Ok, lot of probabilities this week. If you’re looking for context on this, here’s Wednesday’s explanation of how it works. The short version is that we rigged up a spreadsheet that uses FiveThirtyEight’s SPI to try to estimate Burnley’s relegation probability given different combinations of results in the games tomorrow.
We’ll start with the simple one, which is the result over which Burnley has some control. Here are our estimates if Burnley wins, loses, and draws tomorrow, independent of the results of Leeds and Everton:
Burnley Result | Burnley Relegation Probability |
Win | 4% |
Draw | 26% |
Loss | 36% |
Pretty encouraging, honestly. With a win or a draw, the probability drops (it’s currently at 31%, per SPI). Even with a loss, it only rises slightly. Then, though, Leeds and Everton play.
What Everton does probably doesn’t matter for Burnley. Everton’s probably safe. So, let’s look at the Leeds-Burnley combo’s:
Burnley Result | Leeds Result | Burnley Relegation Probability |
Loss | Win | 70% |
Draw | Win | 61% |
Loss | Draw | 46% |
Draw | Draw | 17% |
Loss | Loss | 17% |
Draw | Loss | 12% |
Win | Win | 11% |
Win | Draw | 5% |
Win | Loss | 0% |
Even with a loss, so long as Leeds doesn’t outright win, Burnley’s on the right side of 50% in our estimates, pointing to the value of the extra midweek game Burnley has ahead and Leeds does not.
This is also exciting because it points out that as long as either Burnley wins or Leeds loses, Burnley’s up to at least roughly a 5-in-6 survival probability. Russian Roulette odds, which aren’t that bad!
Now, the big one. An important caveat we’ll repeat below: Some of these scenarios are so unlikely that because we only run 10,000 simulations, we have a very small sample, creating a lot of noise on the probabilities. The margin of error is very high with the unlikelier scenarios. Please remember this. Please.
Leeds Result Sunday | Everton Result Sunday | Burnley Result Sunday | Burnley Relegation Probability | Scenario Probability |
Win | Draw | Loss | 73% | 4.4% |
Win | Win | Loss | 72% | 7.4% |
Win | Loss | Loss | 67% | 8.0% |
Win | Win | Draw | 64% | 1.1% |
Win | Loss | Draw | 60% | 1.0% |
Win | Draw | Draw | 59% | 0.6% |
Draw | Draw | Loss | 49% | 3.2% |
Draw | Win | Loss | 48% | 5.8% |
Draw | Loss | Loss | 42% | 5.7% |
Draw | Win | Draw | 21% | 0.6% |
Loss | Win | Loss | 18% | 16.2% |
Win | Draw | Win | 17% | 0.6% |
Draw | Loss | Draw | 16% | 0.7% |
Loss | Loss | Loss | 16% | 18.2% |
Loss | Draw | Loss | 16% | 9.3% |
Loss | Win | Draw | 13% | 2.2% |
Draw | Draw | Draw | 13% | 0.5% |
Loss | Draw | Draw | 11% | 1.1% |
Loss | Loss | Draw | 11% | 2.3% |
Win | Win | Win | 10% | 1.0% |
Win | Loss | Win | 10% | 1.4% |
Draw | Win | Win | 7% | 0.8% |
Draw | Loss | Win | 4% | 0.9% |
Draw | Draw | Win | 2% | 0.5% |
Loss | Loss | Win | 0% | 2.8% |
Loss | Win | Win | 0% | 2.4% |
Loss | Draw | Win | 0% | 1.3% |
Again, the estimated relegation probabilities for unlikely scenarios could be off by a good amount, because our simulations only turned up the smallest sample for each of those scenarios.
The thing I’m personally taking away from this is that there’s something like a 63% chance Burnley’s relegation probability drops, and something like a 37% chance Burnley’s relegation probability rises. The bad cases are worse than the good cases, but they’re less likely to actually transpire tomorrow.
Here’s that same table, but with results that occurred in fewer than 200 simulations filtered out:
Leeds Result Sunday | Everton Result Sunday | Burnley Result Sunday | Burnley Relegation Probability | Scenario Probability |
Win | Draw | Loss | 73% | 4% |
Win | Win | Loss | 72% | 7% |
Win | Loss | Loss | 67% | 8% |
Draw | Draw | Loss | 49% | 3% |
Draw | Win | Loss | 48% | 6% |
Draw | Loss | Loss | 42% | 6% |
Loss | Win | Loss | 18% | 16% |
Loss | Loss | Loss | 16% | 18% |
Loss | Draw | Loss | 16% | 9% |
Loss | Win | Draw | 13% | 2% |
Loss | Loss | Draw | 11% | 2% |
Loss | Loss | Win | 0% | 3% |
Loss | Win | Win | 0% | 2% |
This still covers about 88% of possibilities tomorrow, making it arguably the most useful table. It demonstrates the value of Everton losing, and how that only matters if Burnley doesn’t win and Leeds gets at least a point (the Burnley relegation probability being higher with an Everton draw than a win in some cases is an example of the small-sample-size noise we talked about above). It also reminds us again of the combination that would lead to Burnley being officially safe: Namely, a Burnley win with a Leeds loss, as we saw above. If Burnley wins and Leeds loses, it’s over barring Leeds making up a ridiculous goal difference, which will not happen.
Up the Clarets, down the Whites, if I’m remembering correctly that Leeds is called the Whites. Sorry, Paul.