AI Model Ends on High Note, Records Impressive Full Season Record
The Kansas City Chiefs capped off a magical season last Sunday, with a Super Bowl victory. Sportpicker AI also ended the season on a high note, after correctly prediciting the Chiefs win (-1) and 7 of 10 games overall in the NFL postseason, all against the spread. As reported on this blog, the playoff run netted Sportspicker AI subscribers a whopping 225% ROI.
But now that the dust has settled on the NFL season, the question is how did Sportspicker AI fare over the course of the full season. Was the AI model able to match the 70% accuracy rate achieved in the postseason?
The short answer is no… But the fact is that in the world of NFL betting against the spread, 70% over the long term is a fantasy. Point spreads are set so as to make each game a virtual coin flip. So, while hitting 70% over 10 games is a very nice run, it’s simply not possible over the course of a full season. You should be very wary of folks who tell you they can do it–ask them to show you their picks!
So what is a good record for an NFL Handicapper anyway? Most professional handicappers agree that 55% is the appropriate benchmark. Each year, a few will hit this benchmark, but many more will fail. And fewer still will hit 55% year after year. So achieving 55% is the NFL-betting equivalent of an investor beating the S&P 500.
To many amateur bettors, 55% might not sound impressive. After all, to cover the 10% juice in a standard -110 bet, you have to hit 52.4% just to break even. So 55% sounds like a pretty slim profit margin
But sharp bettors understand the math. If you bet 20 units in a week and win 55%, you win 11 units (reduced to 10 by the sportsbook’s juice) and lose 9 units. Your net gain is 1 unit, or 5% of your original 20. So a 55% winning percentage equates to a 5% profit each week. If you can achieve 5% profit each week and reinvest your winnings, over the course of a 17 week season, a $1,000 bankroll grows to $2,292 over a 17 week season–more than doubling your money!
That is why at the beginning of the season, we used the 55% benchmark as our goal for Sportspicker AI – NFL. And we are happy to report that we achieved it! We finished the season at 102-83-4, for a winning percentage of 55.1%. You can see the full log of all our 2019-20 picks HERE, but here is the week-by-week scorecard.
So does that mean our subscribers doubled their money? Well, some might have, but most probably did not. Here’s why.
- Inconsistency – Although we cleared 55% for the season, we were not exactly there every week. Many weeks, the model exceeded the mark, but some weeks it was below. And we had 2 three-week stretches (weeks 5-7 and weeks 13-15) below 55%. That inconsistency hurt our overall ROI.
- Conservative Approach – Most gamblers take a more conservative approach. They risk a relatively small percentage (1%-3%) of their bankroll on each game they bet.
Here’s how a hypothetical $1,000 would have grown for a Sportspicker AI Member who used our picks and wagered 3% of bankroll on each game. As you can see, this approach and these somewhat inconsistent results still produced an almost 30% return on investment, growing from $1,000 to $1,292.
However, there was one more factor in play for Sportspicker AI members. In addition to advising what teams to bet on, we provided bankroll management advice. Using the Sportspicker AI model, we gave pick ratings to each bet, suggesting larger bets on some games than others. Unfortunately, that turned out to be a part of the model that still needs some refinement. For users who applied our pick ratings exactly, the results ended up being just about break even for the season.
Now, break even may not sound too thrilling, but let’s keep in mind that picking NFL games against the spread is probably the most difficult challenge in sports betting. There is so much information and so much action that finding an edge is TOUGH. And remember, the average gambler is lucky to be right 50% of the time, which means losing 4.5% every week (after the sportsbook takes its juice). You can see the affect on bankroll below.
To summarize… As we look back on the first NFL Season for Sportspicker AI, we can unequivocally call it a success. To hit 55% overall accuracy shows that the AI model has real intelligence, and that intelligence should be repeatable. While the pick ratings system needs some refinement, members who simply bet a conservative 3% of bankroll on each pick earned a 29.2% return on investment, which is outstanding.
The overall point is clear: There is a big advantage to Sportspicker AI members who use our AI model’s picks to make sports bets, rather than go it alone.
While the NFL season has concluded, don’t forget that Sportspicker AI has great packages available for NBA Basketball, NHL Hockey, and EPL Soccer. The AI Model is remarkably well calibrated for all 3, so now is the time to join us. Click the button below to learn more.
The news, information, opinion, and recommendations on this website are for education and entertainment only. This information is not intended to violate any local, state, or federal laws, and all persons using this website are solely responsible for complying with the laws of the jurisdictions in which they reside. Unanimous A.I. is not responsible for the accuracy of any predictions made on this website, and does not recommend using the information on this site for wagering. Unanimous A.I. does not intend to encourage or condone gambling in jurisdictions where it is prohibited or by persons who are under the legally required age. In accordance with the Terms of Service, you may not sell, resell, publish, republish, copy, duplicate, reproduce, redistribute, broadcast, license or otherwise commercially exploit or make the contents of this email available to any third party. The Terms of Service that apply to Sportspicker AI may be found here.
Want to learn more about our Swarm AI technology? Check out our TED talk below…