Sports Betting vs. Investing: Where the Expected Return Comes From

“some youngs: the stock market is a totally uncertain, high-risk gamble anyway, so we might as well move our savings to actual gambling, which is more fun…”

@DKThomp, August 2026, posted with a chart of the S&P 500 up 942% since inception.

The conclusion is right and the chart is the wrong way to get there. A picture of the S&P 500 taken in 2026 makes a century of American equity returns look like something an investor in 1910 could have counted on. Nobody could. The argument for owning businesses instead of buying betting slips does not depend on hindsight, and it is stronger without it.

It also matters more than it used to, because people are acting on the premise. In Betterment’s 2026 Retail Investor Survey, 52% of the Gen Z respondents said they had redirected money originally meant for investing into sports betting at least once in the past year, and 26% described sports betting as a deliberate part of their long-term financial strategy.¹ Read that denominator carefully. These are 1,000 self-identified retail investors recruited from an online panel, split evenly across four generations, so each generational figure rests on roughly 250 responses. The number describes 250-odd young people who already invest and agreed to take an online survey, a much narrower group than Gen Z.

Even read conservatively, the gradient is steep, and Bloomberg’s write-up of the same survey found that only about a third of Gen Z investors avoid sports betting entirely, against 63% across all four generations.² So the question is worth answering properly, and it is really two questions that happen to share a subject:

• What portfolio weight should sports betting get? This has a clean answer, and the answer is zero.
• Should a financial plan contain a gambling budget? This is a genuine question about spending, and for some people the answer is yes.

Collapsing them produces the two bad takes that dominate this topic: that anyone who bets is an idiot, or that a little speculation is healthy for your portfolio. The evidence supports neither.

Where the expected return comes from

Start by discarding the hindsight argument, because it is the weaker case even though it reaches the right conclusion. Philippe Jorion and William Goetzmann assembled capital appreciation indexes for 39 equity markets going back to the 1920s, precisely because estimates drawn from US data alone are suspect. From 1921 to 1996, US equity prices appreciated 4.32% a year in real terms, the highest in the sample, against a median of 0.75% across all 39 markets.³ Among the markets whose price histories ran unbroken through the century, without closures, wars, or expropriations, the median was 2.35%. That last number strips out the markets that closed or were expropriated, and it is almost always the one left out. Their conclusion was that the large US equity premium is “at least partly the result of conditioning estimates on the best performing market.”

Note what that does and does not concede. Equity investors did face enormous uncertainty. Some national markets did go to zero. The case for owning stocks was never that the outcome was assured.

The case is structural. A share is a residual claim on a business that sells goods and services, generates cash flows, reinvests some of them, and distributes the rest to its owners. Equity investors earn a positive expected return because they own those cash flows and bear the risk attached to them. Positive expected return is a very different promise from a guaranteed one, and it is the only promise the argument needs.

A sportsbook customer is in close to the opposite position, transacting against an intermediary whose revenue is the aggregate customer loss. In 2025, regulated US commercial sportsbooks reported $16.96 billion of revenue on $166.94 billion of wagers.⁴ Those two figures imply that books kept about 10.2 cents of every dollar wagered. That ratio is not the expected loss on any particular bet, since it reflects the mix of straight bets and parlays and the effect of promotions, but the direction of the aggregate transfer is not ambiguous.

The mechanism is easiest to see on a single straight bet at the standard price of −110. You risk $110 to win $100. If the underlying event is a genuine coin flip:

E[profit] = 0.5 × $100 − 0.5 × $110 = −$5.00

That is an expected loss of 4.55% of the money you put at risk on every bet, and it means you must win 52.38% of your bets simply to break even before taxes.⁶ Nothing about knowing football changes the price.

Sportsbooks are also not neutral middlemen balancing two sides of a market. Steven Levitt studied roughly 20,000 wagers placed by 285 bettors in an online handicapping contest over the 2001–02 NFL season and found that in nearly half of all games, at least two-thirds of the bets landed on one side. Bettors systematically favor favorites and, to a lesser degree, visiting teams, and books price into that bias rather than neutralizing it. Levitt estimated the resulting gross profit at 6.16% of the amount bettors stood to win, roughly 5.6% of the dollars actually risked, against 5.0% for a book that simply balanced its exposure. That is a 23% increase in profit from pricing against a known behavioral tendency.⁶

That leaves a gap between the 4.55% a standard straight bet costs and the roughly 10% books keep in aggregate, and parlays appear to explain much of it. New Jersey’s gaming regulator breaks its revenue out by wager type, and in 2025 New Jersey sportsbooks held 19.2% of every dollar wagered on parlays against about 4.9% on everything else. Parlays were 32% of the money wagered and 65% of the revenue.⁵ The product that gets marketed hardest is the one with roughly four times the margin.

The vig is a transaction cost

There is a useful parallel in the retail investing literature, and it cuts closer than the usual comparison does.

Brad Barber and Terrance Odean studied 66,465 households at a large discount broker between 1991 and 1996. The households that traded most earned an annual net return of 11.4%, while those that traded least earned 18.5% and a value-weighted market index returned 17.9%.⁷ That gap is usually retold as evidence that active traders pick worse stocks. The paper found the opposite. Gross returns were nearly identical between the frequent and infrequent traders. Almost the entire seven-point difference was commissions and the bid-ask spread.

A toll charged per transaction, levied regardless of whether the underlying judgment was any good, is exactly what the vig is. The difference is one of size and escapability. The retail investor of 1995 was paying a few percent a year in frictions and can now avoid nearly all of it with an index fund. The bettor pays roughly 4.55% of every dollar risked at the standard price, on every bet, and there is no version of the product without it. The margin is the business model.

What an asset has to do to earn a place

Someone always raises diversification at this point. Sunday’s results are essentially uncorrelated with the stock market, so does a betting sleeve improve a portfolio?

Low correlation on its own does not qualify anything as a portfolio asset. An investment with a negative expected return can still be worth holding, but only when it reliably pays off in states of the world where an extra dollar is worth a great deal. Homeowners insurance is the obvious case. You accept a negative average monetary return because the policy delivers money in precisely the scenario that would otherwise be ruinous. The payoff is timed to your need.

A Cowboys-Eagles wager does nothing of the sort. It resolves at random with respect to your job security, your housing costs, your children’s tuition, and your retirement date. Adding independent negative-expected-return bets to a portfolio introduces a new source of randomness without introducing a new source of return, and without buying protection against anything you face. In mean-variance terms, an asset earns its weight through expected return or through covariance that is valuable when you need it. Being uncorrelated is neither.

What happened to household balance sheets after legalization

Until recently the case against sports betting rested on arithmetic about the house edge. Since Murphy v. NCAA in 2018, 38 states have legalized mobile betting at different times, which gave researchers something better: a staggered natural experiment across millions of household balance sheets. Three large studies landed in 2026, and they agree on direction while disagreeing substantially on magnitude.

Scott Baker, Justin Balthrop, Mark Johnson, Jason Kotter and Kevin Pisciotta used transaction data from 184,000 households in Gambling away stability: Sports betting’s impact on vulnerable households, published in the Journal of Financial Economics. Following legalization, betting spread quickly, and it did not displace other gambling or ordinary consumption. It displaced saving. Net investment in brokerage accounts fell about 20%, with the effect concentrated among frequent bettors and low-savings households. The heaviest bettors cut their investment deposits by more than half, implying that roughly 20 cents of every dollar deposited into a betting app never reached long-term savings.⁸

Brett Hollenbeck, Poet Larsen and Davide Proserpio approached the question through credit files instead, using a panel of roughly 7 million consumers in Management Science. General legalization was associated with an average credit score decline of about 0.7 points. Where online and mobile betting was introduced in states that already permitted retail betting, the estimated decline was about 12 points.¹⁰ That conditioning gets dropped in most summaries, and it matters. The 12-point estimate measures what changes when a sportsbook moves into everyone’s pocket in a state that already allowed betting in person. They also report increases in bankruptcy filings, debt sent to collections, credit card delinquencies and auto loan delinquencies, and find that lenders respond by restricting credit.

Jacob Goss and Daniel Mangrum, working with the New York Fed Consumer Credit Panel, found smaller effects again. Legalization raised sportsbook spending roughly tenfold and take-up by 3.1 percentage points. Delinquency, defined as being at least 90 days past due on any credit product, rose 0.31 percentage points against a 10.71% baseline. Median credit scores fell by about a point, though that estimate is not statistically distinguishable from zero, so the delinquency result is the one carrying weight. The damage was concentrated among younger borrowers: for those under 40, credit card delinquency rose 1.02 percentage points and auto loan delinquency 0.55 points.¹¹ Because only about 3% of the population newly takes up betting after legalization, scaling that population-level effect by take-up implies delinquency increases on the order of 10 percentage points among the people induced to bet.¹²

Where the studies disagree

A 0.7-point average credit score decline and a 12-point decline are not the same finding, and neither is a 1-point median decline. The three studies use different data, different comparison groups, and different definitions of the treatment. Anyone telling you the household cost of legalized sports betting is settled to within an order of magnitude is overstating what exists. What the studies do agree on is the sign: easier access to betting reduces saving and worsens credit outcomes, and the burden falls hardest on younger borrowers and households with little cushion.

They also disagree about whether the damage reaches the severe end. Hollenbeck and his coauthors report increases in bankruptcy filings. Goss and Mangrum tested for exactly that and found nothing: their bankruptcy and foreclosure estimates are small, wrong-signed, and statistically insignificant. They give two reasons, and both are worth carrying. “Bankruptcies are relatively rare events that typically occur after extended periods of financial distress, so the null result may also reflect insufficient time for effects to materialize.” And substantively, the harm they can measure sits in unsecured products like credit cards and auto loans rather than in mortgages and home loss.¹¹

It is worth contrasting all of this with the survey evidence that preceded it. In July 2019, MagnifyMoney asked 1,082 Americans about risk and found that 55% considered investing as risky as gambling, and one in four thought the odds of making money in the stock market were the same as the odds of making money gambling. It also found that self-described gamblers were more likely to invest than non-gamblers, 72% against 56%, and far more likely to trade actively, 57% against 30%.¹³ That is a cross-section, and it describes a personality rather than a mechanism. People with an appetite for risk do both. The 2026 studies identify something different, which is substitution at the margin when access suddenly gets easier.

The same impulse shows up inside a brokerage account

The most useful thing about this research for a DIY investor is that calling something a stock does not make buying it investing.

Alok Kumar showed that individual investors disproportionately demand lottery-type stocks, which he defined not by any single threshold but by a percentile sort on three characteristics at once: low price, high idiosyncratic volatility, and high idiosyncratic skewness. The socioeconomic factors that predict spending on state lotteries also predict investing in these stocks, and a one standard deviation increase in the weight assigned to them corresponded to an additional 3.276% of annual risk-adjusted underperformance.¹⁴

The substitution runs in both directions, which is the part that makes this more than a metaphor. Daniel Dorn and Paul Sengmueller found that brokerage clients who reported enjoying investing or gambling turned over their portfolios at twice the rate of their peers, and the effect survived controls for gender and for proxies of overconfidence.¹⁵ In a later study with Anne Jones Dorn, they found that when Powerball and Mega Millions jackpots swell, small-trade participation in the stock market falls, and the effect appears in individual stocks and options but not in bonds and mutual funds.¹⁶ Xiaohui Gao and Tse-Chun Lin ran the same test on Taiwanese lottery jackpots and quantified it: when jackpots exceeded NT$500 million, trading volume fell between 5.2% and 9.1% in the stocks individual investors prefer, and between 6.8% and 8.6% in lottery-like stocks.¹⁷

In other words, some people have a fixed appetite for a particular kind of thrill and will satisfy it wherever it is cheapest to reach. The same five researchers behind the household transaction study returned to this in a 2026 review of retail betting markets, arguing that sports betting, prediction markets and retail options trading are converging and share the same behavioral drivers.¹⁸

That produces a more useful taxonomy than sorting activities by whether they happen in a brokerage account.

Two rows in that table describe activities with a positive expected return, and both of them require an edge you can demonstrate rather than assert. Which raises the obvious question.

What a real betting edge would look like

Nothing in the mathematics says every individual bettor must lose because bettors collectively lose. A sufficiently skilled bettor can identify mispriced probabilities and earn a positive expected return, the same way a skilled market maker profits from other traders.

Lisandro Kaunitz, Shenjun Zhong and Javier Kreiner documented exactly this. Rather than build a forecasting model to beat bookmakers at their own game, they treated the average odds across many books as a consensus probability estimate and bet only where a single book’s price implied a positive expected payoff against that consensus. The strategy returned 3.5% across 56,435 bets in a ten-year historical simulation on closing odds. Then they staked real money: 265 bets over five months at $50 flat stakes, returning 8.5% for a profit of $957.50.¹⁹

What happened next matters more than the return. Bookmakers began restricting their accounts. The authors documented maximum permitted stakes of $11.11 at one book, $10.45 at another, and $1.25 at a third, on a strategy that violated none of the books’ rules. A demonstrable edge existed, was published with its code and data, and was administratively switched off. The sportsbook is a retailer that can decline your business, which is a constraint with no close analogue for an ordinary long-only investor.

Meanwhile, evidence for skill among ordinary bettors is thin. In Levitt’s handicapping contest, the distribution of bettor outcomes was consistent with independent coin flips, past performance had no predictive value for future performance, and bettors in the top quartile of prior results were projected to win just 49.0% of subsequent bets.⁶

Kelly sizing becomes relevant only after an edge exists. The growth-optimal fraction is the edge divided by the odds, so a strategy with a negative expected value produces a Kelly fraction at or below zero, which is the formula telling you to bet nothing.²⁰ Practitioners generally use some fraction of Kelly anyway, because the formula is acutely sensitive to errors in the estimated probabilities and overbetting is punished far more severely than underbetting. We have a longer treatment of that in position sizing.

Why a winning season proves so little

Bloomberg’s article on the Betterment survey profiled a 32-year-old who researches his bets, caps his stake at $100, says he applies the logic of investing to betting, and is up roughly $2,500 on the year.² That result is compatible with having an edge. It is also compatible with not having one, and nothing in the number distinguishes the two without knowing how much was wagered, across how many bets, at what prices, and against a complete rather than a selectively remembered record.

The calculator below makes the point concrete. Set it to a bettor placing three $100 bets a week for a full year at −110 with no edge at all, a true 50% win rate, and the expected result is a loss of about $709. Yet roughly 29% of seasons like that still finish in the black, and one season in ten finishes about $818 or more ahead. A profitable year is an ordinary draw from a losing distribution.

The more punishing output is what it takes to tell the two cases apart. Move the assumed win rate to 55%, comfortably above the 52.38% break-even, and establishing that the edge is real rather than luck takes roughly 2,200 bets at a one-sided 5% test with 80% power. At three bets a week, that is about 14 years of complete records.

The 2026 tax rule changes the arithmetic

This deserves attention from anyone who thinks of betting as a small business, because it took effect this year and many bettors have not priced it in.

Section 70114 of the One Big Beautiful Bill Act amended the wagering-loss rule in the tax code. For tax years beginning after December 31, 2025, the deduction for gambling losses equals 90% of those losses, and remains capped at gambling winnings.²¹ The IRS reflects the change in the 2026 edition of Publication 505: “Beginning in 2026, your gambling loss deduction on Schedule A (Form 1040) will be limited to lesser of (1) 90% of your gambling losses or (2) your gambling winnings.”²²

Consider a bettor who itemizes and finishes a year with $100,000 of winning wagers and $100,000 of losing wagers. Economically the year is a wash. The deduction is limited to $90,000, so $10,000 of taxable income remains on a year with no gain. Two qualifications travel with that example. It assumes the bettor itemizes, and a filer taking the standard deduction gets no gambling loss deduction at all and is taxed on the full $100,000 of winnings. And the disallowed amount is 10% of losses, which coincides with 10% of winnings only in the exact break-even case.

A bill to restore the full deduction, the FAIR BET Act, was introduced by Representative Dina Titus in July 2025 and referred to the House Ways and Means Committee, where it has sat without a hearing or a floor vote. In February 2026 Titus filed a motion to discharge the committee, Petition No. 119-16, a procedure that needs signatures from a majority of the House to pull a bill onto the floor over a committee’s objection.²³ The 90% rule is in force for 2026, and anyone planning around a repeal is planning around a bill that has not moved. For a high-turnover bettor, grinding out a small edge now has to clear the book’s margin and a tax on losses that were never recovered.

Where a gambling budget goes

None of the above says that spending money on gambling is irrational. Suppose someone gets $300 of genuine enjoyment out of losing an expected $100 over a year. That is a good trade. Concert tickets, restaurant meals and ski passes all have a financial return of exactly negative one hundred percent, and nobody files them under portfolio construction.

There is a further argument, put well by Tyler Sawyer in the exchange that prompted this piece:

“If a person can satisfy that with a few hundred bucks of wasted money per year on actual gambling, and they have a sane financial plan that doesn’t involve gambling with their brokerage account…it seems like the EV of the system as a whole is better than satisfying the gambling desire with short-term options/trading using money earmarked for ‘retirement’.”

@tdlsawyer, August 2026.

This is the most interesting argument in the debate and also the one most likely to be oversold. If someone has a durable appetite for lottery-like payoffs, and the research above says many do, they may end up better off with a fixed annual betting budget plus a boring systematic retirement portfolio than with no explicit outlet and a brokerage account that periodically becomes an options casino. The substitution literature makes that plausible. It does not make it proven. No study has randomized people into a betting allowance and measured their lifetime wealth.

The evidence on the precommitment half is also weaker than most people assume. In the one randomized trial of the idea, prompting new customers at a Finnish operator to set a voluntary deposit limit raised the share who set one from 6.5% to as much as 45%, and left 90-day net losses statistically unchanged.²⁴ The same study found that customers who set limits without being prompted lost more than those who did not, which is what you would expect if limit-setting is a symptom rather than a treatment. A review of the wider literature reached the same place: voluntary limits have little empirical support, while mandatory ones show real benefit.²⁵ Treat the whole arrangement as a behavioral hypothesis about yourself, and check it against your own record rather than your intentions.

The population it would apply to is large. A February 2026 Siena College and St. Bonaventure survey found that 27% of Americans, and 52% of men aged 18 to 49, hold an active online sportsbook account.²⁶ The National Council on Problem Gambling’s 2024 survey found that 10% of men, roughly twice the rate among women, reported at least one indicator of problematic gambling many times in the preceding year.²⁷ That is a screening measure rather than a clinical diagnosis, and the overall rate fell from 11% in 2021 to 8% in 2024. For someone who repeatedly cannot hold a limit they set for themselves, a controlled allowance is the wrong tool and the helpline is the right one.

The policy that follows is deliberately plain.

1. Target allocation inside the portfolio: zero. Sports betting does not belong on an asset allocation chart next to equities, bonds, or anything else. It has a negative expected return and hedges nothing you are exposed to.
2. Fund it from entertainment spending, not investable assets. The right question is what you would happily spend on a hobby if you got nothing back, which has an answer. “What percentage of my portfolio?” does not.
3. Never fund it from retirement contributions, emergency liquidity, college savings, or debt. This is where the household research bites hardest, because crowding out saving is the specific harm it identifies.
4. Use a hard annual cap set in advance, with no top-ups. There is no evidence supporting a universal 0.5% or 1% or 2% of income figure, so do not adopt one. Size it the way you would size any other discretionary category.
5. If you think you are a profitable bettor, treat that as a hypothesis. Keep a complete prospective record, log the price you got, and ask whether the edge survives the margin, promotions, account limits, taxes, and the fact that your probability estimates are noisy.
6. If the cap keeps failing, the cap is the wrong tool. Difficulty respecting a limit you set yourself is information, and what it tells you is that a controlled allowance is not doing the job you assigned it.

For sizing the number itself, the machinery is the same one we use for any other discretionary category in how much to spend on a vacation: work out discretionary capacity first, then decide how you want to spend it. And if the underlying appetite is for volatility rather than for sport, the more relevant framework is in compensated versus uncompensated risk, which deals with risks you are paid to bear and risks you are not.

Key Takeaways

• The dividing line is the source of the expected return. Equities are a claim on business cash flows. A sportsbook wager is a transfer to a counterparty that set the price, and at −110 that expected transfer is 4.55% of every dollar risked.
• Do not argue it from a chart of past returns. Across 39 markets from 1921 to 1996, the US had the highest real capital appreciation in the sample at 4.32% against a 0.75% median, which is the definition of a survivorship problem.
• The published crowd-out estimate is about 20 cents on the dollar. The 2024 working paper reporting more than $2 per dollar is superseded, and much of the coverage still circulating uses it.
• A winning season is weak evidence. For a bettor with no edge placing three $100 bets a week at −110, roughly 29% of years still finish in the black, and confirming a 55% win rate would take on the order of 2,200 bets.
• Being uncorrelated is not a reason to own something. A negative-expected-return position earns a place only if it pays off when you need it most, which is what insurance does and a parlay does not.
• Gambling can be rational spending. The portfolio weight is zero, and the budget line is a legitimate question answered out of entertainment money with a cap set in advance.

Frequently Asked Questions

Is investing just gambling with extra steps?

No, though both involve real risk and uncertain outcomes. A diversified equity portfolio has a positive expected return because it represents ownership of businesses that generate cash flows, and investors are compensated for bearing the risk attached to them. A sportsbook wager has a negative expected return because an intermediary prices in a margin before you place it. That difference does not guarantee any particular investor a good outcome, and individual stocks, single countries and rich valuations can all disappoint badly.

What percentage of my portfolio should I allocate to sports betting?

Zero. The framing itself is the problem, because an allocation is a claim that something belongs among your assets. If you enjoy betting, fund it from entertainment spending with a fixed annual cap, the same way you would fund any other hobby with a negative financial return.

Can anyone actually make money betting on sports?

Yes, and it is rare and looks nothing like fandom. Documented profitable strategies exploit pricing inefficiencies across books rather than superior knowledge of the sport, and the researchers who published one had their accounts limited to maximum stakes as low as $1.25 once they started winning. Treat “I know a lot about football” as roughly zero evidence of an edge.

I made money betting last year. Does that mean I have an edge?

Not on its own. Variance over a single season is large enough that a bettor with a negative expected return finishes ahead a substantial fraction of the time. Distinguishing a genuine edge from luck at typical betting volumes takes thousands of wagers and a complete record, not a good year.

How did the 2026 tax change affect gambling losses?

For tax years beginning after December 31, 2025, the itemized deduction for gambling losses is limited to 90% of those losses and still cannot exceed gambling winnings. A bettor who breaks even economically can therefore owe tax on 10% of the losses, and a filer taking the standard deduction gets no deduction at all. Rules differ for professional gamblers and by state, so this is general information rather than tax advice.

Does having a small betting budget protect my retirement account?

It is a reasonable hypothesis and it is not established. Research does show that people substitute between gambling and speculative trading, so a contained outlet plausibly keeps some of that impulse away from a retirement portfolio. No study has shown the arrangement improves long-run wealth, and precommitted limits work far better for people who already respect them. If your cap keeps moving, it is not working.

Related Guides

• A 1,000% Upside and a 100% Downside Does Not Make It a Good Bet works through the expected-value arithmetic behind speculative payoffs in a brokerage account.
• Compensated vs. Uncompensated Risk separates risks that carry an expected return from risks that simply add variance.
• Position Sizing covers Kelly, fractional Kelly, and why the size of a speculative bet matters more than the pick.
• Most Stocks Lose to T-Bills explains why the lottery-like distribution of individual stock returns is an argument for diversification.
• How Much Should You Spend on a Vacation? is the method for sizing any discretionary category out of real capacity.

Sources and method

1. Betterment, 2026 Retail Investor Survey, August 12, 2026. Online survey fielded March 27 to April 3, 2026, among 1,000 US retail investors split evenly across four generations, panel provided by Sago. Respondents held at least one qualifying investment; 401(k)-only holders were excluded. This is an opt-in online panel rather than a probability sample, and the generational figures each rest on roughly 250 responses.
2. Song, Z., & Amponsah, M., Gen Z Is Moving Money From Stocks to Sports Betting in Wealth Plans, Bloomberg, August 12, 2026.
3. Jorion, P., & Goetzmann, W. N. (1999). Global Stock Markets in the Twentieth Century. The Journal of Finance 54(3), 953–980. The 4.32% and 0.75% figures are real capital appreciation returns and exclude dividends, so they understate total returns and should not be read as an estimate of what US equities returned in full.
4. American Gaming Association, Commercial Gaming Revenue Hits $78.7 Billion in 2025, February 26, 2026. The AGA publishes revenue and handle; the ~10.2% ratio is our own arithmetic on those two figures and is not an AGA statistic.
5. New Jersey Division of Gaming Enforcement, DGE Announces December 2025 Gaming Revenue Results, January 16, 2026, completed-events win and handle by category. The 19.2% parlay hold is published directly; the ~4.9% figure for non-parlay wagers is our subtraction from the same table. DGE notes these figures are unaudited and subject to later amendment.
6. Levitt, S. D. (2004). Why are Gambling Markets Organised so Differently from Financial Markets? The Economic Journal 114(495), 223–246. Levitt reports gross profit rates per 100 units of potential winnings; the corresponding shares of dollars at risk are 5.6% and 4.55%. The 52.4% break-even figure is his footnote 14.
7. Barber, B. M., & Odean, T. (2000). Trading Is Hazardous to Your Wealth: The Common Stock Investment Performance of Individual Investors. The Journal of Finance 55(2), 773–806. Returns quoted are annualized geometric means; the market benchmark is the Fama-French value-weighted NYSE/AMEX/Nasdaq index.
8. Baker, S. R., Balthrop, J., Johnson, M. J., Kotter, J., & Pisciotta, K. (2026). Gambling away stability: Sports betting’s impact on vulnerable households. Journal of Financial Economics 183, article 104330.
9. Baker, S. R., Balthrop, J., Johnson, M. J., Kotter, J., & Pisciotta, K. (2024). Gambling Away Stability, NBER Working Paper 33108. The earlier working-paper version, whose larger crowd-out estimate circulated widely in 2024 and 2025 coverage.
10. Hollenbeck, B., Larsen, P., & Proserpio, D. (2026). The Financial Consequences of Legalized Sports Gambling. Management Science, published online July 21, 2026. No volume or issue assigned at the time of writing.
11. Goss, J., & Mangrum, D. (2026). Sports Betting Across Borders: Spatial Spillovers, Credit Distress, and Fiscal Externalities, Federal Reserve Bank of New York Staff Reports no. 1184, March 2026.
12. Goss, J., & Mangrum, D., Sports Betting Is Everywhere, Especially on Credit Reports, Liberty Street Economics, March 25, 2026.
13. MagnifyMoney, 55% of Americans Think Investing Is As Risky As Gambling. Online survey of 1,082 Americans conducted by Qualtrics, July 22 to 26, 2019. “Gamblers” are respondents who in the prior 12 months visited a casino, played an online gambling game, played poker for money, or bet on sports.
14. Kumar, A. (2009). Who Gambles in the Stock Market? The Journal of Finance 64(4), 1889–1933. Lottery-type stocks are identified by a joint percentile sort rather than any absolute price threshold.
15. Dorn, D., & Sengmueller, P. (2009). Trading as Entertainment? Management Science 55(4), 591–603.
16. Dorn, A. J., Dorn, D., & Sengmueller, P. (2015). Trading as Gambling. Management Science 61(10), 2376–2393. Described qualitatively here because the published abstract reports direction and significance without a headline effect size.
17. Gao, X., & Lin, T.-C. (2015). Do Individual Investors Treat Trading as a Fun and Exciting Gambling Activity? Evidence from Repeated Natural Experiments. The Review of Financial Studies 28(7), 2128–2166.
18. Baker, S. R., Balthrop, J., Johnson, M. J., Kotter, J. D., & Pisciotta, K. (2026). Retail Betting Markets, NBER Working Paper 35520.
19. Kaunitz, L., Zhong, S., & Kreiner, J. (2017). Beating the bookies with their own numbers, and how the online sports betting market is rigged, arXiv:1710.02824. A preprint rather than a peer-reviewed paper. The ten-year 3.5% figure is a historical simulation on closing odds; the 8.5% figure is the realized return on 265 real-money bets.
20. MacLean, L. C., Thorp, E. O., & Ziemba, W. T. (2010). Good and bad properties of the Kelly criterion; and Thorp, E. O. (2006), The Kelly Criterion in Blackjack, Sports Betting, and the Stock Market, in Handbook of Asset and Liability Management, Vol. 1, 385–428.
21. 26 U.S.C. § 165(d), as amended by the One Big Beautiful Bill Act, Pub. L. No. 119-21, § 70114, effective for taxable years beginning after December 31, 2025.
22. Internal Revenue Service, Publication 505 (2026), Tax Withholding and Estimated Tax, “What’s New.” Note that IRS Topic No. 419 had not been updated to reflect the change at the time of writing.
23. H.R. 4304, Fair Accounting for Income Realized from Betting Earnings Taxation Act, 119th Congress. Sponsor Rep. Dina Titus (D-NV-1), introduced July 7, 2025; referred to House Ways and Means. Latest action February 12, 2026: motion to discharge committee filed by Ms. Titus, Petition No. 119-16. Status current as of August 14, 2026; check the bill page for any later action.
24. Ivanova, E., Magnusson, K., & Carlbring, P. (2019). Deposit Limit Prompt in Online Gambling for Reducing Gambling Intensity: A Randomized Controlled Trial. Frontiers in Psychology 10:639. N = 4,328 customers of a Finnish operator tracked for 90 days. The pooled intervention group did not differ from control on the proportion with a positive net loss (OR = 1.0, p = 0.921) or on the size of net loss (B = −0.1, p = 0.291).
25. Delfabbro, P., & King, D. L. (2021). The value of voluntary vs. mandatory responsible gambling limit-setting systems: a review of the evidence. International Gambling Studies 21(2), 255–271.
26. Siena College Research Institute and St. Bonaventure University, American Sport Fanship Survey 2026, Release 2: Sports Betting, April 13, 2026. 3,084 responses collected February 16 to 27, 2026 from an online panel. A non-probability opt-in sample reported with a credibility interval rather than a margin of error.
27. National Council on Problem Gambling, National Survey on Gambling Attitudes and Gambling Experiences (NGAGE 3.0), Key Findings, 2024. Online survey of 3,013 US adults conducted by Ipsos between January 26 and March 20, 2024. The measure is self-reported endorsement of at least one of four problematic gambling behaviors “many times” in the past year, which is a risk screen and sits far above clinical prevalence estimates for gambling disorder.

Method. Break-even win rates use the standard conversion from American odds, so a −110 price requires 110/210 = 52.38%. Season outcome distributions in the calculator are exact binomial distributions over the number of winning bets, treating each wager as an independent flat-stake straight bet at a single price, which ignores parlays, correlated wagers, promotional credits, line shopping, and state taxes. The sample-size figure for establishing an edge uses a one-sided test at the 5% level with 80% power. The 2026 tax illustration assumes an itemizing filer and covers federal tax only.

Author disclosure

Educational content, not investment, tax, or legal advice. I write both Summitward and the Engineer Investor account on X, and this piece grew out of a public exchange there. Summitward has no relationship with any sportsbook, betting operator, or brokerage named or implied above. Cited studies cover specific markets, samples, and periods, and their findings do not generalize to every bettor or investor, which is why the disagreement between them is reported rather than resolved. Figures produced by the calculator describe the inputs shown and are not general results. If gambling has stopped being entertainment, the National Council on Problem Gambling operates a confidential helpline at 1-800-522-4700. Facts verified against the cited journals, working papers, regulator publications, and the US Code as of August 2026.