Sports Predictions Guide: Best prediction sites, apps, and their downfalls
Just to be clear up front: I don’t make sports predictions.
I don’t have the computer programming, math skills, or motivation to try to create something better.
This guide is to study sports prediction rankings and models.
Who has the most accurate predictions for today’s college football/NFL/college basketball/NBA/baseball/etc games?
I’m on a journey to research who is most accurate, but also how much can I trust computer model rankings. How do their predictions compare to Vegas? What do prediction models include? Should I use them to place sports bets? If so, what should I consider before betting?
Are prediction models better than Vegas?
I haven’t found one.
And let’s be real, if there was a model better than Vegas, would they make it public?
They’d keep it private, so they could make millions beating Vegas. And even if they weren’t greedy money-grabbers, the second their world-beating prediction model was public, Vegas would incorporate the model into their own spreads. And when Vegas takes it into account, the model no longer has an edge.
So great start here… you’re on an article to read about the best sports prediction models, and I’m telling you they won’t help you beat Vegas.
Actually, that’s not exactly what I’m saying. The best prediction websites and apps can help you beat Vegas, you just have to know you can’t rely on them 100%. They won’t make your picks for you. It’s going to take a little work. So let’s cover what that work is…
So why learn about prediction models?
I’ve hit home the point that a sports prediction site or app that beats Vegas doesn’t exist.
The cool thing about the sports prediction world is there are so sports fans out there–and some of them are genius mathematicians, statisticians, and computer programmers. This means there are communities of people trying to build the best sports predictions.
The result is multiple extremely accurate sites and apps out there. And good news for us… since sports predictions are fun for many people, you can find some very good prediction models offering their results for free or at least part of their results for free.
But if there’s no model that can consistently beat Vegas, what’s the point?
Score and ranking predictions can be a valuable “tool” to make educated sports bets, win a bet against a friend, or fill out a March Madness bracket. Each model is like a different tool in your toolbox. Sometimes you need a wrench, sometimes you need a hammer. The key is knowing the strengths and weaknesses of different models. What do they include? What might they be missing?
Let me use one simple example… let’s say you use TimeTravelSports Prediction Machine (this doesn’t exist). You know it is one of the most accurate prediction sites for football, but it doesn’t include or update based on game conditions. It’s predicting the Patriots are very slight favorites over the Dolphins… but not by much. That doesn’t convince you to bet on the Patriots because Vegas is saying the same and the Patriots are one-point favorites. But you live in Boston and know 4 inches of snow is coming in. And you’ve watched the Belichick-coached Patriots play in the snow, and they always look better in these conditions. Now, you know the accurate model and Vegas agree, but you also know that your accurate model is underestimating the weather conditions. Looks like a Patriot bet to me (and I hate betting on the Patriots).
When you know what a few models include and what they don’t include, you can get more of a general “story” on a game, and this can give you insights into what Vegas might be missing.
What information do sports prediction models include?
Sport prediction models use statistics to crunch a little or a lot of information depending on the model. There are two many categories of information: historical and “live.”
Historical information is the past stats for the teams, and this is used in every model I know of and almost always the most influential part of a model. After all, what better way to rank teams than how they’ve played (or in the case of early games in the season, how their players have played).
“Live” information isn’t used as often but includes information about the specific game. The most common piece of live information is the home team. Almost every prediction takes into account home field because it’s known to make a definite impact in-game outcomes. But there is a lot of other live information that can be included:
- what is the weather?
- are there any injuries to key players?
- any coaching changes good or bad?
- motivation factors: is one team tanking? on the borderline of making the playoffs? rivalry? etc
- many more (many I probably don’t know)
Motivation is the most interesting to me because it’s difficult to quantify in numbers. There would be a few ways models could address it, but they most likely use humans to manually input something like -1, 0, or 1 into the model on how “motivated” the team is. Negative 1 might be a team that’s had fighting in the locker room. Plus 1 might be a team that has good “billboard material” where the other team talked trash the past o
The Top Sports Prediction Sites
I want to be clear on something first:
Many people are searching for the sports prediction model that will beat Vegas. And I hate to be the spoiler, but it’s not out here. That said, using expert gambling strategies and then taking into account the best sport prediction models is going to make you a much, much sharper better.
One other note worth mentioning is there isn’t a consensus “winner” or “top 5.” Models that have been tested over multiple seasons often vary in their rankings. ‘Worlds Best Sports Prediction Model’ may rank #1 in NCAA basketball in 2017, #8 for NFL in 2017, and in 2018 rank #13 for NCAA basketball and #2 for NFL football. There can be a lot of confusion. But these are models we consistently trust.
With that out of the way, here are the top sports prediction sites:
Donchess Sports: best of both world model
Donchess sports is my favorite ranker right now. They are 100% straightforward on their methodology, but they use both statistics analysis of past performance (like most models), but they also use human polls
The human element is important. Not just because it feels good to say humans still have value, but they do. Computers have a harder time understanding the intangibles that might go into a game like motivation to win (what’s on the line), injuries, drama, etc.
You can check out the DRatings here.
ESPN Ratings: best complex data model
ESPN has ratings with different names per sport. I used to think it was a few of the talking-head “experts” they had coming up with who they thought were the best teams, but it’s actually much more of a Vegas-like rating than that.
ESPN uses its own proprietary mathematical model to predict the winner of games in most sports. It’s a complex mix of data they don’t reveal such as offensive and defensive efficiencies, margin of victory, and takes into account things like players lost. It seems more complex than most, but usually is a good predictor
ESPN doesn’t have a main page for all their predictions, but individual pages around their site. Check out ESPN’s FPI (football power rankings) and BPI (defensive power rankings) here: college basketball, college football, NBA, NFL.
Massey Ratings: best no-frills model
Massey ratings are one of the more “basic” ratings that have proven the test of time. By basic, I’m referring to the number of variables and “bells and whistles” thrown into the algorithm. The statistical analysis is complex. And it provides more accurate results than most models.
The primary variables Massey takes into account are scores and margin of victory. Then, home or away is used to help come up with a final prediction. This can work pretty well, and comes close to Vegas many years in accuracy.
I think the Massey ratings are worth checking out to see what the fundamental math is telling us about who’s going to win.
You can check out the Massey Ratings here.
Why do I only list 3 prediction models?
There are 100s of prediction models you can find with a quick google search.
The problem is – data overload. There are only so many ways you can crunch past sports data to come up with a score prediction.
You just have to find a few you trust and like. The 3 above are my go-to’s (along with Vegas stats). I think adding any more information just makes things more confusing, and these 3 models give 3 different “styles” to crunching the numbers: lots of data, less data, and including a human element.
Is there an app for that?
Not really. There are things called “sports prediction apps,” but the apps I’ve seen don’t predict sports scores, they let you predict them. In other words, they’re like betting apps where you don’t have to wager true money but can win prizes.
FAQ: Which sport is most predictable?
I get this question a lot–which sports is easiest to predict?
I haven’t found a scientific answer that says prediction models say [insert sport here] is the easiest to predict. However, there is analysis on which sports are easiest to pick the champion. The easiest to predict is tennis, and the most difficult to predict is golf. Of the major US sports, NBA is by far the easiest. Interestingly, NCAA basketball is one of the most difficult to predict (maybe because the number of teams — but we all know the majority of teams aren’t a factor early on).
Now, before you go out and become a tennis betting expert and give up on college basketball, you have to think one step further.
Vegas knows tennis is the easiest to predict. It knows NBA is easier than the NFL and college football and basketball. So the sportsbooks pay you less to predict the tennis winner. They pay you more to predict the college basketball winner.
Makes sense.
And that’s another reason it’s so difficult to use prediction systems to beat Vegas, but it’s still valuable information. Just like the sports games prediction, the championship predictions can help give you a steady guide on who’s good through the year when there’s a lot of mid-season noise (random upsets, injuries, etc).
Where you’ll get your edge
Every sports bettor should use prediction models to help make their picks.
But smart sports bettors know that Vegas also knows these predictions, so they’re a good tool but not an edge that will make you profitable.
So how do you get an edge over Vegas? I’ve laid out everything I’ve learned in my free guide to being profitable in sports betting. It’s continuously being updated and covers everything you need to know like how often pros win, and how you can work your way up to matching them.