Last Year Recap and This Year's Improvements
What can I do for you? It's the question many business ask their clients for customer retention. Most companies don't know the answer but for this blog the answer is easy. Accurate NFL spread predictions! Yet last year was one of the most horrendous years in my 8 year career doing this. It all started as a research project and now is a part-time job, a hobby, an obsession.
The Bad
Lets analyze last year's results and discuss what we'll be doing to improve accuracy and deliver great free picks to all of you. We ended with a miserable 46% ATS, week 14 went 0-4 and there were only two weeks with perfect 3-0 picks. Although we never bet the bottom 3 teams in terms of ATS Standings, we did losing bets on San Francisco, Atlanta, and Tampa Bay. Imagine if we would've bet on Minnesota and Cincinnati every week?!
The Good
From week 15 to the Super Bowl we went 14 of 21 for 67% ATS! That's a huge improvement and I did one major thing we've never done before that we'll continue doing this year. I started including an NFLpickles expert average into our models. That is, we took the top 15 handicappers from a site, scraped their average spread, included in our models to adjust our predictions. What this does is reassure a prediction of our that's accurate (something distant from the spread) and remove games that may seem good but after the prediction adjustments they do seem too close to the spread. Other things we'll do to improve:
The Ugly
This year, as we used to do, will not post official picks until week 5. I will run the data and give you results but I suggest waiting to week 5. Good luck!
The Bad
Lets analyze last year's results and discuss what we'll be doing to improve accuracy and deliver great free picks to all of you. We ended with a miserable 46% ATS, week 14 went 0-4 and there were only two weeks with perfect 3-0 picks. Although we never bet the bottom 3 teams in terms of ATS Standings, we did losing bets on San Francisco, Atlanta, and Tampa Bay. Imagine if we would've bet on Minnesota and Cincinnati every week?!
The Good
From week 15 to the Super Bowl we went 14 of 21 for 67% ATS! That's a huge improvement and I did one major thing we've never done before that we'll continue doing this year. I started including an NFLpickles expert average into our models. That is, we took the top 15 handicappers from a site, scraped their average spread, included in our models to adjust our predictions. What this does is reassure a prediction of our that's accurate (something distant from the spread) and remove games that may seem good but after the prediction adjustments they do seem too close to the spread. Other things we'll do to improve:
- Change the algorithms to treat outliers differently
- Pick a pony (team) early on and bet on them more often
- Integrate experts average spread into scoring models
The Ugly
This year, as we used to do, will not post official picks until week 5. I will run the data and give you results but I suggest waiting to week 5. Good luck!
Comments
We're being a bit more public with our math and picks this year as well. If you're interested you can check it out at https://blog.therealline.com/
Proost!