Current Totals:
Click here for direct access to Google Spreadsheet.
September 4. 2014 (Post-Draft Edition)
Okay, well, that was nuts. First thing’s first, please double-check (the normal kind, not the discount kind) the spreadsheet to make sure I entered your teams correctly.
Second thing’s second. I have no idea what the right strategy was there, but let’s take a look at how folks did. Below I’ve listed everyone’s collection of teams, along with two win predictions: first, the Vegas O/U number and then the ESPN/538 “Elo” wins prediction (which is explained here).
Ziff (34, 35): PIT(8.5, 9.1); MIA(7.5, 7.7); BAL(8.5, 8.9); CIN(9.5, 9.3)
Scotty (32.5, 33.7): CAR(7.5, 9.4); TB(7, 6.5); NO(10.5, 9.5); ARI(7.5, 8.3)
Allison (33, 32.8): IND(9.5, 9.3); SD(8, 8.8); DET(8.5, 7.5); NYJ(7, 7.2)
Mushtaq (34, 29.7): HOU(7.5, 6.2); CHI(8.5, 7.7); GB(10.5, 7.8); DAL(7.5, 8)
Rahul (35, 33.8): DEN(11.5, 10); PHI(9, 8.6); NYG(8, 8); BUF(6.5, 7.2)
Nigel (33, 33.8): MIN(6.5, 7.5); NE(11, 10.6); SF(10.5, 10.3); JAX(5, 5.4)
Yoder (33.5, 31.1): SEA(11, 10.9); WAS(7.5, 5.9); STL(6.5, 7.5); ATL(8.5, 6.8)
Willie (26.5, 25.8): KC(8, 7.3); CLE(6.5, 5.9); OAK(5, 4.7); TEN(7, 7.9)
So it looks like Nate Silver has me as the favorite, with Nigel, Rahul, and Scotty right behind. While Vegas likes Rahul to win, with me, Mushtaq, and Yoder on his tail.
Of course, as Brown graduate Chris Berman would say: THAT’S WHY THEY PLAY THE GAMES.
Some additional thoughts on the draft and strategy: So if the average team gets 8 wins, and we each have four teams, then the average player should have 32 wins. Spreading those 32 wins over our $100 budget, that means each “win” should be about $3.13 per win. So the “average” team with 8 wins would cost $25.
But I didn’t think that could be right, since the market for teams has to clear, so the worst that could happen is that you get stuck with the worst team in the NFL for $1, which would maybe net you 4 wins. That means you’d have $99 left to fill out your roster with three teams. At $3.13 a win, those remaining three teams at $33 “should” be able to get you about 10.5 wins each. But there’s no way that’s right, at least certainly not for everyone, since there are not that many 10.5-win teams in the league.
So at that point I sort of gave up and figured I’d see what happened at the auction. And then “bad” teams started going for $7 or so, and I had no idea what to do with that. And everyone seemed set on bidding way over any of my estimated win values, whether under the “average” model or the model with $1 for the worst team. At the end of the day, I’m not sure what a win “should” have been valued at. (And in fact I’m pretty sure, particularly in light of the actual action, that the price-per-win function is not linear.) So what did y’all do? (Email or post in the comments?)
And that’s all I’ve got for now. I’ll update the spreadsheet after each game, and do an updated post, perhaps with new Vegas and 538/Silver favorites after each week. Woo hoo!
–DZ
Addenda, after my post-draft conference call with Nigel, who very patiently explained to me what WebX is and how he created a macro-filled draft analysis spreadsheet:
September 1, 2014:
Hello all! So here it is, the official 2014 NFL Team Wins Pool Webpage. Thank you to our sponsor, Ziff Blog, for hosting this webpage.
Here’s what I’m thinking: Every week (or perhaps more often) I’ll post here with an update on the league, and y’all can chat in the comments or just email each other, or whatever if you’d like. I’ve shared the Google Spreadsheet with all of you, which I’ll be using to track the wins/totals throughout the season. It will always be embedded at the top of this page, but you can also view it directly at the link.
Reminder: The draft is at 7:00pmPT/10:00pmET on Thursday, September 4th. I’m hoping that’s just about halftime of the season-opening Seahawks game. Please please please send me your $15 before the draft. Use Paypal: david.ziff@gmail.com.
And that’s about it. Nigel, I’m sorry this interface is not fancier. My coding/web skills are limited.
Oh, and if you haven’t signed up yet under ESPN for the draft, please sign up and join the league! http://games.espn.go.com/ffl/leagueoffice?leagueId=1635495&seasonId=2014
Ziff Blog 
Interface looks pretty good to me. Simple and to the point. Nice work Davey, I’ll put in a good word for you in the Computer lab
Of course, obviously when I use the “win prediction” metrics that I used pre-draft to evaluate the results post-draft, it’s no surprise that my team looks good. Looking at some other predictions with Nige, I seriously undervalued every one of Mushtaq’s teams. So he could be a juggernaut.
Also, if we look at volatility, I think Yoder has the most volatile team. Seattle will be good. Those other three teams could be good or bad. If they hit, they have a chance to do well. Big risk-reward strategy I think.
The real key is how bunched up everybody is. Whoever drafted this year’s KC is going to be the winner.
As I think about, that has to play a part in any sort of pre-draft preset auction “value” like the kind they have on ESPN. That’s why the prices for QBs drop precipitously after the first 12-13-14-or-so in the recommended values, with just $1 guys after that.
That observation would lead to a non-linearity of the price-per-win rate. (Mush, I was talking to Nige about this, and it’s just like in Budget. Yes, the point-per-dollar ration is not as good for the boppers, but you need the best boppers because you can’t just have a team of “value” 1000 guys and still win. Those last few points, or in this case wins, really are worth more.)
And that played out in our draft, since the best teams really did go for a lot more than any linear PPW rate would have predicted.
But how do you model that?! Not surprisingly, Nate_Wilson has started commenting on my post on the real Ziff Blog with some crazy math/econ modeling stuff. We’ll see what he gives us. Underscore!