2007 NFL Week 11 Predictions Summary

NFL Predictions listed on this page are based on the default settings of the Team Rankings GameZone NFL predictions tool, a 100% objective and data-driven system developed by Stanford engineers. Default settings loosely reflect our research into which statistics tend to be good predictors in most cases.

We encourage our users to manipulate the predictions tool for each matchup of interest, and to customize individual factor settings to test alternative strategies for predicting NFL outcomes. Although we will provide performance tracking information at the default settings, keep in mind the following guidelines:

• The system does not incorporate or attempt to model the potential effects of injuries, weather conditions, coaching strategies, specific player matchups, emotional factors such as team morale, or anything else not listed in the statistical factors box. Making any corresponding adjustments to predictions will require your own judgment.
• Every data-driven predictive system has a margin of error associated with it, and this system is no different. We believe that a 3% to 5% safety buffer around odds to win/cover is warranted, as is a 1.5 to 3 point buffer around expected margin of victory. For example, if odds to win for a given team are calculated at 53.5%, margin of error means that the actual win odds calculation may be as low as 48.5% or as high as 58.5%, based on the variance inherent in our data set and models.
• Every matchup is different, and tweaking the factor settings for a matchup based on the specific characteristics of the two teams in question may improve prediction accuracy. Simply adjust the importance sliders and time filters for any factor, and prediction information will update.
• Try to give the "Help" section a good read. It's lengthy, but it will enable you to understand how this product works and how to use it best. We're confident that if you read the help section, you'll be convinced of the analytical power and sound mathematical foundation of this tool.
• Please remember that luck and probability mean that anyone can win or lose big over one week, month, or even season. The key is to adopt a consistent strategy that you can measure, refine, and improve over the long term, and to filter emotion out of the process of making picks. This product can provide the basis to do so.
• Have fun, and please let us know your comments, questions, suggestions, and feedback!

Last Week In Review

So that's what a correction feels like. Ouch. As we mentioned in the Week 10 notes, we felt a down week may be on the horizon given the inescapable realities of probabilty, and it didn't take long to arrive. At default settings, the prediction tool's analysis was 4-7 (36.4%) ATS in 60%+ odds to cover games, and expected win margins were off by an average of 14.6 points, compared to an average difference of 12.6 points for Vegas spreads. Thank you Seattle for at least putting a cherry on top of bad ice cream.

The relative bright spot in last week's results was that despite a pretty dismal ATS performance, the tool still went 7-7 picking straight up winners in a pretty crazy week, a mark that was as good or better than the picks of 13 out of the 14 "experts" that we track and better than our consensus picks benchmark as well. So most Pick'em players using the tool's analysis should have at least held their ground. Also, the tool is still 32-21 (60.4%) in 60%+ odds to cover games since its pre-Week 3 launch.

You'll never hear any excuses from us for a bad week. First, let's calculate the probability of last week happening. Of the 11 games with odds to cover of 60%+, the average odds to cover prediction was 68.3%. The tool's analysis at default settings got four games right and seven wrong. We can calculate the probability of getting exactly four out of these eleven games right, as well as at most four out of the eleven right. The former is referred to as an "eleven choose four" with an individual success probability of 68.3%.

We'll spare you the gory details, but if the cover odds are indeed "correct," the odds of getting exacly four out of these 11 predictions right are about 2.3%, and the odds of getting at most four games right are around 2.9%. If, in reality, the average odds to cover for these games was more like 60% and not 68% (which is probably a fair assumption given the tool's performance to date, see below), then the odds of getting four or less right soar to almost 1 in 10. And ironically enough, last week was Week 10.

Want to know why a lot of gamblers who think they know what they are doing lose money? Even if you can consistently pick at 55% ATS, which is very hard to do long term, your odds of losing five bets in a row are about 1.8%. Your odds of losing three bets in a row are about 9%. Your odds of losing at least 8 out of 11 bets are more than 6%. Those are extremely dangerous numbers to people who don't practice sound bankroll management and prepare for down streaks.

It does not insult our manhood or fanhood or football ego when down weeks happen, and we're not going to flame the refs or the coaches or posit some wild conspiracy theories that explain why things went wrong. In fact, we just flat-out assume that bias and insider information is woven throughout the NFL at all levels, just like any other pro sport. Their existence is an unavoidable reality, as is being wrong a fair amount of the time when you are trying to predict the future.

The real question is, can you can develop a systematic approach to predicting outcomes that can give you an edge in the long term. (See Season to Date results.) For example, Dr. Bob Stoll, a handicapper we very much respect and who had an article about him published in the Wall Street Journal, has won 55% of his NFL Best Bets over a 20-year career. We know a lot of people who would see that figure and say, "What? I could EASILY pick at least 55% ATS." Guess what? It's a lot harder than you think. You make money in Vegas at a 55% win rate, and if Vegas didn't make money on sports gambling, then there would be no Vegas sports books. Yet miraculously, they still exist!

With that said, we can definitely learn from last week's underperformance. We do believe that a few games, such as the Indy-San Diego mess, were statistical anomalies. For example, what is the probability that Peyton Manning throws 6 INTs, the Indy defense gives up two kick runback TDs, and Adam Vinatieri goes 0-for-2 on field goals, all in the same game? It's clearly greater than zero, but what odds would you need to see to bet on that proposition? And shoot, the Colts still had a chance to win and cover on the final drive.

Similarly, Pittsburgh outgains Cleveland 401-163 and barely pulls out a 3 point win late in the fourth quarter. Wow. Without a doubt, unlikely outcomes like this are always going to occur over the course of a season featuring 200+ games. However, being smart and playing for the long term means that you play the probabilities. We don't have the exact data on hand, but we would wager that more often than not, with a yardage discrepancy that big, Pittsburgh would at least push a -10 spread, and the tool's prediction seems well grounded in that case. But this week, it was wrong.

There are two things we want to investigate more, though. First, weather, bye weeks, and injuries may have played a more significant role in Week 10 than in other weeks. The ball was clearly slipping through receivers' hands in San Diego, and all three underperfoming teams coming off a bye week (Miami, St. Louis, and Chicago) outperformed expectations. The impact of injuries will always be difficult to quantify (especially since it's impossbile to know who may actually play until game time), but the breakdown of the banged-up, usually rock solid Tennessee defense makes us wonder if we can find a systematic way to at least make minor adjustments. This is an open issue.

Second, the tool is finding more high-odds situations that we expected at the beginning of the season, and so far very high odds-to-cover predictions are not correlating tightly with actual results. (The sample size is still small, though.) In other words, the overall performance of 60%+ odds to cover predictions season to date has been solid, but 65%+ odds to cover predictions have been no better than 60%+ predictions. As a result, we feel that the best strategy for using the tool at this point is to apply "soft" analysis (injury adjustments, etc.) to all of the 60%+ games, and then pick a subset of them to focus on regardless of the exact odds to cover prediction.

As always, send us your comments and suggestions and don't forget to check out our new fantasy football player predictions.

NFL Week 11 Preview

The GameZone prediction tool's analysis is indicating a lot of close contests this week, where cover odds and even win odds are in the neighborhood of 50% for both teams. Please remember that any odds prediction in the range of 45%-55% is probably not meaningful on account of margins of error inherent in the calculations.

Unfortunately, we will not be able to deliver full Notes from the Nerds this week, so please make sure to test a few different settings for each game, and see if meaningful changes in predictions result. More importantly, remember to assess the injury and potential weather impacts on every game, especially those where odds to cover are very high and win margin calculations deviate greatly from the Vegas spread.

NFL Week 11 Predictions Summary

Below is our Team Rankings GameZone predictions summary for all NFL Week 11 games. Although we update this page periodically, predictions can change multiple times per day based on new information such as shifting NFL betting lines. Click on "View Predictions" to see the most recent lines and predictions and to customize factor settings.