Add clinching or elimination scenarios?
First off, thanks for a fantastic site!
My question is, do you have any plans for providing potential clinching or elimination scenarios along with projection data? This is easy to find on other sites for the NFL, but I would love for this site to be my one-stop shop for all things playoffs. It would also be very handy for fantasy football leagues that we have uploaded to your site.
kendroberts, what if you add a few worst-case scenario seasons for each team? You already simulate "what if team T loses every remaining game?", this is just going a little farther. In these seasons, you would not only ensure T loses every game, but also try as hard as possible to make other teams catch up to them.
For an example, let's look at T = Boston Bruins in today's NHL standings.
Let A(T) be the set of teams ahead of T in the conference, currently 2 teams in our case.
Let B(T) be teams behind T, but which have enough points available to potentially catch up, it looks like it's 11 teams.
Eight teams advance from the conference, but |A(T)| = 2 are already ahead, so if 8 - 2 = 6 more of them catch up to Boston, Boston will be out.
Now for one worst-case scenario, let W(T) be a subset of 6 teams chosen from B(T). Then run a season where Boston always loses, teams in W(T) always beat teams outside of W(T), and other games are weighted as usual.
You could do perhaps 50 such worst-case seasons for each team T, with a different selection of W(T) from B(T). You probably want to weight the selection of W(T) so that the strongest teams from B(T) are much more likely to be selected. If T makes the playoffs in every such season, you can consider them to have clinched a spot.
Hello Paul, you are most welcome.
I would love to show clinching scenarios. That would be wicked cool. The times I've thought about it in the past I wasn't able to see a clear path to automating it.
I can't just brute force go through all the w/l/t combinations because even with only 32 games left that's 3^32 = 1,850 trillions combinations. To make the program smart enough to pair that down, like you would when figuring it out by hand, seems hard (for me anyway).
Maybe I could use the existing simulation but remember all the game results (each simulated season) pared with weather the bubble teams made it or not. I might be able to get that working if I came up with a way to "summarize" all that data along the way as it was cranking through the millions of seasons, too keep it from growing huge and slowing down the sim.
Or... maybe I could attack it from withing the tiebreakers, tracking which ones I saw, who came out ahead and why.
Anyone have ideas?
TLDR: Great idea and I don't yet know how.