It’s that time of year again when managers are advised to fasten their seatbelts, and hold on tight, as the FPL schedule is about to become hectic. Next week sees the first set of midweek fixtures this season, which comprise the sandwich filling of 3 gameweek deadlines in 9 days. Consequently, I’ll not have time to produce this blog for the next 2 gameweeks, so I’ll try to include enough pointers to them here to keep you going until I return for GW15.
In the interest of transparency I again attach 3 appendices at the end of this blog showing how my spreadsheet’s predictions fared in the previous week compared to other better known predictive models and algorithms. Despite bettering our season average so far, my model’s match result predictions last week were 3% less accurate than those of fivethirtyeight.com (see Appendix I), with that site drawing level for the season on a average mean absolute error of 0.36. Fivethirtyeight.com were also the only one of the three models I compare my spreadsheet’s score forecasts with to outperform them last week (see Appendix II). My model was the only one to get 2 scorelines correct though, which might have been 3, but for a “disgraceful” decision to award SOU a late penalty for a foul that happened outside the box. Whilst my spreadsheet’s clean sheet probabilities again proved more accurate than the bookies best odds posted by @FPL_Salah, they were less accurate last week than those same odds with the effects of bookies margins removed by @fplreview (see Appendix III). Suffice it to say that my spreadsheets are still counting down to the launch date of GW17, which is the point they will be fuelled by this season’s data only.
Meanwhile, there are five teams deemed more likely than not to score at least 2 goals in GW12, and maybe they can act as the crawler transporter moving my spreadsheet predictions to the launch pad.
All five are shown in the table below as expected to win their matches this weekend, but the probability of a 1 – 1 draw is actually higher than a 2 – 1 win for both WHU and LEI.
Apologies if you’re new to my blog, but after eleven weeks of doing so, I cannot bring myself to account for this apparent paradox to my readers yet again. They must be as bored of reading it as I am of writing it, so please refer to previous blogs for an explanation.
Unusually, there are only 3 correct score forecasts in GW12 that are not relatively straight forward. The one not already mentioned is the Manchester derby, which is reckoned to be 0.1% more likely to finish 2 – 0 than 3 – 0 (see table below), despite the Citizens being deemed 0.3% more likely to score 3 goals than 2.
Note that the probability given for the visitors winning 4 – 0 is more than double the chances of MUN winning (see table above). Old Trafford threatens to be more like the Theatre of Nightmares this weekend for Bruno owners. Let’s get #prayforsolskjaer trending, people!
LEE vs WHU, and LEI vs BHA are set to be much more evenly contested if the correct score forecast probability matrices below are anything to go by:
Here are the projected correct score forecast tables for GW13 and GW14:
The table below shows how many points players are expected to get on average (excluding bonus points, and yellow/red cards). The red ‘P’s indicate where predicted scores have been inflated by the effects of recent penalty kicks, and the numbers in the column to the right of expected points show what these scores would have been if the distorting effects of penalties are removed. In other words, these are the average expected points assuming there are no opportunities to score from the penalty spot.
KDB, Salah, and Vardy were among the top half dozen ranked attacking players in this table last week, and are so again this week if we follow Pep’s example, and ignore Torres and Foden.
As can be seen in the column furthest to the right, the points boost KDB receives from spot kick duties is significantly larger than is the case for Salah. If we knew in advance that neither player will be called upon to score from the penalty spot in GW12, then Salah‘s points prediction of 7.6 points would be a full point higher than KDB‘s. In the absence of penalties, however, Mane‘s points prediction actually edges Salah‘s by 0.1.
Wilson and Watkins are the highest rated forwards this week, trumping the more premium strikers Vardy, Kane, Jesus and Werner.
Most of the players mentioned above feature near the top of my table for predicted points between now and my next blog (see below). The main worry for owners of Salah, Mane, and Robertson during this period has to be rotation, especially with the GW14 fixture away to Palace coming just 62 hours 30 minutes after the “six pointer” vs Spurs in GW13. Remember, 72 hours is considered by sports scientists to be the minimum safe recovery time for footballers, and Klopp made the wrong kind of headlines the last time his team was faced with a similar situation.
Few fixtures hold more appeal currently than a home game vs WBA, but coming just 73 hours and 30 minutes after the Manchester derby, and only 65 hours before an away game against in-form SOU, who beat them in the corresponding fixture last season, the Pep Roulette threat level has been raised to Severe!
In my opinion, Wilson, Watkins, Kane and Grealish are the players in the table above likely to play the most minutes due to the recovery times between their games being least unfavourable.
My new rolling data charts made an immediate impact in the captaincy debate last week, emphatically challenging the story told by recent underlying stats that WOL were one of the league’s best defences. With time against me this week, I’m going to present a similar chart for FUL vs LIV in the hope of being spared having to run through my FFS stats tables. I did several charts last week and none looked remotely like the one below!
The chart above shows two trendlines, one for the home defence strength weighting for FUL showing how many goals they are expected to concede on average vs an average away attack, and the other, for the away attack strength weighting for LIV showing how many goals they are expected to score on average vs an average home defence. These ratios are such that an average team would be expected to concede/score one goal vs another average team, but as can be seen, the Cottagers have a much worse than average home defence, and the Reds have a much better than average away attack. In none of my charts last week was a team’s defence strength weighting trendline above a team’s attack strength weighting trendline.
Last week, LIV had the highest attack strength weighting, and WOL the worst defence strength weighting, but it’s not quite as clear cut this week. There are actually 3 teams with a higher attack strength weighting than LIV this week (MCI, CHE, and WHU), but FUL have by far the worst defence strength weighting of any team, and so, the case for a Salah (or Mane) captaincy looks open-and-shut to me.
Of the four defenders who appeared in my players points table last week, only Robertson justified his place there, and he is one of only three defenders this time around, with Cancelo and Dias above him despite facing a Manchester derby at Old Trafford. Up against a team who did the league double over them last season, MCI are nevertheless reckoned by my spreadsheet to be more likely than not to keep a clean sheet. In order to do so, Pep will need to have learned his lesson with regard to MUN‘s counter-attacking threat.
The chart below validates the clean sheet prediction for Pep’s men. The red trendline describes the sharp decline of MUN‘s home attack strength weighting, followed by a levelling off that is well below average, while the blue trendline shows MCI‘s away defence strength recovering, and consolidating, its position as the best in the league.
MCI continue to dominate my longer term clean sheet table, with good to strong chances of adding to their tally of 4 clean sheets so far in 4 of the next 6 gameweeks.
The patience of Martinez owners who held onto him despite AVL‘s blank last weekend looks set to be rewarded soon enough, as AVL are ranked second best for clean sheets over the next 6, with good to strong prospects in half of them. Meanwhile, LEE draw nearer to the 3 good clean sheet chances in a row (GW13-15) signposted last week, and Dallas and Ayling owners probably ought to stick to their guns.
Robertson and Cancelo continue to be the only defenders predicted to score at least 20 points (before bonuses) over the next 4 gameweeks (see table below).
That’s all for now, folks. May the GW12 flop, GW13 turn, and GW14 river be with you.
Coley a.k.a. FPL P0ker PlAyerFollow @barCOLEYna
Appendix I – mean absolute error of result probabilities
Appendix II – mean absolute error of predicted goals
Appendix III – mean absolute error of clean sheet probabilities