Friday, October 07, 2005

NFL Pool

It says somewhere above that I talk about sports once in a while... so here we go. I run a pool through CBS Sportsline for about 35 people. It's simple: $5/week, $60 for the season (which is only 14 weeks - Weeks 1, 6, and 7 are their own mini-pools). The point system is what I like to call "WITS-style", because that's what we did in WITS [until told in no uncertain terms that pools were not allowed... okay, maybe a little after, too]: Basically, you pick the winners of each game, and assign a different number up to 16 (14 when there are byes) to each game. Whoever gets the most points wins, so you usually want to stick your higher numbers by the teams you are most confident in, and the lower ones in the others.

Many people go by their own instincts and research every week, others use systems, others research, etc. Still others do it almost randomly - and yet, in the end, it never seems to matter all that much. Someone got lucky with a couple of upsets and survived while everyone else fell out of the top spot. I used to use my own brain, and while in WITS, I finished off one season winning 4 of 5 weeks and the overall (which is money set aside each week to those who accumulate the most points over the season). Last year, I re-opened the pool after a 3-year hiatus, and there were about 25-30 people in it each week. I didn't do as well overall, starting with a few atrocious weeks - but I did win one week, and tie another (splitting the first and second place money), which is decent for a 12-week season.

This year I decided to follow a system I implemented in the middle of last season. Back in WITS, a friend of mine (now a math major in Lander) decided to use the average scores of teams to predict how they would fare - this offense vs. the other's defense, and vice versa - and then assigned the 1-16 values based on the closest to furthest margin of victory. The weeks he was not in the pool, he would have finished first twice and second once. While this system would not work in this year's pool (the more people, the more pure average systems are unable to win), it got me thinking. What if I could find a better measuring stick? Last year, I believe I found it - the Aikman [Efficiency] Ratings.

If you want to know about them, read the articles - but suffice it to say, they were a far better measure of predicting which teams will win ballgames: And which ones have a very good chance of being upset. Ironically, I could not find them at the beginning of this season, and for Week 1 (its own pool) I used the numbers from last year. They were a disaster, and I finished in the middle of the pack. In Week 2, the first week that counted for the overall, I made my own picks. Ironically again, I won the week by 7 points (a solid margin), with a 15-point cushion over 5th place. I was 10-6 with 93 points - not a great week, but for a Week 2, quite solid. Week 3 I found the Aikman Ratings, and used them - and at 7:00 PM Sunday, I looked to be on my way to another win. I was 9-2 with just 9 points lost - and Pittsburgh blew it, the Giants got crushed, and the Chiefs got run over, dropping me toward the bottom of the pack for the week (but still tied for second overall) with 66 points.

For Week 4 I tweaked the Aikman Ratings a bit, adjusting them a certain way which is important but Aikman's Ratings cannot measure. The tweak probably got me an extra 4 points, and I had a very solid week - going 10-4 with 78 points, with only Oakland's upset of Dallas keeping me from coming in third place. The winner had 91 points, and now holds the overall lead - I'm in third, just 11 points behind. The greatest positive is that the Aikman Ratings improve as the season progresses - the statistics balance out to a more accurate measure of a team's abilities. Therefore, I am in excellent position, remaining this high this early in the season.

Now, for this week's picks:
3 Bal @ Det
12 TB @ NyJ
7 Mia @ Buf
5 Ten @ Hou
11 Chi @ CLE
14 NE @ Atl
2 NO @ GB
1 Sea @ StL
13 Ind @ SF
8 Was @ Den
6 Car@ Arz
10 Phi @ Dal
9 Cin @ Jax
4 Pit @ SD

The obvious concerns: 14 against the Patriots, 11 against the Browns. The fairly safe bets: 13 on Indy - and that's it. This week is a little disconcerting, with so many road teams the favorites, despite my adjustments.... But we'll have to wait until Sunday. Good Shabbos!

Technorati tags: , , , , , .


  1. I am in a pool at the office. Been in second for three weeks now. Dolphins and Raiders both screwed me up.

    FWIW, I go by my gut on my picks.

  2. In a way, the systems are easier on the mind... If you win, it's "Ha - my system worked!" If you lose, you say, "Ha - stupid system... I would never have done that!"

  3. Yo Ez,

    I've been playing (and winning) fantasy football for 11 years now, and I just found out about Don't know if you're as into individual player performances as predicting team outcomes, but I love this "player's stock" idea on

    I've heard that hard-core options, futures, and derivatives traders are using this thing, so I can't wait to pit my subject-matter expertise against their valuation abilities!

    Check it out!

  4. I'm not as into the individual performance, if only because I don't do fantasy football this year. I used to win every year, but I don't find it as fun for some reason... maybe I was just never in a great league.

    I heard about protrade - that does sound cool. Can you start it in the middle of a season?