Week 7 Machine Learning Point Spread Picks

The miraculous Titans cover on Sunday and a perfect 2 for 2 on MNF put us at 75% ATS for the top 4 picks last week. The models' overall accuracy of 71% ATS across all games gives me lots of confidence going into week 7. This week, 3 games with high confidence and a few more worth mentioning. Let's talk about it.

Philadelphia -4.5
The predictions across the ensemble range from 7 to 10 all with great confidence of 60%+. The Eagles have had a much tougher schedule this year already faced Rams, Steelers, and the Ravens last week. This week they get DeSean Jackson back playing against a #23 passing defense in the Giants. Doug Pederson will out-coach the Giants and has done so on TNF twice in his career and is 5-0 with the Eagles on TNF. Eagles getting 65% of the tickets and 72% of the money. Fly Eagles Fly!

Bet Eagles on TNF. If it's right, signup you won't regret it. You'll make 1000x than the cost of $9.99/mo and since there's only 3 months left it'll be $30 for the season. Get all picks at rokfin.com/nflpickles

Week 6 NFL Machine Learning Point Spread Picks

One of my favorite spots with a 60% ATS trend since 2013 is picked up by the models in 3 games. The 4th pick shows home field AND recency bias. This week looks big with opportunity. Let's dig. Get all picks at rokfin.com/nflpickles

Green Bay -1 
The best offense by far gets back Davante Adams and is facing a cry baby back-to-last year Brady. Although Godwin is back for the Buccaneers, they're "great" defense finally gets to defend a good offense (Bears, Chargers, Broncos and Panthers all have below avg offenses). This looks like a shoot out and Rodgers looks like the likely winner here. All my models have this game going GB by over 10, the average close to 13. The confidence is high because of the situation (model high on the visiting team and successful across years). Although the Packers are getting the majority of the tickets, it's a good sign that even a higher percentage of the money (big bets) are also going to the Packers.

Week 5 NFL Machine Learning Point Spread Picks

Last week, we stuck to our strategy of finding stronger teams beat much weaker teams with single-digit spreads and it paid off big time last week. Bills, Seahawks, Saints, Packers, KC, and Cincinnati all covered. The only disappointment being the Cardinals. This week there's a bit of that, but mostly the theme this week is recency bias. 

Recency bias is a plague and we have spotted it. Line makers and the public seem to only remember one week. You're smarter than that. 4 great opportunities spotted, let's roll.

Here are the games that my models have picked with significantly higher confidence:

Indianapolis -1.5

A home field and recency bias all in one. Cleveland is home therefore they get at least 3 points but probably more by line setters. 49 points last week for the Browns! WOW. Try that against the #1 defense according to DVOA. All my models have the Colts winning by at least 10 points. A better spread would probably be -3 so I'll jump on those points. I've hated on Rivers before and his 24th ranked offense is nothing to be impressed about. The Browns have won 3 in a row (vs Bengals, "W Team", and Cowboys), seems like a great spot to fade them.

For all other picks, go to Rokfin.com/NFLpickles and subscribe. It's probably the best investment you'll make after bitcoin :)

Week 4 NFL Machine Learning Point Spread Picks

The models have detected 5 opportunities. Double-digit predictions on single-digit point spreads. It's only week 4, I know we're just warming up. We're sprinkling some here and there, nothing serious yet. Still, these are some eye-opening numbers. Let's dig in.

New Orleans -4.5

I get it, the Saints defense isn't good. Surprisingly they're still ranked #12 in DVOA. Mostly because they can stop the run, but are having issues with passing. But you know what, Michael Thomas is back! Without him, they still were able to put 30 points on the board. This team scores 40 and Detroit won't be able to get anywhere near that. Detroit's D is the opposite. First, they aren't good either. They're better at stopping the pass which might be bad for Drew Brees. This might either mean they haven't faced a Thomas or that they will depend more on Kamara. This smells blowout, the Saints by two TDs.

Arizona -3.5

A big asterisk here for Hopkins. If he plays, then it's a play if not stay away. I like Arizona coming off a loss and the Panthers coming off a win. People including myself talk a lot about Murray and Hopkins and how dynamic they are. But the Cardinals defense has allowed 61 points in 3 games ranked 7th close to KC and PIT. They're defense DVOA is also ranked 7th and is 22% better against the run. Initial money is going to Carolina which is another reason to wait and see if you can get it at 3. I got Arizona also by more than 2 TDs against potentially the worst defense in the NFL.

For all other picks, create a free account on Rokfin.com/NFLpickles

Week 3 NFL Machine Learning Point Spread Picks

Last week's most confidence pick, the Rams, was such an easy win. A strong offense and pretty good defense against a weak Eagles team all around. Money.

We have two weeks of data under our database. DVOA metrics are out. The strength of schedule is a bit wild but we're starting to look at it. We use all of the tools at hand to continue to dump on mismatch ups with single-digit spreads being careful of a week heavy with injuries. 

LA Chargers -6.5

Carolina is not looking good and now MacCaffrey is out. Chargers according to DVOA are about 5% below average vs Carolina at -20%. They rank similar in offense and the defense is their differentiator. With MacCaffrey out, that offense DVOA should be a lot less. The models even by weighing recent games more heavily, all project the Chargers to win by over two TDs. Chargers are at home too, which is weird why this isn't even a TD for a spread. Milk the weak team. Carolina is 0-2 ATS and will be one of the teams that stays at 0-3 while the Chargers are 2-0 and the market doesn't seem to have caught up. Interesting that most of the money is on Carolina, that's good!

For all other picks, create a free account on Rokfin.com/NFLpickles

NFL Point Spread Picks Week 2, 2020

With only one week in the bag, some of these week's spreads scream recency bias. There are teams that are hyped up too much and others that are just plain garbage. Let's dig in. 

We got 4 visiting teams covering the spread this week. Last year as you may know, we pounded visiting teams because the home field advantage was over rated for some teams. One of the main trends was that strong teams on the road was a killer strategy and it worked. We had over 70% ATS success with visiting teams we found undervalued. Not sure some of these visiting teams are strong, but the home teams seem week. Here are the picks:


I've never been a huge fan of Carson Wentz. Last year they covered 7 of 17 (41%) and last week didn't seem like an improvement. Yes, they have some guys in the OL coming back this week but without Jeffrey they don't seem to be a big threat offensively. They also allowed 27 points against Washington. Meanwhile the Rams looked solid especially on defense. They're gonna come crashing and I expect a few turnovers from the Eagles. The models really like this game and so do the sharps and everyone on the planet but like about 18% which is probably all Eagles fans. Should be an easy win for the Rams.

For all other picks, create a free account on Rokfin.com/NFLpickles

NFL Point Spread Model Results - Week 1, 2020

Week 1 model results are out! They are very interesting. Last year we picked on a lot of visiting teams, but this week the 3 of the top 4 picks with most confidence are home teams. I usually don't bet on week 1 through 4. Definitely when the pick includes teams which changed QBs (e.g. Tampa, Indianapolis) or made significant trades (e.g. Arizona). Stay away would be my advice or sprinkle it in just to make it more fun to watch. 

 There were times I didn't even run the models the first few weeks because only start keeping track of ATS % and ROI until week 5. People have asked me for the results and I think it's good for newcomers to start getting familiar with the system. I didn't deep dive on these as I usually do in looking at DVOA, sharp money, trends and other data I like to geek on. Here are the point spread model outputs for week 1 of the NFL!

You can get the whole list of games from week 1 to week 6 FREE on my channel on Rokfin. Why Rokfin? It's a long story, but it's only $9.99. You get emails and notifications when I post and a bunch of other channels included for the same subscription. If you've been tracking me over the years, you know $50 for the whole year it's a deal when compared to the many fake handicappers out there. I do the research and we all win money.


Why is NFL Pickles so Damn Good?

 Who doesn't like to brag about their accomplishments? I'll make it short, I promise. First a bit of history.

In 2005, as I was finishing my doctoral thesis in Stochastic Games, I met a finance professor who had just published a paper on how he could predict the outcomes of figure skating results. His paper explained the use of an ensemble of models with traditional regression components, human bias, and a situational component. My mind was spinning with ideas and we talked for hours. The ensemble component was amazing and is probably the beginnings of what today is called AI. Human Bias is market inefficiencies; aka prices determined by people. Finally, the situational component (location, ice, competitors) is all the situations that could be captured at the beginning of a game. Although I didn't have much knowledge of the NFL at the time, in a short time I got access to over 20 years of NFL Point Spread Data, and the rest is money.

2019 Results
Thank you to all who support and profit from my work. Last year was an amazing season. Here are some quick highlights:
  • 62.5% ATS for the season
  • 80% ROI if you followed my rules, but likely doubled your money if you spent more during the playoffs
  • 67% during the playoffs with a correct prediction for the Super Bowl which puts me at 12 of 13 during the last 13 years

2019 NFL Spread Record

2019 NFL Super Bowl Point Spread Pick

First a bit of context of who am I and how I've picked the Super Bowl in the past. In 12 years doing this, I've picked correctly the Super Bowl winner 10 of 11 times. Yes, that's 90.9% ATS in the big game. Let that sink in. This season, I got 2nd place in a 1,400 entries content picking every game at 58% ATS. Games with high situational confidence are at 61% ATS and over 60% ROI. Ok enough bragging, good luck going against Mahomes. Here's why the Chiefs are going to win and cover Super Bowl 54.

Kansas City -1 correct NFL point spread pick here's why:
  • Models are detecting a huge superiority in stats for KC. Probably due to the strong offense, but as of late not so bad defense. The situational confidence is at 66.9%.
  • Offense wins championships as analyzed in my previous post. 22 of the last 30 Super Bowls are teams with DVOA higher than 10 aka very strong offense but not necessarily strong defenses.
  • Andy Reid. He's the veteran coach hungry for a ring. His record coming off a bye is 18-3 SU speaks for itself
  • The 49ers defense is amazing, no doubt. The 49ers did struggle against offensive powerhouses in the last few weeks of the season (Saints, Falcons and Rams).
  • KC already had a test in stopping the run against the Titans and they did.
  • The 49ers recently played against a poor coach and depleted Green Bay Packers team. The Packers kept running the ball without success, unable to adjust. Recency bias.
  • Patrick Mahomes vs Jimmy Garappolo, if you bet the 49ers you somehow disregard that the Chiefs have the better captain
  • Experience matters and losing last year in the AFC Championship has given this team the experience needed to win this year. This energized their confidence and attitude which is going to be what matters in this game
I don't do props or other guessing bets. But I still hit 7 of 10 on the podcast double dipping on the Chiefsl I do stick to what I know, the NFL point spread. Here are the models' output for the Super Bowl LIV point spread pick. Let's goooooooo!

Game Vegas Line Estimate Diff Confidence
SAN FRANCISCO - KANSAS CITY -1 -12.6 -11.6 66.9

Defense Wins the Super Bowl: Myth or Fact?

We all want to make a big wager on the big day. I've been really lucky on Super Bowl Sunday in my career going 10-1 ATS. Nevertheless, this year I'm doing a lot more analysis. In a series of posts of deep analysis of this year's Super Bowl 54 posted on my Rokfin channel, we first tackle the elephant in the room.

Does defense wins championships in the NFL?  We couldn't help but noticing that this year's Super Bowl teams are a strong offense against a strong defense. Then the quick thought was "defense wins championships", everyone knows that.  Remember Super Bowl 50 when Payton Manning sucked and yet won the Super Bowl and linebacker Von Miller won the MVP? But as a data scientist, I can't take a mantra and a game I remember and believe it's true.

We dig into the data to prove or debunk this mantra by looking at the last 30 Super Bowl games and each of the team's DVOA.

Check out the article here. It's FREE, but you do have to create an account to read it. It's worth it.