Sweet New Dashboard
First, I'd like to announce the release of my 2013 fantasy football dashboard. It's public this year, in javascript and full of all sorts of interesting data. I included data obtained from:
- Statistics obtained via NFL.com
- Mock Draft data collected from a certain mock draft site, which gives us a draft distribution per player and an average ranking
- Rankings by the top 3 sources, CBS, ESPN, and FFC and the ADP (Average Draft Position, which is just the average ranking and a mediocre measure
- A sparkline of upcoming defensive matchups (see below)
I really wanted to include some of football outsider's KUBIAK data but I am pretty sure it is copyrighted and I want to respect their copyright. Bummer because it's probably the best set of predictions out there.
Defensive Points Against per Game (PAPG) Predictions
So I needed a quick and dirty way to predict which teams were going to have good defensive performances this year and which were not. I could have really gone deep into player and coordinator level data but I don't do this for a living, so I didn't have time. In the short term, I put together a (mediocre) model which seems to capture at least the approxmate rank ordering okay. Here are the picks.
team predicted.fit predicted.lwr predicted.upr
SF 18.32589 17.19841 19.45336
PIT 19.46592 18.49606 20.43578
SEA 19.72042 18.61981 20.82104
GB 20.60593 19.88312 21.32873
MIA 20.62056 20.20545 21.03568
BAL 20.70957 19.91464 21.50449
CHI 20.88071 20.13641 21.62501
CLE 21.21339 20.75442 21.67236
CIN 21.30532 20.71994 21.89070
HOU 21.36536 20.54220 22.18852
ATL 21.58513 20.72699 22.44326
KC 21.91856 21.21568 22.62144
NYJ 21.96500 21.45735 22.47265
NE 21.96581 21.28860 22.64301
WAS 22.10506 21.46445 22.74566
ARI 22.15112 21.37068 22.93156
NYG 22.56559 22.02549 23.10569
NO 22.63798 21.80729 23.46866
MIN 22.73069 21.98883 23.47255
STL 22.75580 22.08969 23.42192
SD 22.90059 22.06447 23.73671
DET 23.08472 22.36155 23.80788
TEN 23.15544 22.23376 24.07712
DAL 23.15567 22.43295 23.87838
CAR 23.20896 22.50117 23.91676
PHI 23.38832 22.61614 24.16051
DEN 23.55729 22.40624 24.70834
IND 23.88955 23.18984 24.58927
JAC 24.36472 23.35697 25.37247
TB 24.71455 23.44008 25.98902
OAK 24.73461 23.82956 25.63965
BUF 25.17504 24.24984 26.10025
These are only point estimates. Margins of error are relatively large but hey, this is a really hard problem and my $R^2 = 0.1586$ sucks but it's better than random! Predictive inputs into the model:
- Current pre-season average PAPG
- Previous 3 seasons average PAPG
If I had infinite time, I could include all sorts of awesome stuff like defensive coordinators, player transactions, offensive matchups, and so on. I don't have that kind of time, so it is at least enough to get us started on an approximate ranking of team defenses. When I look at the top teams drafted in NFL mock drafts, I get:
+-----------------------+-----------+----------+
| player_name | AVG(pick) | count(1) |
+-----------------------+-----------+----------+
| Seattle Defense | 94.4698 | 24470 |
| San Francisco Defense | 102.1096 | 24440 |
| Houston Defense | 109.7391 | 24412 |
| Chicago Defense | 112.3020 | 24398 |
| Cincinnati Defense | 118.7876 | 24354 |
| Denver Defense | 121.3694 | 24196 |
| New England Defense | 130.2573 | 23468 |
| Baltimore Defense | 135.8904 | 22451 |
+-----------------------+-----------+----------+
So it's.. close. Compared the model, Pittsburgh seems to be excluded from the mock draft picks (it wasn't picked in about half the mock drafts) but the model likes them. Same for Green Bay. We'll see how this plays out this year.