Monday, September 16, 2013

Is Safety Rating A Real Thing? An NFL Survivor Primer


Way back in 2008, for whatever reason, I decided to sign up for ESPN.com's Eliminator Challenge. After thinking about my Week 1 pick for probably five minutes, I decided the strategy behind the contest was somewhat interesting, and decided to write a blog post about Week 2. Those two picks both won, as did the next 15.

This unlikely streak had two effects. Most immediately, it thoroughly annoyed me that I hadn't profited from going 17-0, as I was only in the free ESPN contest, and lost the tiebreaker.

The more lasting effect was that my shiny 2008 record caused people to think I knew what I was doing when it came to NFL Survivor contests. Which certainly wasn't the case at the time, and remains overstated even now. But here we are, five years later, and I'm still writing these posts. Considering that most of the people reading at this point surely weren't doing so in '08, and even then I never spelled out (had?) an overall strategy, now seems like as good a time as any to do just that.

The general premise of Survivor pools is quite simple. You pick one team each week, and all you need them to do is win outright (there are no point spreads involved). If they win, you advance to the next week, and attempt to pick another straight up winner. If they lose (or tie), you're eliminated from the pool. The one catch is that you can (generally) only use each team once, so you have to pick your spots with teams that are often big favorites.

In my view, there are three important factors to consider when looking at each game. We'll start with the most obvious.

Win Probability

Well, yeah. The goal is to pick a team that wins, so this seems like an important input. Each week, I use the money line at Pinnacle to determine each team's likelihood of winning. And to answer an inevitable question: if you think those odds are way off, you should probably open up some offshore accounts and do something about that, rather than wasting your time with a $50 Survivor pool.

Recently (inspired by this and probably some other nonsense), I've started including a "Safety Rating" (™) for each team. Yes, SR is a "real thing", I guess. But it's not some top secret formula; it's just the implied winning percentage from Pinny's money line, translated onto a scale from 0 to 10.

Keep in mind that these lines sometimes move quite a bit at Pinnacle during the week; this is especially relevant since the weekly Survivor post usually goes up on Wednesday. I try to mention any relevant post-Wednesday developments in the comments and/or on Twitter.

Future Value

A key aspect of  these pools is that you can (generally) only use each team once. So, especially early in the season, even if a team is a big favorite it's important to consider whether this is the optimal use of your one opportunity to take them. The Texans are favored by nine against the Titans now, but what about when they face the Raiders and Jags later in the season?

I maintain ratings for each team, which I use for purposes of evaluating future value. But, absent that, Survivor Grid is an excellent resource, as it lays out each team's remaining schedule, and also provides an estimate of the line on each future game.

The size of your pool is also very important to consider when looking at future value; if there are only four people left in Week 4, it probably doesn't matter that the Seahawks host the Cardinals in Week 16.

Popularity

Here's the one that I think your average Survivor contestant doesn't take into account at all. Consider this made up, completely unrealistic scenario.

You're in a Survivor pool with 99 other people. For some reason, there's only one game on the board this week; the Broncos are hosting the Jaguars. This seems like the Lock of the Century, and Pinnacle has Denver at 92% to win. Everybody else has already submitted their picks, and they're all taking the Broncos. Which team do you take?

Ignoring future value for a moment, it's just a matter of math to compare the two options with respect to win probability and popularity. If you take Denver, you've got the same chance of winning as anyone else: 1 in 100. If you pick Jacksonville, you're eliminated 92% of the time, and look very dumb in the process, as every other entrant advances via the Lock of the Century.

However, the Jags do win 8% of the time, in the Upset of the Century. And in that scenario, well, your pool is over. You win. So taking Jacksonville gives you an 8 in 100 chance of winning, which you'll notice is quite a bit higher than if you pick the Broncos.

Clearly, this exact scenario will never come up, but the general principle holds: it can be extremely advantageous to go against the crowd in pools like this. And while that doesn't necessarily mean taking a 12:1 underdog, it often means passing on the most obvious favorite of the week. I never really write out the math behind all this in the actual posts, but there is certainly math to be done.

The two main sources I use to gauge popularity are the Office Football Pool site, and the pick distribution for Yahoo!'s "Survival Football" contest. I used to use the consensus numbers from ESPN's Eliminator Challenge, but they allow you to make picks weeks ahead of time, which I think ends up skewing the numbers. These resources are a nice general guide, but it's important to keep in mind that even minor differences in the rules of your pool can have a noticeable effect on who picks whom.

A random, but certainly relevant note: if your deadline is on Sunday, Thursday night games can cause some weirdness here. In that situation, the popularity of a team playing on Thursday night should be discounted, as it will always decrease between 8:25pm EST Thursday and 1:00pm EST Sunday.


So, those are the three most important things to consider. And sometimes they can be analyzed jointly. Early in the season, it's often smart to pass on popular teams with a lot of future value, in order to use them later, when they're again big favorites but fewer people still have them available.

Things can also get quite a bit more complex later in the season, when the relevant details -- exact pool rules, how many people are left, which teams you've used, which teams others have used -- vary wildly from pool to pool. At that point, the "official" blog pick becomes significantly less important, with everybody needing to consider their own personal scenario. And that's how you end up with threads like this and this.

Please post any questions/additions about things I've forgotten in the comments, and I'll add relevant stuff to this post.

1 comment:

  1. Strategy becomes ultimately very important when you are in very large pools, say 3000-10000 entries where you gave to pick two teams a week at some point.

    This is when it is always important from week 1 to zig when your pool members zag which nog only allows you to avoid the big upset but alao allows you to zag later in the pool when your fellow members have to zig.

    ReplyDelete