Stats and game projections for Spurs-Palace

Michael Regan

Every week, I'll be writing up a short statistically-based game preview. What are the likely outcomes? How important is this game to Spurs' top four chances?

Last year, Crystal Palace finished fifth in the Championship with an entirely unimpressive +11 goal difference. That goal difference in the third worst of a promoted Championship side, in the history of the Championship so named. (The worst ever is Hull City, also promoted this year.) I have done a short study of promoted sides, and I've found that while there are no particularly strong estimators for the performance of promoted clubs, goal difference in the previous season is the strongest. So right now, I have Crystal Palace only slightly ahead of Hull City in the projected table, with a roughly three in five chance of being relegated.

To give you an anecdotal sense without showering you in data tables, these are the Championship goal difference numbers of the last fifteen clubs promoted to the EPL, along with their points total in the next Premier League season. (R) marks clubs that were relegated back right off the mark.

Club Season Ch GD PL Pts
Newcastle 09-10 +55 47
West Bromwich Albion 09-10 +41 46
Southampton 11-12 +39 41
Queens Park Rangers 10-11 +39 37
West Ham United 11-12 +33 46
Wolverhampton Wanderers 08-09 +28 38
Reading 11-12 +28 28 (R)
Swansea City 10-11 +27 47
Cardiff City 12-13 +27 ??
Norwich City 10-11 +25 47
Birmingham City 08-09 +17 50
Blackpool 09-10 +16 39 (R)
Burnley 08-09 +12 30 (R)
Crystal Palace 12-13 +11 ??
Hull City 12-13 +9 ??

The relationship isn't determinative, or close to it. but you can definitely see the weaker EPL performances clustering around the sides with lower goal difference numbers in the Championship. From a subjective point of view, Crystal Palace appear to have lost talent as much as they've gained, in moving up a level, and they were by a wide margin the club picked 19th most often in your prediction tables. (They were 19th in about 45% of your tables.)

At the same time, there's a ton of variance in these numbers. We'll know a lot more about Palace once they've played ten or fifteen league matches. And anyone can win a home game, especially on the first day of the season, against a Tottenham Hotspur side that has played a total of one good half of football in the preseason.

My numbers do not account for those subjective concerns, though I do have a quite large home/road effect in my match projection algorithm. So what are the odds?

Match Projections and Odds

Outcome W D L
Odds 63% 23% 14%

Well, that's pretty good. I don't have much more to say. There's a real chance of losing this match, but I think we all basically know that. Even Manchester United last year lost at Norwich. But it would be better if it didn't happen on the first day of the season.

I also have the long-term effects of different game outcomes on Tottenham's likely end-of-season position. This table lists the average points Tottenham accumulated in season with different outcomes at Crystal Palace, along with the chances of the club finishing top four depending on the match outcome.

Outcome W D L
Pts 71 68 67
Top 4% 60% 49% 42%

Experientially, probabilities like 60% and 40% are weird. They sort of matter and they sort of don't. If you had a coin that came up heads 60% of the time and tails 40% of the time, you'd need to log the outcomes in a spreadsheet over a significant sample before you could tell the difference between that weighted coin and a regular one. If Spurs are 60% to finish top 4 or 40% to finish top four, you won't be able to tell the difference over one season. At the same time, if I knew the coin you were flipping was weighted 60/40, and if I were a jerk, I could make some money betting on that coin. So it sort of matters in the long run, but since we're only playing the season once, no one would notice much.

You might also notice that the points numbers are a little bit odd. Why does losing this game make Spurs likely to drop another point over the rest of the season? Why is it 67 projected points for a Spurs side that loses at Palace and 71 for a side that wins?

This is because I slightly randomize team quality in my simulations. I don't run every simulation with every club having the exact same team quality. I use projected team quality as a mean and take a random number sampled around that mean. My logic is, I don't actually know how good any of these teams are. All I have is an estimate. So I let that estimate be an estimate, and allow for real variance.

The effect is the points totals above. In seasons where Spurs got a relatively poor randomized team quality, they were more likely to lose to Palace. And vice versa for seasons where they were relatively strong. So if Spurs lose to Palace, it's somewhat more likely that they're not actually as good as we might hope. Let's hope, instead, that they're as good as we hope.

Not a member? Join Cartilage Free Captain and start commenting | Follow @CartilageFree on Twitter | Like Cartilage Free Captain on Facebook | Subscribe to our RSS feed

X
Log In Sign Up

forgot?
Log In Sign Up

Forgot password?

We'll email you a reset link.

If you signed up using a 3rd party account like Facebook or Twitter, please login with it instead.

Forgot password?

Try another email?

Almost done,

Join Cartilage Free Captain

You must be a member of Cartilage Free Captain to participate.

We have our own Community Guidelines at Cartilage Free Captain. You should read them.

Join Cartilage Free Captain

You must be a member of Cartilage Free Captain to participate.

We have our own Community Guidelines at Cartilage Free Captain. You should read them.

Spinner.vc97ec6e

Authenticating

Great!

Choose an available username to complete sign up.

In order to provide our users with a better overall experience, we ask for more information from Facebook when using it to login so that we can learn more about our audience and provide you with the best possible experience. We do not store specific user data and the sharing of it is not required to login with Facebook.

tracking_pixel_9353_tracker