Projected Premier League table and title odds

Yesterday I laid out my method and asked for your help. Today I've got the projections. I combined your subjective rankings (a little over 60 in total) with regressed, expected-goals-based power rankings from last season to create team quality projections. I then simulated a million seasons. Now I have the results.

Premier League Projected Table and Odds

So what have we got here? First, I think it's much more instructive to look at the "%" columns than the overall table. Spurs do project 4th ahead of Arsenal—Tottenham had slightly better underlying numbers last year—but the difference in percentage chance of Top 4 is really rather negligible. 55% vs 52%. That's a dead heat. Likewise, there isn't much of significance in the gap between Chelsea and United in second and third, and really the run from Chelsea down to Liverpool is pretty close to a tie. There's no way for a projection engine to be accurate past a delta of 3-4 points per team, and that's enough to cover the gap.

This is perhaps particularly important when looking at the middle of the table. Yes, maybe it's sort of odd that Newcastle stand in 10th place after their weak season, but the difference between 10th place Newcastle and 14th place West Ham is too small to draw any conclusions from. Newcastle has 10% chance of relegation by these numbers, West Ham 15%.

Let's Talk Bloomberg

At this point preseason, I think everyone's mostly guessing, me and the big boys at Bloomberg. The variance in any season is huge, and putting a percentage point number on outcomes is an exercise in hubris. I have tried to limit the appearance of confidence in these numbers by rounding everything to whole numbers. Bloomberg has emphasized their proactive synergy by displaying results down to the decimal point, as if somehow that extra .1% chance of Chelsea winning the Premier League was a meaningful figure. In general, if someone hands you a projection, ignore everything after the decimal point. That's just rhetorical, an attempt to convince you that the person who handed you the projection is smart and knows about numbers. It isn't actually information.

But in any case, most differences between our projections and real outcomes can probably be attributed to a combination of random chance and the infinite number of complicated football facts that we did not and perhaps could not account for before the fact.

Nonetheless, I think there are some interesting differences. I'm sure you noticed the first one, which is Chelsea. City and Chelsea are most people's pick to win the EPL this year, but I have a pretty big gap between Chelsea and City. This is for a simple reason. Last year, City had the best underlying stats in the Premier League—slightly better than league champions United—while Chelsea's underlying numbers were unimpressive. They had fewer shots on target in the box than either Spurs or Arsenal, and allowed many more. Chelsea did do an excellent job preventing big chances, but even with that added in, I have them overall as a worse attacking and defensive side than the 4th/5th place finishers. Because a significant part of my projections are the club's stats from the previous year, Chelsea's excellent subjective placement in the table is only enough to lift them to the head of the trailing group, not up to even with Manchester City.

The other thing I see, looking at these tables, is that my numbers rate the league as significantly more wide open. Bloomberg thinks that Chelsea, City and United are unstoppable forces headed for huge nearly +50 goal difference numbers. No clubs last year had a goal difference as high as City, United and Chelsea are projected by Bloomberg. So they have those three dominating the title race (combined odds about 90%) and the top four positions (average Top% likewise 90%). My numbers are a little bit more flattened, perhaps in part because of regression to the mean. I also have Tottenham. Arsenal, and Liverpool as capable possible competitors for at least Chelsea and United, which makes for a notably wide-open race for the top four positions and creates the possibility of an upset title winner.

I'm not sure I necessarily like my numbers better. City and Chelsea really did spend ludicrous amounts of money buying great players and managers. I'm not sure they should have a 15% or 35% chance of missing the top four. At the same time, I think a reminder that one should have a little bit of skepticism toward Chelsea is worthwhile. They were good last year, not great. They've made good additions, but haven't City, who were better to begin with, done at least as well? Obviously there's the Mourinho factor. My guess is that Bloomberg's numbers are driven by some sort of Mourinho adjustment, and I'm not making any special allowances beyond the subjective factor included from your tables. We shall see.

Indeed, with all of this, we shall see. Aesthetically, I like the open season projection a little better, so I'll go with it.

In conclusion, I think it is likely that one or both of Hull City and Crystal Palace are terrible.