins, Bennett, Carruthers

After a couple of promising outings, N'Zogbia once again found himself back on the bench which only serves to underline that unless your squad is very deep, he's a luxury that can't really be afforded at this stage of the season, even if Villa have been better in the second half of the season. The rest of this team is fairly predictable with Sylla the only odd name here, though obviously not one to cause too many fantasy waves despite a promising first full game. Bent's miserable season continues, though he looks like he could be fit in the next week or so, which mind dent any meagre value offered by Weimann or Agbonlahor.

Liverpool
Jones, Johnson, Carragher, Agger, Jose Enrique, Lucas, Gerrard, Downing, Suarez, Coutinho, Sturridge
Subs: Gulacsi, Skrtel, Shelvey, Allen, Henderson, Sterling, Wisdom

Fans of the teams out in front don't often don't want to hear it, but this Liverpool side are very much for real. For a while their production was severely straggling behind their underlying stats though aside from Suarez taking a few too many ill advised shots from long range, I never saw too many concrete reasons as to why this was the case. We can point to a number of factors as to what has changed in Liverpool's play, but Sturridge's arrival seems to have brought a lot of their good points together and since he made his first start in GW23, they have been simply unplayable, racking up 21 goals from 133 shots (42 on target) - a full eight goals clear of the next best teams (Arsenal and Newcastle). Now, that goal/SoT rate looks unsustainably high so some caution needs to be exercised here before pencilling this side in for another 20-24 goals this year, but when you consider their floor, ceiling and - chiefly - the predictability of where those goals might come from, this is an exciting side to target. We might see the midfield trio of Allen, Henderson and Sterling steal a few minutes here and there for the rest of the season but generally this side looks to be very settled and given the aforementioned factors, should be an important part of your considerations when setting up your team for the final push.

Newcastle
Elliot, Debuchy, S Taylor, Yanga-Mbiwa, Santon, Cabaye, Tiote, Gutierrez, Sissoko, Gouffran, Cisse
Subs: Harper, Anita, Perch, Haidara, Marveaux, Obertan, Campbell

We've observed a significant amount of turnover in this side this year, though for the most part it has been caused by injury rather than manager rotation. Newcastle are still missing a couple of first teamers, with Ben Arfa apparently set to come back sooner rather than later, but otherwise this is likely your team for the remainder of the season. Where Ben Arfa fits in isn't too clear, but if you had to bet you might think it would be at the expense of Gouffran rather than Gutierrez, who's been virtually ever-present this year and brings a bit more of the defensive responsibility that his teammates lack. Sissoko has obviously caught the eye as a legit goal scoring threat since coming on the scene but Cabaye has also quietly racked up 18 shots over the last six games (though with 17 of them from outside the box, I'm not suggesting he's in the class of the some of his peers in that same total shot range).

Norwich
Bunn, Martin, Bassong, Turner, Garrido, Snodgrass, Johnson, Howson, Pilkington, Kamara, Holt
Subs: Camp, Whittaker, Hoolahan, Fox, E Bennett, Becchio, R Bennett

Norwich have added a bit of depth up front in Becchio and Kamara though neither have done too much to really spark our interest and time is probably running out for them to be impactful this year. Holt remains an option if you've been priced out of the Benteke/Berbatov/Lukaku market, though that trio offer substantially higher ceilings than the consistent, if underwhelming, Holt.

QPR
Green, Bosingwa, Samba, Hill, Da Silva, Townsend, Park, Mbia, Hoilett, Zamora, Remy
Subs: Murphy, Wright-Phillips, Mackie, Granero, Onuoha, Jenas, Bothroyd

QPR enjoy good depth on the bench here but Redknapp has shown some faith in the majority of his starters this past month, giving us a couple of players to at least look at. I like what I've seen of Remy and believe he's at least worth a look at 5.9m (preferring him to the aforementioned Holt if you're priced out of that 6.5-7.1m group). Saying that, his lack of volume of chances will always hurt his upside and QPR are always prone to simply throwing a blank out against the league's better sides, but without too many of those fixtures left, Remy might be an option if you need to save a bit of cash to free up space for a couple of the elite options with double gameweeks coming up. Outside of Taarabt's mirage-like stat line, the rest of this midfield hasn't shown us a great deal and thus it doesn't appear to offer much value, even at bargain prices.

Reading
Taylor, Kelly, Pearce, Mariappa, Harte, Robson-Kanu, Akpan, Leigertwood, McAnuff, Hunt, Le Fondre
Subs: Henly, Shorey, Karacan, McCleary, Morrison, Guthrie, Blackman

Le Fondre got his second straight start this week, perhaps signalling that Reading are ready to place an emphasis on scoring goals - an advanced concept that is taking the league by storm. Whether or not there's enough certainty here to warrant investment is still a tricky issue, but as noted above, the key driver for these coming weeks could well be the need to fill your team with elite double gameweekers, so the flexibility a 5.0m forward can offer your team is not to be ignored. The narrative for Robson-Kanu (Reading's 4th top scorer) is almost exactly the same as with Le Fondre, with the midfielder offering nice value in the lowest price bracket, if only given a modicum of consistent minutes. His six shots over the last couple of games place him 6th among midfielders, as do his three efforts on target, yet I wouldn't be too surprised to see one of the other dubious options restored next week, so caveat emptor is the name of the game here.

Southampton
Boruc, Yoshida, Fonte, Hooiveld, Shaw, Fox, Schneiderlin, Rodriguez, Ramirez, Lallana, Lambert
Subs: K Davis, Clyne, Forren, Ward-Prowse, Do Prado, Chaplow, Puncheon

It was a big surprise to see both Clyne and Puncheon benched, as - at least in my limited viewings of Southampton this year - that pair have been a couple of consistent bright spots. What to do with them now isn't clear, though my instinct would be that both will be back soon and thus panic moves are probably best avoided, particularly as their budget prices allow them to be benched without too many issues. The timing of them being benched when most fantasy squads were decimated by bye weeks was however frustrating to say the least.

Stoke
Begovic, Cameron, Shawcross, Wilson, Wilkinson, Shotton, Whelan, Nzonzi, Walters, Jerome, Crouch
Subs: Sorensen, Jones, Owen, Adam, Whitehead, Kightly, Shea

Wilson made his third start since returning from injury and my bet is that he will stay in the lineup at the expense of Wilkinson when Huth is back from suspension (though that one isn't set in stone). If that's the case he will represent excellent defensive value at just 3.9m. Not too much else going on here with the likes of Adam, Jones and Kightly still finding themselves on the outside looking in.

Sunderland
Mignolet, Gardner, Bramble, O'Shea, Colback, Sessegnon, Larsson, N'Diaye, Johnson, Fletcher, Graham
Subs: Westwood, Bardsley, Rose, Kilgallon, Vaughan, Mangane, Cuellar

No real issues to note here though with Bardsley, Rose and Cuellar all fit enough for the bench, we're likely to see some movement along the back line in the coming weeks. Sunderland haven't offered too much value at the back this year, though they've been playing midfielders at full back for essentially the whole season, so the return of this group would be a welcome boost. The fixtures don't offer up too many games where a clean sheet looks too likely though, so any enthusiasm for a quick turnaround should probably be tempered.

Swansea
Vorm, Rangel, Monk, Williams, Davies, Hernandez, de Guzman, Ki, Routledge, Michu, Moore
Subs: Tremmel, Bartley, Dyer, Lamah, Shechter, Tiendalli, Gower

Nothing really to note here and it's another so-so performance from this once-darling Swansea side. I don't want to be down on this team, who have been very good at times this year, but over the last six gameweeks they rank just 14th in total shots, scoring just six times and there is a slight danger of this season petering out a little bit unless they can get back on track. In fantasy terms their current 'struggles' make me apprehensive about getting behind anyone other than Michu, though he himself should probably be somewhat questioned given Swansea's tough set of fixtures.

Tottenham
Lloris, Walker, Vertonghen, Dawson, Assou-Ekotto, Parker, Livermore, Bale, Dembele, Sigurdsson, Defoe
Subs: Friedel, Huddlestone, Gallas, Naughton, Holtby, Caulker, Carroll

With Adebayor and Lennon sidelined, Spurs didn't have too many options here, though I did think Holtby had shown enough to maybe get a start here. This is a useful Spurs team who can score against anyone, though it's still tough to see beyond Bale. The defense looks solid on paper, though with just two clean sheets in the last ten gameweeks, it's not easy to fit any of these options in here other than perhaps Dawson given his lower price tag (no, I don't believe Vertonghen's goal threat compensates for the extra 1.7m outlay).

West Brom
Foster, Jones, McAuley, Olsson, Ridgewell, Yacob, Mulumbu, Dorrans, Morrison, Brunt, Lukaku
Subs: Myhill, Tamas, Popov, Rosenberg, Thomas, Odemwingie, Fortune

Nothing much going on here either. I guess however that this is the point in the weekly post where I note that Lukaku now ranks 6th among all forwards in SiB, despite playing between 30-45% less minutes than most of his competitors. The penalty miss was a shame but the bigger issue remains Shane Long's return from injury which could once again cut into the Chelsea loanee's minutes. 

Friday, March 8, 2013

Gameweek 29 Preview, or, while my fantasy team gently weeps

Look guys, it's going to be carnage this week. I would genuinely take 20 points if offered right now, but given the lack of great options available, I don't believe it's a situation where splashing out on four points is going to really help. Even if you have a wildcard, selling out for this week is likely a bad idea as you don't want to be setup with no players from teams like Man Utd, Man City, Chelsea and Arsenal. In short then, just try and minimise your losses and target selling players who have an off week and whose future prospects you aren't crazy about (Everton players spring to mind here given their upcoming fixtures).

To address a point that will almost certainly come up if anyone links to this page, you'll notice that the all conquering Gareth Bale finds himself well down the rankings despite all the talent missing this week. I understand that this goes against your gut, but please remember that this model is based on statistical averages and is not supposed to simply make every decision for us. If you believe Bale is better than his underlying stats and can unlock a Liverpool defense who have generally been outstanding at home, then feel free to captain him, but until I see something concrete to suggest he can outperform all the facts, I'm happy holding him down a couple of places this week (any suggestions of why he can out shoot that averages are welcomed in the comments section).

Good luck, it's going to be a rough week

Clean sheet rankings
Attacking rankings
Individual rankings

Wednesday, March 6, 2013

What's happening in Salford?

Regular readers will know that one of the key drivers behind my team goal model (which ultimately backs up a lot of the individual player data too) is based on shot totals, split between those inside and outside the box. I like shots on target too, of course, but the reason I have opted for total shots in the model is for fear that over a short period, shots on target can be a bit misleading for individual players as, just as we wouldn't expect a player to score every time he hits the target, we equally wouldn't expect him to actually hit the target every time either.

With that in mind, I want to draw your attention to some interesting shot data from this and last year and open up the floor to suggestions as to (a) what is going on and (b) how we should deal with it next year. You'll notice that a lot of the posts for these remaining few weeks are going to be a bit more 'blue sky' than normal, mainly because there's only so many times you can write up the same lineup lessons are suggest that Michu is due for regression.

To start then, let's look at some likely familiar charts from prior year, first showing the strong relationship between created chances and shots inside the box and then between shots inside the box and goals:

The first chart actually shows a slightly stronger correlation than expected as I thought it was reasonable for teams to rack up 'created chances' without necessarily registering a good shot on goal. Anyway, that strength of relationship suggests created chances should probably feature more heavily in the next model as a way of normalizing SiB. The second chart plots SiB against goals and shows a slightly weaker relationship that I had in mind but an r-squared of 0.72 suggests there's enough here to still rely on. The variance between expected and actual goals per this trend line were as below:

The relationship is far from concrete but generally it looks like the better teams were out shooting the expected goals while the poorer teams struggled to meet expectations (Blackburn and to a degree Liverpool were the key exceptions there. In short though, there's a decent relationship here with a couple of exceptions to keep in mind. Let's move onto this year then:

So far, so good as once again we see a strong relationship between chances created and total goals. This again suggests that the better teams who enjoy good possession in dangerous areas and/or threaten from deep balls into the box are getting more SiB and thus goals. The tricky point comes with the next chart though:

What on earth is happening with Man Utd here (the title of this post was a little dig at United's location in the North West for those outside of the UK)? Again, let's look at the variance from the line of best fit:

Once again United are the largest outlier and the only team who have really shown any consistency to over or under perform the SiB forecast goals over the two seasons (Villa's underperformance looks like the next closest match).

This is the point where the analysis ends and the speculation begins. We can all think of subjective reasons why this is the case for United, but are any of them backed up by stats:

  • They have the best forwards, who are the most clinical and thus convert equally promising shots as other teams but at a higher rate
  • They create better chances than other teams which are thus easier to convert
  • They've simply enjoyed a couple of seasons where they've out scored their underlying stats and thus could regress
  • Something else
With almost two full seasons of data I think we're at a point where it's unlikely that United are simply getting lucky but I haven't been able to locate anything in the data to concretely prove, and the predict what in their game makes them so much more efficient with their chances. Once again, I throw the question out to the readers to see what we can come up with.

Sunday, March 3, 2013

Shot on Target ratio

One of the areas we need to work on over the summer is how shots on target are converted into goals, specifically the split between types of shot. We've observed that, on average, shots on target become goals at a league wide steady rate (30%-33% depending on position), but on an individual level things are more complicated. We consistently see players like Adel Taarabt show up as outliers in that they've taken a lot of shots, hit the target but yet not received the final reward.

One of the obvious starting points is to look at the ratio of shots taken inside and outside the box, and this is something that is already baked into the current model. However, playing around with some data recently, I put together a graphic which hopefully lays out some of the questions that remain unanswered, and I hope to spark some debate to push this forward in the offseason.

SiB Ratio - percentage of shots taken inside the box
G/SoT - percentage of shots on target converted into goals (excluding penalties)

 On first glance, the signs aren't too promising here. We don't observe a particularly strong relationship between the two ratios and we see successful players scattered all over the chart. If we focus on those players who've notched more goals this year (represented by a larger circle) though, we can see something develop, with only a handful of players managing to really defy the logical trend of having a lower G/SoT rate is they taken less shots inside the box:

One thing that could potentially explain some of this variance is the fact that this quartet of players all take a large number of set pieces, all ranking in the top 10 among midfielders:

1. Bale 24 attempts, 0 non penalty goals
4. Gerrard 18 attempts, 0 non penalty goals
7. Cazorla 16 attempts, 0 non penalty goals
10. Lampard 14 attempts, 1 non penalty goal

I must say it wasn't particularly intuitive to me to think that taking set pieces would be a bad thing for a player's stats, but it does seem that in terms of their impact on ratios, scoring a long range free kick does seem harder than scoring a long range effort in open play. I therefore think that for future iterations of the model we need to further distinguish between not just SiB and SoB but also shots taken from set pieces. The issue is that from the Opta data I currently use, we don't seem to have a distinction between shots from set pieces in terms of long range efforts and those where a player gets on the end of a cross from a set piece (which presumably have a decent chance of being converted). As I say, the search for a magic formula continues.

At this point in the season there are more pressing issues to think about (like trying to field a GW29 lineup!) but I want to keep these questions in the background to make sure they are ready to be discussed over the offseason. On that issue, I will be guest writing in a new fantasy publication called InsideFPL which launched last week. At this stage it is still a developing idea so any suggestions or comments would be welcome but I encourage everyone to sign up and give it a read as it could represent an exciting forum to discuss fantasy football and there are some excellent writers already on board.