I love betting on baseball, and I want everyone to try it. I think if more people knew about baseball betting models they’d be more inclined to drop a few bucks on MLB games.
Baseball is the easiest sports betting market. Oddsmakers do much better against football and basketball bettors. You can look at revenue reports from sportsbooks for proof. Something about the sport’s long schedule and ever-changing rosters create powerful trends that bettors can work to their advantage. When you look at trying to determine the profitability of a bet on something like who will be the next pope, you’ll see that baseball is even more superior than that.
This post explains the baseball betting model, offering MLB tips, strategies, and systems along the way to help new baseball bettors in 2023 beat the books.
If you just want to win a million dollars, this post might not be for you, but I have a new post about how to win a million dollars.
Table of Contents
What Is a Baseball Betting Model?
When I was a kid I made model kits with my dad. The smelly glue, tiny plastic templates, weird hobby shop employees – the whole nine yards. I was amazed at how a cardboard box full of junk could turn into a cannon that shot little cannonballs or a destroyer that really floated on water.
In some ways, a baseball betting model is like those plastic hobby kits that produced so many submarines, helicopters, and Formula 1 cars. Starting with something abstract, we worked to piece together something functional.
Any system that uses statistics or analysis to create projections of baseball wins is technically a baseball betting system. Baseball betting modelers start with abstract numbers and hope to put together a system that returns a profit with little effort.
The best baseball betting model would spit out profitable picks with no input from the modeler. I don’t think anyone has consistently pulled this off yet – if they have, they aren’t going to sing it from the mountaintops.
So many models like this exist that it’s hard to define or categorize them broadly. The best I can do is to say that the goal of an MLB betting model is to make successful projections.
Examples of MLB Betting Models
I’d guess hundreds of successful MLB betting models exist. I’ll cover just a couple of them to give people an idea of what they look like and how they work.
I know lots of baseball bettors that use some version of the Base Runs system. I’ll start by describing that.
1- BaseRuns Baseball Betting Model
BaseRuns is a formula designed to estimate how many runs a team would be expected to score or allow based on their offensive and defensive stats. Context doesn’t come into play when you calculate BaseRuns. It’s more like a description of how a team “should” perform against a generic opponent on any given day.
A simple baseball betting model would compare a team’s BaseRuns to their actual number of runs produced, and then project forward based on that calculation. If a team scored more runs than their predicted number, you’d back that team over a squad that underperforms its BaseRuns number.
You can calculate BaseRuns for individual players or team performances. You can also calculate BaseRuns for pitchers, useful if you want a picture of how a particular pitcher “should” perform regardless of context.
How is BaseRuns calculated?
It’s complicated.
The formula looks like this:
Raw BaseRuns = [(A * B) / (B + C)] + D
A = H + BB + HBP – (0.5 * IBB) – HR
B = 1.1 * [1.4 * TB – 0.6 * H – 3 * HR + 0.1 * (BB + HBP – IBB) + 0.9 * (SB – CS – GDP)]
C = PA – BB – SF – SH – HBP – H + CS + GDP
D = HR
Confused?
It’s not a big deal. You can find the stats needed to make these calculations for free all over the Internet.
As of the 2021 season, the top players in terms of BaseRuns/game are Mike Trout and Christian Yelich, good for 0.96 and 0.87 runs per game respectively.
2- Baseball Third-Order Wins Model
A team’s third-order win percentage is a projected winning percentage based on that team’s stats and adjusted for the quality of their opponents. In short, it’s a projection of how many games a team “should” win based on things like the number of runs scored and the number of runs allowed.
The goal with third-order win models is to forecast future game outcomes successfully.
You can also use it to look back at a season for analytics purposes.
Look at the 2021 Seattle Mariners:
They scored 697 runs and allowed 748. Based on those numbers, they should have ended the season with a record of 75-87. Their final record was 90-72, missing the postseason by just 2 games instead of 19 or 20. The 2021 Mariners outperformed their Pythagorean win-loss total.
You can use this during a season to identify teams likely to surge or tank. Consider the 2023 Toronto Blue Jays. They’ve scored 139 runs and allowed 147. A third-order wins model suggests the Blue Jays will finish the season with 76 wins. That’s just 56 more wins out of 125 games remaining.
A basic third-order betting model in 2023 would suggest that you fade the Blue Jays since they’re likely to go 56-69 through the end of the year.
3- The ATC Projections
You don’t have to drive yourself crazy with math to use baseball betting models to win more bets. Plenty of betting models exist and are available online, some of them for free use.
I think the best of these freely-available player projections is ATC. Created by Ariel Cohen, winner of the 2019 FSWA Baseball Writer of the Year award and a total fantasy baseball stud, these projections get into the nitty-gritty, projecting traditional and advanced stats for every player in the league.
It’s not hard to put together a simple projection for any team in any situation given the projection numbers available from ATC through sites like FanGraphs. Compare the starting pitchers and batting lineups, come up with your own projection of how the game might go, and bet based on your new insight.
Do Baseball Betting Models Work?
Yes, baseball betting models work if you use the definition I’ve written above.
It is possible to create a successful baseball betting model.
But I don’t think anyone has created a perfect baseball wagering system yet.
They require tweaking, maintenance, and attention.
A good baseball model is like a garden, with new needs popping up and old habits dying off, responding like a garden does to changes in the weather. It’s a long season, moving from cold to hot and back to cold.
It’s hard to create a successful model that runs more than a single season. Changes in league dynamics and rules, not to mention changes in rosters and the shifting abilities of players as they develop and fade, mean a system is only as good as its input.
So, how do we know if a betting system is working?
A Simple Explanation of Return on Investment (ROI)
Successful baseball modeling is all about profit.
If your model doesn’t help you make money betting on baseball, then it’s just a fun mathematical toy that nobody will care about.
Return on investment, often shortened to ROI, is a good way to measure the success of a baseball model. You calculate ROI by dividing your profit by your expenses. If you made $1,000 but it cost you $1,200 in bets to get there, you have a negative return. If you only bet $800 to win that $1,000, you have a positive return.
The goal of every MLB model is to create a positive return throughout the long baseball season. Models don’t account for MLB rain days, but they’re still plenty useful, regardless of whether you’re betting at offshore sportsbooks or with a local bookie.
How to Build Baseball Betting Models
You have to start with statistics.
To keep things simple in this example, I’ll focus entirely on offensive statistics. If you want to go deeper, you can add in defensive stats and pitching performances.
I look at the past three seasons of a team’s performance when coming up with these numbers. This year, I’ve chosen to focus on slugging percentage, isolated power, and on-base percentage. I feel like these three stats go a long way toward predicting run production. I’ll keep things simple by taking an average of three years of stats in those categories.
I’ll project run production for the Houston Astros based on their stats from 2019, 2020, and 2021.
The Astros’ average slugging percentage in those years was .449. They averaged a .334 OBP and an ISO of .185. Based on these stats and my own model’s calculations, the Astros should produce 892 runs in the 2023 season. That should be good enough for 119 wins.
I don’t stop there. I like to look at a really basic Pythagorean model of wins and losses. In years past, the Astros underperformed their Pythagorean number by 4-5 wins. I’ll go ahead and book the Astros for 115 wins just to be safe.
Using the same system above, I calculated that the Angels (the Astros’ biggest competition in the AL West) are set up to win 112 games.
It’s going to be a tight race in the AL West in 2023.
Conclusion
A guy I know who uses a complicated MLB projection model was thrilled when he went 254-258 last season. That’s a winning percentage of 49.6%. How’d he put together a more than 8% ROI with that record?
He stuck to bets in the neighborhood of +140, as much as possible. Remember that favorites usually get plus money on the run line. Good projections can help bettors identify valuable money line dogs, too.
Therefore, baseball models are an achievable strategy for winning more MLB bets. A good baseball model returns a profit of some kind, usually by helping bettors identify value bets. Use baseball betting models to find valuable plus money situations that help wipe out the impact of sportsbook vig.
