Last edited Sat Jan 10, 2015, 07:59 PM - Edit history (1)
Not really picks but thoughts.
I hope Baltimore wins, I think Carolina can keep it close--their defense has been the second best unit since Seattle lost to KC. Also that other running back (Stewart, if I'm remembering the right one) is healthy and they already have 2 really good backs.
I'm rooting hard for Dallas, I think Green Bay is more likely to win but Dallas is certainly capable, a lot of people will disagree with me (I think) but Romo gives them a chance of winning games they had no business competing in but I wouldn't sell them short, they are one of the top teams in turning possessions into points but the team that is tops in category is Green Bay.
Denver, total points I don't know. Denver has a good D but Colts passing is hard to slow down, like with Dallas, very efficient in turning drives into touchdown. I'll go 75.
Brian Burke has a probability model which weighs all the key stats giving most of the weight to net/ypa, there is home field edge factored into it. I don't know every tool out there but there is nothing that I ever came close in any other sport (well, the Trading Bases book had a pretty good model for betting on baseball. He uses sabermatrics for the foundation but it is very detail orientated, focusing on things that a normal sabermetic wouldn't, he uses the baseball pythagorean theorem to estimate seasonal projections but it also gives advice on investing into baseball betting like a "hedge fund". His models updated for lineup changes would come up with an estimate of the true odds and the biggest differences from the Vegas spread he invested "2% of the fund" which was his max, even something like it gave a team 40% chance of winning, he invested a portion of the fund but I didn't mean to get sidetracked as there is more I can say on the book, but it wasn't very successful in picking winners (52%) but it was very successful in turning a profit (former Wall-Street guy wrote the book)
Burke's model is very effective at picking winners but since it isn't subjective but based on data on full-season up-to-the-day team stats, like it might give a team an edge because of the net/yards per pass attempt but if the backup QB is playing, it doesn't really apply (or a backup QB dragging down the YPA numbers for several games but a returning healthy starter QB would have lower odds based on the model then actual odds). Even with that, he gives great overviews on what to expect, team/strength weaknesses, and he'll mention the limitations of the model based on the factors of the game.
PANTHERS: 10 (25% win probability)
SEAHAWKS: 24 (75%)
VEGAS: Seahawks (-11)
This looks like one of the more lopsided playoff games in memory. It's true that Carolina's losing record in the regular season doesn't reflect its improvement in the latter half of the year, but Seattle's defense is still significantly better. Seattle is slightly better against the run and considerably better against the pass.
Although the Seahawks offense is league-average in passing efficiency, it is still effective thanks to a very low turnover rate and a devastating running attack. The Panthers can run well and have about average turnover rates, but they haven't been able to pass efficiently all season.
The only critical category where Carolina has an edge is in penalty rate. Seattle has the third-worst penalty rate in the league, but that may be, at least partially, by design. Seahawks defenders are known for their physical aggressiveness against receivers, and that can pay off despite the risk of a few flags.
One of the things I like is sometimes stats such as penalty rate which no one pays attention to has an impact on probability. A low vs a high penalty rate changed a big underdog to a 40%-45% underdog. In addition to the Seattle stat (which he called the rampart interference a smart strategy based on Bayesian economics http://archive.advancedfootballanalytics.com/2014/01/thomas-bayes-would-approve-of-seattles.html).
COLTS: 17 (28%)
BRONCOS: 27 (72%)
VEGAS: Broncos (-7)
Both offenses are very effective, but the difference in this game is all about defense. Indianapolis' defense is much better than it is given credit, allowing 6.1 net yards per attempt, which is slightly better than average. But Denver's defense is probably the best in the league, allowing only 5.3 net yards per attempt. The Broncos are also more effective against the run than the Colts, and have been able to generate takeaways more frequently.
Because of their defensive numbers, I think the Broncos may be even better than their 12-4 regular season record suggests. If you ever wondered how good a Peyton Manning-led team could be with a solid defense, this is it.
I think a lot of people are selling Denver short this year when they had just about a "worst to first" improvement on defense. If they have to fly out to Foxboro that is going to be a killer. MST & Pacific teams have horrible winning % in the EST and Denver does too who didn't win a football game in the state of Florida until the mid 2000s.
COWBOYS: 24 (31%)
PACKERS: 31 (69%)
VEGAS: Packers (-6)
Statistically it would be hard to tell these two offenses apart, except for one key facet of play. Each team's passing efficiency, run success rate, and interception rate are as follows:
Dallas: 7.4, 43%, and 2.2%
Green Bay: 7.5, 41%, and 1.1%.
Both interception rates are better than average, but it is Green Bay with the much lower rate at 1.1 -- the best in the league.
On defense, the Packers have a big advantage against the pass, while the Cowboys have a big advantage against the run. But passing is what tends to win games in the NFL, so on net the Packers have a sizeable edge when on defense.
http://www.sportsonearth.com/article/105976062/nfl-divisional-round-man-machine-patriots-colts-panthers-ravens-broncos-seahawks
Both teams have among the league's best offenses but seems the team that is the best, also has the better pass D going into the game. I actually wasn't expecting Dallas odds to be so low searching for this (i was expecting near 40)
This one surprised me initially but immediately remembered they have Kubiak who is one of the best offensive minds out there that consistently shows results
The Machine
RAVENS: 20 (43%)
PATRIOTS: 24 (57%)
VEGAS: Patriots (-7)
I bet you didn't know Baltimore has a slightly better net offensive pass efficiency than New England. It's little misleading because the Ravens had a favorable schedule, matched up against a weak NFC South this season. The "Machine" accounts for opponent strength, but it doesn't make a very large difference by this point in the season, because after so many games things tend to average out.
Most other ranking or prediction systems would have the Patriots as the undisputed top team in the league. That's because they have the highest point differential, averaging 9.7 points per game more than their opponents. But this model doesn't rely on points or wins, which themselves can be misleading. Garbage-time blowouts tend to skew score-based systems. Analysis at a per-play level is likely to give us a truer picture of team strength.
Both teams appear about equal in terms of passing efficiency on both sides of the ball. The differences between the two teams are in run success rate. New England is more consistent running the ball, but Baltimore is more consistent stopping the run. In total, it appears that each team's strengths and weaknesses are largely a wash, and what is left is home-field advantage for the Patriots.
This analysis is starkly different than consensus estimates that New England is a decidedly better team. The reason for the disagreement is that the Patriots began the season struggling to move the ball, and the "Machine" weights performance equally across the season rather than over-weighting recent performance. There are advantages and disadvantages to this approach, but the bottom line is that if you believe New England's early difficulties reveal vulnerability then this game could be a much more even matchup than many think. But if you believe New England's early-season difficulties are not representative of their current level of play (as I personally do) then their advantage should be bigger than the Machine's numbers suggest.
So many things I want to highlight but the model is most effective when it differs from consensus. I won a week in the DU pool, I used a strategy in picking Atlanta @ Carolina because I had a feeling most people were going to pick Atlanta & the were favored but according to the model, Carolina was a heavy favorite going into the game. they had the edge in efficiency numbers + home field is why they had the edge in that game.
http://www.sportsonearth.com/article/105976062/nfl-divisional-round-man-machine-patriots-colts-panthers-ravens-broncos-seahawks
If I were to pick up football betting his weekly game probabilities would be my bible so my picks based on the model.
I'm going a little away from the model and go with recent subjective criteria. Carolina has won 5 straight with the second best performing defense in the league in that time span (Seattle #1). The running game has come back from the dead. The other side of the Baltimore played the NFC South is Carolina played the AFC North. Under.
Carolina
Green Bay (though I'd like to see Dallas win, it seems likely a push or more)
Denver (65 points)
Baltimore
On edit - I changed Denver points from 75 to a more reasonable 65.
= new reply since forum marked as read
