A Data-Driven Look at the Sportsbook Market for the Tech-Curious

Start with a single number. In 2024, people in the United States legally wagered close to $148 billion on sports, and the companies that accepted those bets kept roughly $13.7 billion of it. Both figures set records, and both grew by more than a fifth in twelve months. If you spend your time thinking about platforms, pricing engines, and the way software quietly reshapes old habits, that pair of numbers is more interesting than any single highlight-reel bet. It describes an industry that went from mostly illegal to mostly digital in well under a decade, and it did so on the back of probability models, location software, and real-time data feeds.

This is a market worth reading the way you would read any other technology sector, by the metrics rather than the marketing. The clearest way to do that is to separate the money people put in from the money operators keep, and then watch how those two lines move relative to each other over time. A useful starting reference is the running sportsbook market analysis that Lineups maintains, which tracks handle, revenue, and state-level results rather than promoting any single app. Treating the sector as a dataset rather than a pastime is the only way to distinguish hype from structure, and structure is what this article is about.

None of what follows is betting advice, and none of it assumes you have ever placed a wager. The point is to give a technically minded reader a clean mental model of how the sportsbook business actually works as a system: what the numbers mean, where the models come from, why almost all of it lives on phones, and how to judge the confident market claims you will keep seeing in your feed.

How a Niche Pastime Turned Into a Data Business

For most of modern American history, betting on sports outside of Nevada existed in a legal gray zone, run through informal networks and offshore websites with no oversight. The structural change came in 2018, when the Supreme Court struck down a federal law that had blocked most states from authorizing the activity. That single ruling handed the decision to individual state legislatures, and the buildout that followed looked less like a cultural shift and more like a software rollout repeated fifty times.

What made the rollout move so fast was that the product was ready to be digital from day one. A sportsbook is, at its core, a pricing system attached to a payments system attached to a compliance system. Once a state said yes, operators did not need to pour concrete or open buildings. They needed to ship an app, connect data feeds, and pass a regulator’s technical review. The barrier was code and licensing, not real estate, which is why a market that barely existed in 2017 was handling nine figures a year by the early 2020s.

That origin matters for how you should read the sector now. This was never a slow analog business that later added a website. It was born as a data product in nearly every state that legalized it, which is exactly why the growth curve looks more like that of a consumer app category than the traditional gambling industry. The numbers below are the output of that system, and they only make sense once you know what each one is measuring.

Reading the Market by Its Two Core Metrics

The first thing to learn is that the headline dollar figures describe two completely different quantities, and confusing them is the most common mistake in coverage of this industry. The first is handle, which is the total amount wagered. Every dollar placed on a bet counts toward handle, including dollars that are won and immediately wagered again. Handle is a measure of activity and volume, not profit. When you read that a state did billions in betting, that almost always means handle.

The second metric is revenue, sometimes called gross gaming revenue. This is what operators actually keep after paying out winning bets. It is a far smaller number, and it is the one that matters for whether the business is healthy. The ratio between the two, revenue divided by handle, is called hold. Nationally, hold in 2024 sat around 9.3 percent, meaning that for roughly every $100 wagered, sportsbooks kept a little over $9 and returned the rest as winnings. Hold is the closest thing this industry has to a margin, and it drifts slowly upward as operators get better at pricing and as bettors gravitate toward higher-margin bet types.

Understanding the hold lets you double-check almost any claim about the sector. If someone quotes a giant betting figure, ask whether they mean handle or revenue, because the two differ by a factor of roughly ten. A state can post an enormous handle and still generate modest revenue if its bettors win that month, since hold fluctuates with real-world results. Over a long enough window, the percentage stabilizes, which is why analysts trust annual holdings far more than any single month. The metric, not the anecdote, is the signal.

What the Numbers Actually Say

With the definitions in place, the recent figures are no longer abstract. The table below compiles the core data points from the 2024 reporting year, the most complete annual picture available, along with the units in which each is measured. Read it as a snapshot of a system rather than a scoreboard, and notice how the small revenue figure sits underneath the very large handle figure.

A few patterns jump out once the figures sit together. The gap between handle and revenue is the hold percentage made visible, and it is narrow on purpose, since a competitive market keeps margins thin. The growth rate tells you the category is still expanding quickly rather than leveling off, driven partly by new states coming online and partly by existing bettors wagering more often. And the online share, sitting above ninety-five percent, is the single most important fact for a technically minded reader, because it confirms that this is overwhelmingly a software market wearing the costume of an old vice.

The state concentration is its own story. A handful of large states account for a disproportionate share of the national total, with New York alone generating revenue in the low billions. That concentration reflects population, tax structure, and how early each state opened, and it means national averages can hide enormous local variation. Any serious read of the market has to account for the fact that a few big jurisdictions move the aggregate numbers far more than dozens of smaller ones combined.

Where the Models Actually Come From

Underneath every odds line is a probability estimate, and this is the layer that should interest anyone who works with data. A betting price is not a guess pulled from the air. It is a forecast of how likely an outcome is, encoded as a number, and then shaded to account for the operator’s margin. When a line implies a team has a 60 percent chance of winning, that figure comes from statistical models fed by historical results, player data, injury reports, and, increasingly, live in-game inputs. Reading odds is really reading a probability distribution with a fee attached.

Those models have grown sharper as the surrounding technology has matured. The same broad shift that is reshaping every other corner of software, the move from systems that respond to prompts toward systems that plan and adjust on their own, shows up here too. The plain-language way TechAcute explains how IBM frames generative and agentic AI is a useful way to think about what modern pricing engines are becoming: not static calculators but adaptive systems that continuously update their estimates as new information lands. A line that moves the moment a starting player is ruled out is a model re-pricing in real time, not a human scrambling.

For a data-literate reader, this reframes the whole product. The bettor on one side is making a forecast. The operator on the other side is making a more disciplined forecast, then charging a spread between the two. Over millions of events, the side with the better model and the built-in margin wins on average, which is precisely why hold stays positive year after year. The interesting question is not who won a given game. It is whose probability estimate was better calibrated, and what data made it so.

Why the Whole Market Migrated to Phones

The statistic that more than ninety-five percent of legal wagering now happens online is easy to skim past, but it explains nearly everything about how the industry behaves. Physical sportsbooks still exist in casinos, yet they are a rounding error compared to the mobile total. The market did not gradually digitize. It launched digitally and barely looked at the counter.

This matters because it puts the sector squarely inside the same design and distribution pressures as any consumer app. Customer acquisition is funded by advertising and promotional credits. Retention runs on push notifications, live betting, and interface design. The competition that decides who wins a state is less about who offers a marginally better price and more about who builds the stickier product and the smoother onboarding. The digital share of the broader commercial gaming business has climbed steadily, from a small slice a few years ago to roughly a third of all commercial gaming revenue, with sports betting as the engine driving that growth.

There is a regulatory dimension to the phone-first reality as well. Because federal rules generally require a legal bet to stay inside the state where it is placed, every app has to confirm your physical location before it will accept a wager. That means a constant background layer of geolocation, identity verification, and age checks, all running invisibly each time someone opens the app. For a technology reader, that compliance stack is half the actual engineering challenge, and it is the part that rarely shows up in the marketing.

The Geography Problem: One Country, Many Rulebooks

The 38-states-plus-DC headline hides a messier truth. Legal status is not a single switch but a spectrum, and the differences between jurisdictions are large enough to break any tidy national summary. Some states allow full online betting with many competing operators. Some permit only a single state-run app. Some authorize betting only inside physical casinos, with no phone option at all. And a meaningful number still entirely prohibits the activity.

This fragmentation is a direct consequence of the 2018 ruling, which handed the decision to each legislature rather than setting a single national standard. The result is that an operator does not build a single product for a single country. It builds and maintains a separate, separately licensed version for each state it enters, each with its own tax rate, approved bet types, and regulator to satisfy. The engineering and legal overhead of running what amounts to dozens of parallel deployments is a real and underappreciated cost of doing business in this market.

For anyone analyzing the sector, the lesson is to distrust any clean national story. Tax rates alone range from modest to punishing depending on the state, and those rates shape which markets are profitable and which are barely worth entering. A company can be thriving in one state and quietly losing money in another, and the blended national figures will tell you none of that. Real analysis lives at the state level, which is exactly why the better data tools report results by jurisdiction rather than as a single number.

What a Technical Audience Should Watch Next

If you want to track where this market goes, watch a few specific signals rather than the promotional noise. The first is the hold percentage. As operators push higher-margin bet types, such as same-game parlays that combine several outcomes into one ticket, the national hold has been creeping upward. A rising hold means the industry is extracting more revenue from the same handle, which is a structural change, not a one-month swing.

The second signal is the spread of live, in-play betting, where wagers are placed and re-priced continuously during a game. This is the most data-intensive part of the product, demanding low-latency feeds and models that update second by second. Its growth is a direct proxy for how much computational sophistication is moving into the sector. The third signal is consolidation. As promotional spending normalizes and the cost of competing in each state remains high, expect fewer operators to hold larger shares, reflecting the same maturation curve most app categories follow once the land-grab phase ends.

The last thing to watch is the data layer itself. Independent platforms that aggregate odds, model outcomes, and report results have become a meaningful part of how both casual followers and serious analysts read the market. That mirrors what happened in finance, where independent data and analytics tools grew up alongside the exchanges they tracked. As the sportsbook market matures, the value increasingly lies in who can interpret the numbers well, not just who can take the bets.

How to Read Market Claims Without Getting Fooled

Because this sector is loud and money-soaked, it generates a steady stream of confident statistics, many of them mangled. A short checklist keeps you grounded. First, always ask whether a quoted figure is handle or revenue, since the two differ by roughly a factor of ten and are routinely swapped to make a number sound bigger. Second, check the time window. A monthly figure distorted by a few lucky bettors is not an annual trend and only stabilizes over long periods.

Third, ask whether a number is national or state-specific, because the heavy concentration in a few large states means national averages can badly misrepresent any single market. Fourth, trace the source. The most reliable figures come from state regulators, who collect them as a condition of licensing, and from industry bodies that aggregate those filings. The trade reporting behind the record totals from the 2024 season traces its figures back to that regulatory data rather than to operator press releases, which is the standard any serious claim should meet.

Apply that skepticism, and the sector becomes far easier to read. Strip away the promotional language, and you are left with a clean, fast-growing software market built on probability models, location verification, and state-by-state compliance. The bets are the surface. The data is the business. For a reader who already thinks in systems, that is the version of the story actually worth following, long after this season’s records have been broken by the next ones.

Frequently Asked Questions

What is the difference between handle and revenue in sports betting?

Handle is the total amount of money wagered, counting every dollar placed, including winnings that are bet again. Revenue is the much smaller amount operators keep after paying out winning bets. The two differ by roughly a factor of ten, so any market claim is far clearer once you know which one is being quoted.

Why does almost all legal sports betting happen online?

The market was built digitally from the start because launching an app is far cheaper and faster than opening physical venues. Over 95% of legal wagering now happens on phones, which places the whole sector under the same design, distribution, and retention pressures as any other consumer app category.

How do sportsbooks set their odds?

Odds are probability forecasts encoded as prices, generated by statistical models fed with historical results, player data, injury news, and live in-game inputs, then shaded to build in the operator’s margin. Modern pricing engines update continuously as new information arrives, which is why a line can move the instant a player is ruled out.

Is legal sports betting the same everywhere in the United States?

No. Legal status varies widely from state to state, since the 2018 Supreme Court decision handed the choice to individual legislatures. Some states allow full mobile betting with many operators, some permit only a single app or in-person wagering, and some prohibit it entirely, each with its own tax rate and rules.

Where do the most reliable sports betting statistics come from?

The most trustworthy figures come from state regulators, who collect them as a licensing requirement, and from industry bodies and independent data platforms that aggregate those official filings. Numbers sourced this way are far more dependable than figures pulled from operator marketing, and they are usually reported state by state rather than as a single national total.

Photo credit: All images shown were taken by Idris Bello.