2014 NFL Betting Picks - Week 9

Pick 1: San Diego +1.5 incorrect NFL point spread pick
Pick 2: Indianapolis -3 correct NFL point spread pick
Pick 3: Philadelphia -2 correct NFL point spread pick
Pick 4: Kansas City -9.5 correct NFL point spread pick
Two losing weeks in a row starts to get me a bit nervous. Regardless of the long-term strategy, we need to pick up the pace this week with 4 solid picks. I felt that all three picks last week were the right ones given the circumstances and the data. This week I'm also feeling comfortable with the games that were selected by the probability of covering the spread as well as the final pick. Below you will find the free nfl point spread picks for the non-premium picks and members should be receiving the premium picks shortly.

This week we again use 18% of bankroll split evenly across our 4 premium games giving us a bet of about $420 per game. Below are the computer-based NFL point spread predictions for week 9 of the NFL:

GameVegas LineEstimatePrediction-VegasConfidence
SAN DIEGO @ MIAMI -1 3.9 4.9 61.0%
INDIANAPOLIS @ NY GIANTS 3 8.4 5.4 60.8%
PHILADELPHIA @ HOUSTON 2.5 5.5 3.0 58.5%
NY JETS @ KANSAS CITY -10 -23.6 -13.6 57.5%
OAKLAND @ SEATTLE -15 -20.0 -5.0 54.3%
DENVER @ NEW ENGLAND 3 6.3 3.3 54.1%
WASHINGTON @ MINNESOTA 0 2.9 2.9 50.0%
TAMPA BAY @ CLEVELAND -6.5 -5.8 0.7 50.0%
BALTIMORE @ PITTSBURGH 0 2.3 2.3 50.0%
ARIZONA @ DALLAS -3.5 0.1 3.6 49.7%
NEW ORLEANS @ CAROLINA 3 -5.0 -8.0 49.1%
JACKSONVILLE @ CINCINNATI -11 -18.5 -7.5 48.4%

How to read the table:
  • Vegas Line: A NEGATIVE number implies the point spread favors the HOME team
  • Estimate: NFL Pickles' point spread prediction
  • Pred-Vegas: Subtraction: POSITIVE implies VISITING team will cover point spread.
  • Confidence: The probability that the point spread pick is on the correct side.


Jaime said…
2-1 entering tonight's game
johnbart said…
Best of luck on tonight's game. My line is pretty much spot on with the Vegas line for this one so I'm interested to see what you're at.
Unknown said…
I suspected SD, IND, PHI, KC this week. Consistent with the 2-1. We shall see. 3-1 would be a nice bounce back
TotalBS said…
Im sure it's INDY but when you see all the public money on one side, head for the hills.

Just like last week when they were all over Indy against Pittsburgh.

Same thing tonight.

Public was all over San Diego this week and, well, errrr....
Unknown said…
TotalBS - I always get hesitant over those games. I was looking over the model I have built yesterday, and there were a couple of games I identified over the last few weeks, where the Vegas line just did not make sense (i.e. I couldn't manipulate the inputs in my model in any manner that would have yielded the Vegas line). Naturally, it was those games where Vegas got it right, and I suspect all the money was on the other team.
Jaime said…
Basing a decision like that on last week's games goes against everything I do here. I do remember reading a scientific study on this that found a very small advantage over a 10 year period.
Unknown said…
There's a few sites out there who track performance of various lines/experts picks over time. I don't have time to look it up right now, but I think I've seen that the initial line typically beats the line at kickoff. I'm guessing it's because the odds makers know best, and the public money moves it away from the best estimate. Lines which move due to injury would make this difficult to analyze.
TotalBS said…
I do statistical analysis using my laptop. I have built numerous numerical models as well. The problem has always been that Stats don't play the games, humans do. And that is why just relying solely on "what the numbers say" is dangerous.

Humans are not robots, they have ups and downs. If the games were decided by the stats then in theory we could figure out the likely winners consistently. But this year especially you can see that these teams are very inconsistent sans a few.

The guys making the lines can sucker the public in by knowing how the public is thinking. With stats you can see where the line doesn't make sense, but that doesn't mean you are still on the right side!....
Unknown said…
You raise a valid point regarding oddsmakers being able to play off of how people bet. That's an interesting thing about the lines though. Casinos are businesses, and they aren't interested in making "bets" - they are in the business of earning their commission. They would prefer to set a line that centers around where the public's money will go than necessarily get the line "right".

You are right about humans playing games, not robots. The counter argument could be that you are using historic data of games, also played by humans, to make decisions.

Jaime has referenced a few times the term 'vegas finish'. If I understand it correctly, it's when a game is completely a blowout, and ends up finishing near the line. These games drive me crazy. Last night's game started to worry me, just because I've been screwed so many times by these Vegas finishes. Indy was up by 30(?), and my wagers looked like a certain win. Next thing you know, it's 16 points, and the Giants are getting the ball back.

I truly prefer betting games straight up, because teams play to win, not to win by 7.5, or to cover a +3.5 line. I think straight up games are where the human factor aligns closer with what the computer generates. Unfortunately, being able to predict that Denver will take out Oakland this week isn't rewarded very well.
TotalBS said…
My point is, using stats alone will only ever get you so far.

Fine if the teams play exactly to their stats but....

As we know they don't week in and week out, and so you need to combine BOTH statistical analysis as well as the situations the teams are in.

The other thing KILLING statistical analysis in the NFL is INJURIES. It is getting impossible to compare teams performances with teams fielding different players each week. Comapring Team A in week 5 when team A had two backups on the offensive line, versus Team A in week 1 when healthy are two distinct animals. This is the biggest challenge for me as a statistical handicapper.
johnbart said…
I think when it comes to injuries there are a surprisingly small number of players in the league that actually could affect a multi-point move in the spread if they went out. With that said it's still hard to sift those games out of your dataset if you're not hand tweaking everything which can be impossible.

The only approach that I think will work is to dive down into a play level analysis of each game to get a full understanding of how valuable each individual player is per team and then base your team ratings on those. This has the benefits of providing you a great starting point at the beginning of the season when rosters have been shuffled.

I'm no where near that point since I'm still using team based stats for my models but it's a good goal to have.
TotalBS said…
I actually have a scrapper to load all the play by play data, box score data and drive chart data. I do the same for the nba. But to try and figure out what each player is worth and his effect on team performance in a practical way? Good luck with that one! Let me know when you figure something out! :)
Jaime said…
TotalBS and Jonathon, this is a very interesting discussion. My reaction:
Stats are not perfect but the hope is that they give us a slight advantage over the market.

Injuries are my weakness. Many of you that have followed me know that I try to stay away from games where there are significant injuries, i.e. injuries from main QB or RB. I wish I had player level data, but even if I did would need time and help in making sense of it.

When the line doesn't make sense could be an opportunity or could be something that the model is completely ignoring. That's why you need humans, I agree. But too much human intervention and you stray away from stats which defeats the purpose. All I do is stay away from a game that has injuries, is not at home, there's huge weather issues or some other thing not accounted by my models that will significantly change the outcome of the game.