Last Week In Review
Week 9 was probably the GameZone prediction tool's best week yet. With no less than 13 games showing one team with 60%+ odds to cover, the tool's analysis
at default settings went 10-3 ATS for those games, which includes the Jacksonville-New Orleans game for which we issued a strong warning in our Week 9
Predictions Summary. Even at the lowest threshold (50%+ odds to win), analysis at default settings correctly predicted the straight up winner in 12 of last week's 14 games,
a better performance than Vegas spreads (10-4), all nine ESPN NFL experts, and four out of five Yahoo! Sports NFL experts. (Congrats to
Michael Silver for also going 12-2.)
In the 98 NFL games since its pre-Week 3 launch, the tool's analysis at default settings is now 66-31 (68.0%) picking straight up winners and 28-14 (66.7%)
against the Vegas spread in 60%+ odds to cover games. (See NFL predictions results to date.) It also has correctly picked the last 9 out of 10 NFL upsets as defined by the Vegas spread. Again, these
results were achieved with the tool having no knowledge whatsoever about injuries, weather, or bye weeks. In the spirit of transparency, we keep
all past games and predictions up for users to review, although please note that if you go back and view Week 3-6 games, the current default settings
no longer reflect the settings that were in place during those weeks.
In our opinion, last's week's biggest data-driven analysis triumphs included predictions that Pittsburgh
would blow out the Ravens and cover a large spread, that Minnesota would cover against a surging San Diego (the tool still had San Diego as slight favorites to win, however),
and that Houston would bounce back from its crushing defeat in San Diego to win and cover on the
road in Oakland.
We also reached another milestone after NFL Week 9. For all games since its pre-Week 3 launch, the prediction tool's expected margin of victory
calculations at default settings have been more accurate overall than the margins implied by Vegas spreads. Across 98 games, our win margin predictions
were off by an average of 10.73 points; closing Vegas spreads were off by an average of 10.79 points.
This exact difference of .06 points is statistically insignificant,
but the key point is that even counting every single NFL game during these weeks -- not just the select few games of our choosing (attention sleazy handicappers!) -- the tool is doing
about as good a job at predicting win margins as the oddsmakers and the overall betting markets at large. That makes us feel very good, especially given that the bookies and the public
both know about injury situations and the tool doesn't. In Week 9, win margin predictions were correct
to within seven points in 8 out of 14 games; closing Vegas spreads were less than seven points off only five times.
We are excited about the performance to date of the NFL GameZone predictions tool. We've already received several emails from users who
are achieving fantastic results by testing and customizing strategies using different factor importance weights, time filter options, and personal judgment regarding
injury impacts. That is the whole reason
why we designed the tool to be user-configurable.
However, after three straight strong weeks, now is a good time to remind all of our users about the importance of sample sizes.
Right now the tool's default analysis
is outperforming even our own internal expectations. We would not be surprised if a down week, or two, lurks in the near future. In fact,
we received a related email from a user this week that we wanted to share. He wrote:
"I have been watching the results to contemplate whether or not to use your selections in betting situations. (I am monitoring, because I am sure you are well aware that
approximately 99.99999% of the handicappers on the Internet are nothing more than rip off artists.) You have posted the correct results and actually have a very good ATS record.
Statistically, there have still not been enough selections to say that one can "bank" on your selections. However, what I have seen so far this season is impressive to say the least.
Your honesty and intellect are refreshing."
His point about sample sizes is well taken, as is the one about 99.99999% of handicappers. It's still too early to tell how the tool will perform for the entire NFL season, and the odds are
stacked very high against anyone who thinks he/she can achieve a 60%+ ATS record long term. And by long term, we mean making multiple predictions a week for multiple seasons. Most handicappers are a total sham because
they employ misleading tactics
to convince people to buy their picks based on meaningless short term or situational performance. (We're actually working on an article to expose a lot of these tricks to our users.)
Being math dorks, we're the first people
to admit that the GameZone prediction tool's ATS performance at default settings is still unproven from a long term standpoint. However, we do believe that the information and analysis it provides are far more
powerful, mathematically credible, and sophisticated than most if not all other NFL game analysis web sites or services available to everyday fans. And we're planning many more improvements.
On that note, enjoy Week 10, and please keep the comments and suggestions coming. We'd love to hear how you are using the GameZone predictions tool and
what you think will make it better.
Talk to us.
NFL Week 10 Preview
We would like to thank our users for all the comments and feedback last week. Keep it coming! For some reason, we're having trouble getting super excited about this
week's games; it must be WAPS (Week After Patriots-Colts) Syndrome kicking in. However, we've got at least a few games where the GameZone predictions tool
is indicating a strong favorite.
Sorry for the delay in posting the predictions commentary this week, but we've been busy little bees. The big news is that we just launched the alpha version of the TeamRankings.com FantasyZone,
our very first fantasy football analysis tool. Please check it out and email us with your thoughts!
NFL Week 10 Predictions Summary
Below is our Team Rankings GameZone predictions summary for all
NFL Week 10 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.
 Buffalo Bills at  Miami Dolphins
|
View Predictions |
|
Based on a point spread of Buffalo -3, analysis at default settings
shows Buffalo as the favorite to win the game, with win odds of 71.4% and an expected
margin of 6.1 points. Similar games analysis indicates Buffalo as the slight favorite
to cover the Vegas point spread with 58% cover odds.
Notes from the Nerds: Miami's predictive power rating suggests that there indeed may be a win lurking in its future, but
probably not this week. The spread is a relatively common one, and sensitivity analysis on power ratings
and recent win momentum (just play with the sliders) -- which couldn't be more different for these two teams -- doesn't meaningfully change the status of the Bills
as slight favorites to cover.
|
 Denver Broncos at  Kansas City Chiefs
|
View Predictions |
|
Based on a point spread of Kansas City -3.5, analysis at default settings
shows Kansas City as the favorite to win the game, with win odds of 84.5% and an expected
margin of 10 points. Similar games analysis indicates Kansas City as the favorite
to cover the Vegas point spread with 74.2% cover odds.
Notes from the Nerds: Is Larry Johnson worth 6.5 points more than his replacement? The Vegas line says yes. The Football Outisders
say no. Although there have been some instances
of the team like Denver winning in this situation, almost all of the prediction tool's angles favor Kansas City by around 10, assuming no injuries
or adverse weather. Denver will probably be without starting QB Jay Cutler, the Chiefs without LJ. Your judgment call.
|
 Cleveland Browns at  Pittsburgh Steelers
|
View Predictions |
|
Based on a point spread of Pittsburgh -10, analysis at default settings
shows Pittsburgh as the favorite to win the game, with win odds of 83.4% and an expected
margin of 12.6 points. Similar games analysis indicates Pittsburgh as the favorite
to cover the Vegas point spread with 68.7% cover odds.
Notes from the Nerds: Here we go again with the Steelers facing a tall spread at home, only this week they face a streaking
Browns team that power ratings indicate is a more formidable foe than the Ravens of one week past. The tool's similar games analysis presents
quite the polarized picture: some instances of the team like Cleveland barely squeezing out the upset, more instances of the team like
Pittsburgh winning in a blowout, and not much in between. We are excited to see how it plays out.
|
 Philadelphia Eagles at  Washington Redskins
|
View Predictions |
|
Based on a point spread of Washington -3, analysis at default settings
shows Washington as the favorite to win the game, with win odds of 65.6% and an expected
margin of 4.5 points. Similar games analysis indicates Washington as the favorite
to cover the Vegas point spread with 60.5% cover odds.
Notes from the Nerds: Neither of these teams has been playing well. Donovan McInconsistency and the Eagles got blown away by Dallas,
and the Redskins needed overtime to beat the Jets. Power ratings are close to equal for these teams, but even when you decrease the importance of those
factors, the tool's analysis still indicates Washington as the slight favorite to cover.
|
 Atlanta Falcons at  Carolina Panthers
|
View Predictions |
|
Based on a point spread of Carolina -4, analysis at default settings
shows Carolina as the favorite to win the game, with win odds of 82.3% and an expected
margin of 8.3 points. Similar games analysis indicates Carolina as the favorite
to cover the Vegas point spread with 66.8% cover odds.
Notes from the Nerds: Carolina's QB situation is serious enough that any prediction for this game needs to be taken with a
grain, or perhaps a dollop, of salt. At full strength, Carolina looks like a decently strong pick according to the tool's situational analysis,
but with even third-string QB Vinny Testaverde struggling with an injury, that's not exactly the kind of insurance policy we like. We'd probably steer clear of this one.
|
 Jacksonville Jaguars at  Tennessee Titans
|
View Predictions |
|
Based on a point spread of Tennessee -4, analysis at default settings
shows Tennessee as the favorite to win the game, with win odds of 81.8% and an expected
margin of 8.7 points. Similar games analysis indicates Tennessee as the favorite
to cover the Vegas point spread with 69.6% cover odds.
Notes from the Nerds: Not much to say here. These teams aren't very different in terms of quantitative power ratings. However, all of the tool's
situational angles give the Titans strong odds to win and cover, and they are healthy.
|
 St. Louis Rams at  New Orleans Saints
|
View Predictions |
|
Based on a point spread of New Orleans -11.5, analysis at default settings
shows New Orleans as the favorite to win the game, with win odds of 90.4% and an expected
margin of 13.6 points. Similar games analysis indicates that the Vegas point spread is
efficient for this game.
Notes from the Nerds: If the Rams keep up their current level of performance, we are going to need to redefine the scale of our power ratings.
Sweet Jemimah, they are bad! The return of RB Steven Jackson (if he can stay healthy enough to play the whole game) should help them, and the Saints are battling
a few injuries to key personnel. Nevertheless, analysis at default settings shows that the odds are stacked pretty highly against a Rams upset. However, the tool's
analysis gives New Orleans only 50/50 odds to cover the tall spread, despite the fact that Drew Brees might throw for 728 yards.
|
 Minnesota Vikings at  Green Bay Packers
|
View Predictions |
|
Based on a point spread of Green Bay -6, analysis at default settings
shows Green Bay as the favorite to win the game, with win odds of 83.5% and an expected
margin of 9.8 points. Similar games analysis indicates Green Bay as the favorite
to cover the Vegas point spread with 74.6% cover odds.
Notes from the Nerds: Hold on a second while we take off our Adrian Peterson colored glasses. OK. We think AP is a beast too, and the somewhat banged
up Packer secondary may well help his cause in this game. However, we math nerds are always thinking about probabilities and regression to the mean. If AP's performance
continued at the same level of his peformance against San Diego last week, he would be on a pace to rush for 4,736 yards per 16 game season. That figure is more than double Eric Dickerson's
current NFL record, so the odds of it happening are probably similar to a major league baseball player hitting 140 home runs in a season.
Peterson aside, the Vikings have QB health issues to deal with, with third-stringer Brooks Bollinger potentially
getting his first start since January 2006. Unless you think AP will deliver another off-the-charts performance, situational angles favor Green Bay as a moderate to
strong favorite to cover.
|
 Cincinnati Bengals at  Baltimore Ravens
|
View Predictions |
|
Based on a point spread of Baltimore -4, analysis at default settings
shows Cincinnati as the nail-biter favorite to win the game, with win odds of 56% and an expected
margin of 0.6 points. Similar games analysis indicates Cincinnati as the favorite
to cover the Vegas point spread with 62.8% cover odds.
Battle of the Underperformers #1 was sent to us by the power ratings gods to test our conviction in them. The Bengals (2-6) actually have better power ratings
than the Ravens (4-4), and in these situations, that fact leads to Cincy having a small odds to cover advantage at the tool's default settings. However, while the Ravens are a
lot worse team than their record shows, the Bengals just seem like an emotional train wreck right now. This is not an odds to cover prediction we feel extremely confident about, but the
numbers are what they are.
|
 Chicago Bears at  Oakland Raiders
|
View Predictions |
|
Based on a point spread of Chicago -3.5, analysis at default settings
shows Chicago as the favorite to win the game, with win odds of 60.2% and an expected
margin of 3.9 points. Similar games analysis indicates that the Vegas point spread is efficient for
this game.
Notes from the Nerds: Battle of the Underperformers #2 offers us very little to get excited about. The tool's angles don't indicate any significant ATS advantage
for either team. However, we did just notice that as of Friday midday, 82% of Yahoo! Pro Football Pick'em users had chosen the Bears to win this game. That's a big enough discrepancy with the tool's
calculated win odds to make Oakland to win a decent strategic upset pick for a straight-up pick'em pool.
|
 Dallas Cowboys at  New York Giants
|
View Predictions |
|
Based on a point spread of Dallas -1.5, analysis at default settings
shows Dallas as the favorite to win the game, with win odds of 77.3% and an expected
margin of 7.5 points. Similar games analysis indicates Dallas as the favorite
to cover the Vegas point spread with 69.1% cover odds.
Notes from the Nerds: Ladies and gentlemen, welcome to fascinating game of the week. A key factor affecting this game is the fact that the Giants
have played a lighter schedule lately than the Cowboys (see our NFL strength of schedule power ratings), so their
power ratings are significantly lower despite an impressive record. In a sentence, the performance level of New York, season to date, is less proven than that of Dallas.
If you use the sliders to decrease the importance of power ratings for this game, you will see the dramatic
impact on calculated predictions. In short, the tool is telling us that power rankings matter in this situation. And the Giants haven't had as many
chances as the Cowboys have had recently to perform well against strong opponents and drive up their power ratings. So if you believe that New York
would have performed similarily to the Cowboys given a like schedule, then this game is a toss-up. If not, Dallas is the strong favorite. Can't wait! |
 Detroit Lions at  Arizona Cardinals
|
View Predictions |
|
Based on a point spread of Arizona -1, analysis at default settings
shows Detroit as the favorite to win the game, with win odds of 69.8% and an expected
margin of 5.6 points. Similar games analysis indicates Detroit as the slight favorite
to cover the Vegas point spread with 58.2% cover odds.
Notes from the Nerds: Just so you know, there is not a bug in the predictions here, in terms of the Lions (with a +1 spread) being shown as having
greater odds to win than to cover. In short, the prediction tool's algorithms look at winning and covering as both dependent and independent events. When a
result like this occurs, it means that at least some identified similar games feature spreads that are close to the current game spread, but with the underdog and
favorite reversed. (For instance, similar game #20 at default settings features the team
comparable to the Lions, the New York Jets, facing a -1.5 spread instead of +1.)
Mathematically, this situation occasionally can cause illogical-looking final predictions. In the end,
just be aware that this situation can occur, and that the exact numbers presented in final predictions are derived from a wide range of underlying data. The tool does not simply try to determine who is going to win and by how much,
and then compare that result to the spread to determine cover odds. The algorithms are a lot more complicated than that.
OK, enough of the pocket protector talk. The tool's angles indicate that the line looks pretty efficient here, and that this game should be closer than
most people seem to think. Yes, the Lions are hot and the Cardinals are cold. But play with the "Win % - All Games" slider while set to "Last 3," and you'll see that
recent win momentum has not played a signficant role in determining outcomes in similar game situations. Based on the data analysis alone, this game would not be
one of our strong picks.
|
 Indianapolis Colts at  San Diego Chargers
|
View Predictions |
|
Based on a point spread of Indianapolis -3.5, analysis at default settings
shows Indianapolis as the favorite to win the game, with win odds of 86.3% and an expected
margin of 9 points. Similar games analysis indicates Indianapolis as the favorite
to cover the Vegas point spread with 74.5% cover odds.
Notes from the Nerds: By only losing to the Patriots by four, the Colts' overall power rating actually increased. While Peyton and company
again may face some health situations with key personnel, injuries didn't stop them from performing up to expectations vs. New England. In the
end, unless you think the close loss to the Patriots will trigger an emotional meltdown, the data indicates that a -3.5 spread is generous to Indy.
|
 San Francisco 49ers at  Seattle Seahawks
|
View Predictions |
|
Based on a point spread of Seattle -10, analysis at default settings
shows Seattle as the favorite to win the game, with win odds of 90.2% and an expected
margin of 14.1 points. Similar games analysis indicates Seattle as the favorite
to cover the Vegas point spread with 61.6% cover odds.
Notes from the Nerds: The best part about this game is that someone has to win, which means a brutal overall losing streak for the NFC West must end. While
it may be hard to believe that the Seahawks could have a slight edge to cover a 10 point spread against any team (the cover odds are a little sensitive to power ratings), the tool's analysis
indicates two things: (a) yes, they do, and (b) yes, the Niners are really that bad. Injury situations seem relatively similar for both teams.
|
|
