Premier League Power Rankings and Projections, Week 11: Wide Open

It's wide open. Manchester City and Tottenham Hotspur both lost matches they should win, and while Chelsea were gifted a draw in the final minute of extra time, that's still two dropped points for the Blues. Arsenal had a chance to open up a huge lead in both the title and champions league races, but they put in a drab, tired performance at Old Trafford and came away deserved 1-0 losers.

The big winners from last weekend are the clubs that just didn't screw up. Liverpool won easily home to Fulham and Southampton did the same home to Hull City. It takes a lot in my database for Fulham to actually project to get worse, but this did it. Liverpool put eight shots on target from inside the box, Fulham none. Liverpool created five big chances, Fulham again none. Southampton likewise beat up on Hull City. Both clubs now look like legitimate top four competitors.

The outcome is that my projections give eight clubs a greater than 10% chance of making the Champions League. That seems like a lot, doesn't it? Could be a crazy season.

I think I've written sufficiently on Spurs' crazy loss home to Newcastle, but as you can see the underlying stats from that loss so improved our projection that we only dropped a negligible couple points in our projections. The stats are saying, if Spurs keep creating goal threats as the club did against Newcastle, it's likely that the outcomes of these attempts will be better and we'll win a few extra matches.

Projections and Power Rankings

Do remember that because of rounding, not all the numbers necessarily add up quite right.

• Expected Goals methodology
• Season Simulation methodology

• Average points for positions of importance. Title: 81 points; 4th place: 70 points; 17th place: 35 points.
• That 17th place number is low compared to the points of the 17th place club in recent seasons, but it wouldn't be much of an outlier historically. West Ham in 2010 and Hull City in 2009 both escaped relegation on 35 points, and West Brom in 2005 stayed in the Premier League with just 34 points. My stats think the bottom five clubs in the EPL are really bad (Team+ under 70), as was the case in those earlier seasons. Odds are good that three of those five very bad teams end up having 30ish point seasons, allowing someone else to escape with well under a point per match.
• Fulham have shockingly awful numbers everywhere you look. They've not only allowed the most shots on target from inside the box in the Premier League, when I break it down with the Shot Matrix zones, they've allowed the most Danger Zone SoT in the league and the most SoT from each of Zones 1, 2, and 3 individually as well. I am kind of shocked that Martin Jol keeps running the same basic club out there to get steamrolled every week. I think the wonderfully avuncular Jol is precisely the wrong fit for this club, they need a ruthless sort to tell the high-priced over-30 starters that they're on the bench until the defense improves. Run out the kids, anyone, it can't be any worse than this.
• If you've been following my articles, you know I've got this fun new Shot Matrix database, from which the above numbers were drawn. My power rankings and projections here are still running based on SiBoT, SoBoT and big chances. This is because I want to renoobulate my team quality stats all at once, using the best regression methods I know. That will take time, and I'm still working through the database. So it won't be for another month, probably, before I have a totally new expected goals / expected points method. I'm not expecting particularly large divergences between the methods, but I guess I won't know until I've built the new one.
• Arsenal.
• I pointed out last week that Norwich and Aston Villa had played the two most difficult schedules in the league by far through ten matches. It gets better for them from here. Both clubs obliged with good wins over mediocre to poor competition. Thanks guys.
• You might notice that Manchester City and Sunderland moved less than you'd expect given the shocking result of the weekend. This is not because Manchester City should have won, they totaled zero SiBoT at Sunderland and the underlying stats reflect a roughly even game. Rather, it's because of my methodology. Every week, I not only add this week's results in to my team quality formulas, I also increase the weighting of 2013-2014 stats in the projections. Sunderland have been much worse this season than I originally projected, and City have been a lot better. So while their 2013-2014 numbers moved in the expected directions this week, increasing the weight on their overall 2013-2014 numbers also means increasing the weight given to City's demolishings of Norwich and Newcastle, and Sunderland many and varies crappy matches. So it adds up to lost points for City and gained points for Sunderland, but fewer that you'd probably expect.
• Crystal Palace drew home to Everton. That's nice for them, but they're still getting relegated.