FanPost

Where will Liverpool place? You decide! (sort of)

I've done a couple of posts on the results of my model that simulates out the rest of the Premier League season, most notably here (you'll find some stuff on the methodology there) and here (some more in-depth methodology stuff there).


With the Spurs game behind us and the rest of the season being down to a somewhat manageable 9 games I've decided to crank the dial up to 11 and simulate the rest of the season 250,000 instead of 5,000 times to allow users to build their own scenario generator.

You can download an Excel spreadsheet with the data here (Warning: Hitting "download" on that page will trigger a 35 MB download-- 250,000 simulations are a lot! That also means that the spreadsheet can not be opened in Office 2003 or earlier, which support only 64,000 lines.)

Following is a description of how to use the sheet, but if you've ever been in an Excel sheet it should be fairly self-explanatory.

The only cells you really need to touch are in the tab "results", G6:O6 (prepopulated with stars). That's where you select from a drop-down menu what you think the results of Liverpool's remaining games will be-- either a Liverpool win, a win for the other team, a draw-- or you can just leave the cell as a star, which leaves the outcome of the game up to random chance-- the odds that the model uses for that can be seen by hovering with the mouse over cells G4:O4.

I have protected the rest of the sheet to keep people who aren't too savvy with Excel to inadvertently hit some buttons that make the spreadsheet not work, but you can unprotect it with the password "liverpool".

Once you have selected the outcomes you think most likely, the odds of Liverpool placing 2nd to 12th place (a better and worse finish weren't found in the 250,000 simulations) are displayed in cells G9:H20 and tallied up into "Champions League straight (2nd+3rd place), CL qualification (4th), Europa League (5th) in cells J9:M12

Importantly, cell A12 shows how often your exact scenario came up in the 250,000 runs. If you picked a really weird scenario that happened less than 100 or so times, the results will obviously not be particularly statistically significant. Take with a grain of salt and consider leaving another game up to chance.

Lastly, there's a problem in a dozen or so of the 250,000 runs that will in some weird edge cases make the chance of placing in some spot a negative, very small number. I'm not sure what's going on there, but those cases are functionally zero. It doesn't matter, affecting literally just one of 20,000 runs. If I'd thought about that before uploading the spreadsheet to Google Drive, I'd have built in a function that floors values at 0%, but it's been uploading for half an hour and I don't want to repeat that.

Let me know what you think and what you come up with.

Below is a screenshot of the sheet, to visualize what I'm talking about.

Liverpoolsheet_medium

Have fun!

Please make any posts here geared towards creating discussion. Any works that are copied and pasted from other sites will be deleted. This isn't a place to advertise services or your blog, it's an extension of our site. Please treat it as such.

In This FanPost

Teams