2016-12-20

Superbowl LI Projection, Week 15, Final

New England over Dallas

Carolina refused to eliminate themselves with a loss last night, so I felt compelled to simulate todays' runs with ties included. I implemented that in my software a few years ago, but it's a parameter I can set when I run them, and I usually leave it at 0 to hit as many edge cases as possible just involving wins and losses. But, these numbers include a 1% shot of any given game ending in a tie. There's probably a smarter way to do this, having closely matched games more likely to end in a tie and mismatches less likely, but I haven't quite worked out the math on that yet. For an approximation though, I'd say 1% is a decent to generous overall estimate. Carolina is still less than 0.005% likely to make the playoffs.

I had a well-reasoned comment from Josh D on yesterday's post, asking why I sort on strength instead of one of the columns that represents a team's chances, like making the playoffs, or winning the Superbowl. That would bump all the eliminated teams to the bottom. In an ideal world, I'd figure out how to get JavaScript table sorting working, much like what you see on many Wikipedia tables. I've got the script, but the last time I tried to make it work on blogger it didn't take. I'd kind of like to see both views, to be honest. This one shows something interesting, that it's not always the "top 12" teams that make the playoffs, either due to divisional or conference alignment, or unluckiness of results despite a theoretically strong team (like 5-9 Philadelphia that is thousandths of a point stronger than 9-5 Detroit). On the other hand, it'd be nice to see a view that shows me New England, Dallas, and Seattle in order as the teams most likely to win the Superbowl, instead of stronger-but-more-weakly-positioned Atlanta in their midst. Assuming it's not a ton of work, I think I'll change the sorting starting next week and see how I like it.

I've been thinking about fully revamping my software for awhile now, to add a few features, make the C++ less terrible, and possibly make it generic enough to handle any league via configurable tiebreakers and playoff spots. I'd also make it handle numbers bigger than 32-bit, so I can do more than 2.1 billion runs. To allow for that, I'd probably make it multi-threaded, or even distributed, but that depends on me finding a whole bunch of free time.

TEAMRecordStrengthDivPlyffWCQtrSemiChamp
NE12-20.756++++++70.5748.2029.70
DAL12-20.71097.18++98.9865.2238.2720.63
ATL9-50.66782.1197.2169.7936.5318.979.33
SEA9-4-10.649++++89.6948.5723.5711.12
PIT9-50.63470.3887.0753.4827.4112.155.77
KC10-40.59833.7697.3768.4030.9612.795.60
BAL8-60.59729.6245.0522.8210.074.261.87
DEN8-60.595-16.908.112.621.350.58
BUF7-70.585-1.110.500.160.080.03
OAK11-30.57466.24++85.0139.4214.596.06
NYG10-40.5552.8299.8150.3918.587.462.83
GB8-60.54442.9455.8126.329.333.641.35
IND7-70.5433.504.111.890.650.250.10
PHI5-90.536------
DET9-50.53657.0677.4438.5413.865.251.90
TEN8-60.53352.9453.8824.388.573.161.19
ARI5-8-10.529------
NO6-80.523-0.130.060.020.010.00
MIN7-70.512-3.931.560.480.190.06
WAS7-6-10.504-22.108.462.490.960.32
MIA9-50.502-49.6720.966.062.280.80
SD5-90.500------
CIN5-8-10.489------
TB8-60.48217.8943.5716.224.911.680.53
CAR6-80.472-0.000.000.000.000.00
HOU8-60.39643.5644.8314.453.500.880.23
CHI3-110.341------
JAC2-120.304------
NYJ4-100.270------
SF1-130.224------
LA4-100.219------
CLE0-140.182------

First to 10% chance of Superbowl win: New England (Week 3, Thursday)
First to 20% chance of Superbowl win: New England (Week 8, Sunday)

First to 80% chance of Playoffs: New England (Week 3, Thursday)
First to 90% chance of Playoffs: Minnesota (Week 4, Final)
First to 99% chance of Playoffs: New England (Week 8, Sunday)
First to 99.9% chance of Playoffs: Dallas (Week 12, Thursday)
First to 99.999% chance of Playoffs: Dallas (Week 13, Thursday)
First to clinch Playoffs: Dallas (Week 13, Sunday)
First to clinch a First Round Bye: New England (Week 15, Sunday)

First team eliminated from winning a division: Cleveland, San Francisco (Week 11, Sunday)
First team eliminated from the playoffs: Cleveland (Week 11, Sunday)

[Week 1] San Francisco over Houston
[Week 2, Thursday] San Francisco over Houston
[Week 2, Sunday] Pittsburgh over Minnesota
[Week 2, Final] Philadelphia over Houston
[Week 3, Thursday] New England over Philadelphia
[Week 3, Sunday] Philadelphia over New England
[Week 3, Final] Philadelphia over New England
[Week 4, Thursday] Philadelphia over New England
[Week 4, Sunday] Philadelphia over Denver
[Week 4, Final] Minnesota over Denver
[Week 5, Thursday] Minnesota over Denver
[Week 5, Sunday] Minnesota over Pittsburgh
[Week 5, Final] Minnesota over Pittsburgh
[Week 6, Thursday] Minnesota over Pittsburgh
[Week 6, Sunday] New England over Minnesota
[Week 6, Final] New England over Minnesota
[Week 7, Thursday] New England over Minnesota
[Week 7, Sunday] New England over Minnesota
[Week 7, Final] New England over Minnesota
[Week 8, Thursday] New England over Minnesota
[Week 8, Sunday] New England over Dallas
[Week 8, Final] New England over Dallas
[Week 9, Thursday] New England over Dallas
[Week 9, Sunday] New England over Dallas
[Week 9, Final] New England over Dallas
[Week 10, Thursday] New England over Dallas
[Week 10, Sunday] Dallas over New England
[Week 10, Final] Dallas over New England
[Week 11, Thursday] Dallas over New England
[Week 11, Sunday] Dallas over New England
[Week 11, Final] Dallas over New England
[Week 12, Thursday] Dallas over New England
[Week 12, Sunday] New England over Dallas
[Week 12, Final] New England over Dallas
[Week 13, Thursday] Dallas over New England
[Week 13, Sunday] New England over Dallas
[Week 13, Final] New England over Dallas
[Week 14, Thursday] New England over Dallas
[Week 14, Sunday] New England over Dallas
[Week 14, Final] New England over Dallas
[Week 15, Thursday] New England over Dallas
[Week 15, Saturday] New England over Dallas
[Week 15, Sunday] New England over Dallas

No comments:

Post a Comment