Should You Bet Every Underdog in March Madness?

Blindly betting every March Madness underdog does not guarantee positive returns. Profit depends on the prices you accept, how often those bets win and exactly which games your strategy includes—not simply on the number of upsets.

A published hypothetical 2024 men's main-bracket ledger finished slightly ahead, while a separate publisher's longer sample reported a loss. Neither establishes a reliable future advantage. The useful question is whether a clearly defined strategy earns enough from its winners to cover every losing stake.

What counts as an underdog bet?

An underdog is the team with the longer decimal price within a two-team market at your chosen bookmaker and observation time. A tournament seed is a ranking, not a betting price: a lower-seeded team can still be the betting favorite.

Do not classify underdogs solely by a plus sign in American odds. Although underdogs commonly have positive American prices, both sides can have negative prices. Compare the pair, converting prices consistently if needed with the odds converter.

The market also changes what counts as a win:

• Moneyline: the team must win outright.
• Point spread: the team must cover the spread. A team that loses by four points with a +5.5 handicap wins its spread bet but loses its moneyline bet.
• Push: if the final margin exactly matches a whole-number spread, the stake is returned under the applicable settlement rules.

Check your bookmaker's actual rules for overtime, periods, cancellations and settlement. A seed-based upset is not necessarily a bookmaker-defined upset, and moneyline results cannot be substituted for ATS results.

For matchup analysis rather than a blind selection rule, see the March Madness betting strategy guide. Bracket-pool strategy is separate because pool scoring does not pay the odds on individual wagers.

Why prices matter more than win frequency alone

The same win frequency can produce either a profit or a loss because higher-priced winners return more money. For a winning wager at decimal odds D, gross return is stake × D, while net profit is stake × (D − 1).

Consider this independent hypothetical comparison: ten bets, each risking $10, with three winners and seven losers. Every bet has the same price within each scenario. There are no pushes, voids, fees or taxes, and the figures below require no rounding.

This chart represents only the hypothetical ten-bet model—not historical tournament results.

For a ledger with different prices, calculate profit = sum of winning net profits − sum of lost stakes. Calculate return on total staked as ROI = profit / sum of accepted stakes × 100. In this guide, the denominator includes accepted stakes later returned through pushes or voids if those wagers are logged; document that convention when comparing results.

For decided bets without pushes, the break-even win probability at a single price is 1 / D. With mixed prices, a single overall win percentage is insufficient to establish profitability.

What does −110 require?

At −110, decimal odds are approximately 1.90909 and the break-even rate is 52.38% of decided bets—not 50%. A hypothetical 51.8% win rate at that price produces approximately −1.11% expected ROI.

With 100 wagers each risking $10 at −110:

These figures are calculated before per-ticket cent rounding, with no pushes, fees or taxes. Actual settlement rounding can change the total slightly.

Keep staking conventions consistent. Risking $10 at −110 is not the same as risking $11 to win $10. Switching between those conventions makes comparisons misleading.

What does the published historical evidence show?

The evidence shows sample-dependent outcomes, not guaranteed profitability. Our arithmetic audit of Adam Hensley's NCSharp ledger found that its hypothetical 2024 men's main-bracket underdog bets returned +$20.20: 63 wagers at $10, 21 wins, 42 losses, $630 staked, $650.20 gross return and 3.21% ROI.

The article was published March 21, 2024, and updated February 12, 2025. We recomputed all listed American prices, including Colorado −102. These were published hypothetical FanDuel NC bets—not our wagers or independently verified historical prices. First Four was excluded.

Oakland's +720 winner contributed $72 net profit and $82 gross return. Excluding that entire wager retrospectively leaves 62 bets, $620 staked and −$51.80 net profit: −8.35% ROI. This diagnoses concentration; it is not an advance selection strategy.

Does a longer sample support blind betting?

A separate longer sample reported a loss. Chase Kiddy's BetMGM article, published March 6, 2026, reports men's moneyline underdogs going 122–279 across 2019–2025, with 2020 canceled: 30.4% outright wins and −4.9% ROI.

Those are the publisher's figures, not an independently verified backtest; no raw odds export was available. The 401 reported decided games should not be treated as an automatically complete set of every possible game.

Different prices, filters and sample periods prevent combining this result with the 2024 ledger as if they were one continuous dataset. Removing unfavorable years after seeing their results introduces selection bias. Neither sample establishes a universally profitable seed, round or forecast for a future tournament.

How much would betting every underdog involve?

Betting every eligible men's main-bracket underdog at $10 means $630 in turnover if all 63 games qualify. Betting only the main bracket's first round means 32 wagers and $320 in turnover—not 64 games.

According to the NCAA's expansion announcement, the men's and women's fields expand from 68 to 76 teams beginning in 2027. The expanded Opening Round still leaves a 64-team main bracket.

The whole-field counts follow the single-elimination rule games = teams − 1. Actual eligible wager counts may be lower because of missing prices, pick'em markets or exclusions under your void policy.

Turnover is not the same as required starting bankroll. Settled bets can release money for later wagers, but simultaneous bets create concurrent exposure. Losses, settlement delays and bookmaker limits can prevent that money from being recycled. These figures illustrate volume, not a recommended stake or deposit.

How can you check an assumed edge with the calculator?

Use the Value Bet Calculator to evaluate a price against your entered probability assumption. It does not discover the true probability, validate an NCAA forecast or supply live bookmaker prices.

For a reproducible manual example:

1. Select Decimal as the odds format.
2. Select Your probability.
3. Enter 3.50 in Your price (the odds you can bet).
4. Enter 30% in Your true win probability. Despite the label, this is your estimate—not a verified truth.
5. Set Stake to 10 and Bankroll to 1000.

English calculator inputs showing decimal odds 3.50, an assumed 30% win probability, stake 10 and bankroll 1000

Localhost capture of manual example inputs. The values are assumptions, not production prices or personal bets; the dollar setting involves no currency conversion.

What does the positive result mean?

The assumed 30% win probability at decimal odds 3.50 produces +5% EV, or +$0.50 expected value on a $10 stake. That conclusion holds only if the probability estimate is accurate.

The price implies a 28.6% win probability, rounded for display. The probability gap is +1.43 percentage points—not 1.43% ROI. Fair decimal odds under the entered assumption are 3.33.

English calculator result showing positive expected value for the manually entered 30% probability at odds 3.50

Localhost result from the manual assumptions above. This is a calculator demonstration, not a production-market assessment or a record of betting returns.

The calculation is EV = stake × (p × D − 1), where p is a decimal fraction. Here, 10 × (0.30 × 3.50 − 1) = 0.50. This assumes a binary win/loss wager with no fees, taxes or pushes; displayed probabilities and fair odds are rounded.

A winning ticket would return $35 gross, including $25 net profit. EV is an average under the probability assumption, not the outcome promised for that ticket.

The optional Kelly display gives full Kelly of 2%, or $20 of a $1,000 bankroll; half Kelly is $10 and quarter Kelly is $5. Those outputs are not automatic staking advice. They inherit the uncertainty in your probability estimate.

What happens if the probability estimate is lower?

Changing only Your true win probability to 25% turns the same price into a negative-EV bet. Keep odds at 3.50, stake at 10 and bankroll at 1000 to isolate the effect of that assumption.

The result is −12.5% EV, −$1.25 EV on the stake and a −3.57-percentage-point probability gap.

English calculator sensitivity result showing negative expected value after reducing the assumed win probability to 25%

Localhost sensitivity capture with manually entered 25% probability. It illustrates assumption risk, not a real NCAA prediction or personal wager.

A five-percentage-point change in the estimate reverses the conclusion without any change in price. For more on judging that input, see the value betting strategy guide.

How should you test the strategy reproducibly?

A meaningful test fixes its rules before results are known, records every eligible wager and preserves a future season as an untouched out-of-sample check. Choosing the winning year, round or team afterward tests hindsight rather than a usable strategy.

Write down these rules before collecting results:

Save an individual row for every eligible selection: game, market, timestamp, both quoted prices, selected team, accepted odds, stake, result, settlement status, gross payout and net profit. Include every eligible loser, not just notable upsets. If testing actual execution, distinguish a quoted price from the price the bookmaker accepted.

Prices can move between observation and acceptance, changing both returns and which team qualifies as the underdog. Settlement timing affects available cash and concurrent exposure, even when final ROI is unchanged.

Report the full predefined sample first. Round-by-round results and removal tests can reveal concentration, but do not convert them into profitable selection rules without a new untouched test.

What should determine the final decision?

The decision should rest on evidence of an advantage at executable prices, not the appeal of backing outsiders. A small positive historical total or a favorable calculator output is insufficient on its own.

Before committing money, check:

• Are the market, bookmaker, observation time and eligible games defined?
• Are calculations based on accepted prices and complete settlements?
• Does the probability estimate have support beyond a team's seed or an upset narrative?
• Could one long-priced winner account for most of the profit?
• Has the strategy been checked on an untouched future sample?
• Can the available bankroll withstand losses and overlapping exposure?

Blind underdog betting is easy to describe but not inherently profitable. Without defensible probability estimates or reproducible out-of-sample evidence, treat it as an unproven wagering rule—not a dependable source of returns.