2019-09-10

Super Bowl LIV Projection, Week 1

Baltimore over Green Bay


What's this? Here's a quick rundown originally taken from 2014's introductory post.
Each year I project the Super Bowl match-up after each day of games. The method is not of my own design, but the implementation is. Each team's strength is determined based on points scored and points allowed, and scaled to make the average of all teams .500. Then, the remainder of the season and playoffs are simulated millions of times, and I report to you the chances of each team reaching each round of the playoffs. It's had mixed results in the past, but that's the nature of taking 1 weighted random sample out of millions of potential outcomes.
In week 1, the top teams are the ones with the biggest point differential, which could simply be a function of playing one very bad team. New England won 33-3. They might be the strongest team, or Pittsburgh might be the weakest team. Or both could be true. Pittsburgh avoids being lowest on the list because of a currently weak division, 1-3 in aggregate. Also, I only show 3 decimal places here, but with a few more New England has the higher strength by just a little bit over Baltimore, who won 59-10, and is the AFC favorite for now, just like in week 1 last year. Also like week 1 last year, we have a tie. I really don't like ties that much, because some of the other hastily coded numbers I output for my own use assume a whole number of wins and losses summing to 16. But, Arizona and Detroit have uglied the charts just like Cleveland and Pittsburgh last year.

Also, because I know you're going through my post history, you'll see that the week 1 leader always has a .530 or .531. Why is that? New England, to 5 decimal places, is .53121. The highest possible in my system after one week is .53125, which is 1/16 of the way from .500 to a perfect 1.000, and is achievable only by a shutout, but by any score by the winning team. That's because the strengths are regressed back to .500 a lot at the beginning of the year, so that something like a 3-0 win doesn't result in a team that appears completely dominant.

Anyway, things are mostly even as usual, with a few outliers. Chicago is the rare team that is below 1% likely to win after just the first week. It's not unprecedented, but it doesn't happen all that often. Buffalo 2018 and New York Giants 2017 were both listed at 0.97%, and they wound up 6-10 and 3-13, respectively. Don't hold out much hope for Chicago, is what I'm saying.

TEAMRecordStrengthDivPlyffWCQtrSemiChamp
BAL1-00.53149.8161.0244.1423.1312.086.31
TEN1-00.53047.2259.8943.4522.7011.826.14
NE1-00.53141.5657.9042.0522.0711.526.02
GB1-00.53036.7253.0137.9219.8910.435.40
DAL1-00.52540.6154.2938.4519.9510.345.30
MIN1-00.52735.0752.1937.1919.4010.105.22
SF1-00.52330.9350.2635.6618.459.504.87
OAK1-00.51731.3750.5335.0817.849.044.61
KC1-00.51829.7149.2234.1717.428.874.51
LAC1-00.51029.2648.6833.4316.728.364.19
PHI1-00.50835.0948.9433.3516.688.344.14
NO1-00.50342.5951.2034.3316.928.334.10
BUF1-00.50332.5948.7433.0716.278.003.96
LAR1-00.50526.2044.9030.6915.277.573.73
SEA1-00.50224.7742.4328.6814.177.003.44
ARI0-0-10.50018.1033.7221.9810.805.302.59
DET0-0-10.50019.7333.1121.3810.525.192.53
CIN0-10.49820.4530.7319.139.304.522.20
HOU0-10.49718.4828.8317.978.724.252.07
CAR0-10.49521.7328.7717.448.464.101.99
NYJ0-10.49714.3226.4616.357.963.871.90
IND0-10.49018.1927.5616.718.003.831.85
WAS0-10.49213.3223.0514.036.783.281.58
JAC0-10.48216.1124.6914.656.893.251.55
TB0-10.47718.3824.6314.316.683.111.45
DEN0-10.4839.6621.0912.575.932.801.33
ATL0-10.47317.3123.0913.306.162.841.32
CLE0-10.47015.1322.3212.845.892.701.25
PIT0-10.46914.6122.0012.625.772.641.22
NYG0-10.47510.9919.4911.415.312.471.15
MIA0-10.46911.5320.3211.765.392.471.13
CHI0-10.4708.4916.949.874.562.100.97

No comments:

Post a Comment