Sports betting ROI: how to calculate it and check the record

Sports betting ROI tells you how much you won or lost compared with the amount you risked: net profit ÷ total staked × 100. If you staked 100 units across your settled bets and lost 8 units, your ROI is -8%. The number of winning picks cannot tell you that on its own.

That is the first thing I check when someone shares a record. What were the prices? How much was risked? Are the losing picks still there? A 60% win rate can look great and still leave you down money.

Here is how to calculate the number, check a record you are following, and keep your own research honest. Every worked example below is fictional. None is a WagerProof result or a suggested bet.

How to calculate sports betting ROI

Start with one clearly defined sample, such as all cash bets settled during a calendar month. Add the actual stakes. Then add the net profits and losses from those same bets.

• Net profit is what you received back, including returned stakes, minus what you staked and any separately charged fees.
• Total staked is the amount risked across the included bets, not your opening account balance.
• ROI is net profit divided by that total, multiplied by 100.

This is the stake-based calculation used in Sports Insights' ROI explanation. You may also see it called yield. Check the denominator before comparing two published records.

Suppose you start with 20 units, reuse some returned money, and place 100 units of bets over the month. A 5-unit profit is 5% of the amount staked, not 25%. The 25% figure measures profit against the starting balance. Both calculations have a denominator, but they answer different questions.

A unit is just a consistent reporting amount. It does not make a stake safe. Keep its value fixed within a units-based comparison, and retain actual currency amounts if that value changes.

Why a 60% win rate can lose money

Imagine five separate bets, each risking one unit at -200. Three win and two lose. There are no fees, bonuses, pushes or pending bets.

At -200, a one-unit stake earns 0.5 units of profit on a win. It does not earn a full unit.

The win rate is 3 ÷ 5 = 60%. The ROI is -0.5 ÷ 5 × 100 = -10%.

Nothing is wrong with the arithmetic. The winners simply did not pay enough to cover the losers. For equal stakes at -200, you need to win two thirds of resolved bets just to break even before fees. Different prices change that threshold. Our implied probability guide explains the price conversion.

Keep the price attached to the result. A row that only says “win” is missing half the story.

Add stakes before calculating the percentage

Averaging individual bet returns can give the wrong answer when stakes differ.

In this second fictional example, a one-unit bet at +200 wins two units of profit. A separate four-unit bet loses all four units. No fees or promotions apply.

The combined ROI is -2 ÷ 5 × 100 = -40%. Averaging +200% and -100% would give +50%, even though the record lost money.

The same problem appears when combining weeks, sports or Agents with different amounts staked. Add the relevant profits and stakes first. Then divide. Do not average the percentages.

Decide what belongs in the record

Write your grading policy before reviewing the results. Otherwise it is too easy to change the sample until the percentage looks better.

For the manual review sheet below, use this policy:

These are this sheet's reporting choices, not universal sportsbook rules. Some trackers exclude pushes from the amount staked. State your convention and compare like with like.

For example, add a one-unit push to the two-bet record above. Net profit stays at -2 units, but included stake rises to 6 units. With the push included, ROI becomes -33.33%. Excluding it produces -40%. The underlying money result has not improved.

A parlay is one ticket for profit and stake accounting. You can review its legs separately, but do not count the full ticket stake again for every leg. If you also placed singles, those are separate wagers with their own stakes.

For a fictional full cash-out, suppose 50 units were staked and 40 units came back, with no separate fee. Net profit is 40 - 50 = -10 units, and return on that stake is -20%. If your stated policy includes cash-outs, include those cash flows in ROI but label the ticket separately from ordinary wins and losses. This is an accounting example, not advice to cash out.

Use the record behind the Agent

We built WagerProof's Agents to make research easier to follow from the original reasoning through the result. You choose how an Agent approaches the slate. Its picks are graded automatically, with wins, losses, pushes and net units visible in its history. Open the Agent's detail page to review that record, including the losing picks. Agent performance tracking

I would start there before following an Agent because of one impressive week. Read several losing selections as well as the winners. Check whether the markets and prices match the kind of research you want to do.

Historical Systems give you another way to follow an idea over time: save a supported historical filter as a System and review its ongoing graded results. Keep the historical sample separate from what happened after you saved the idea. WagerProof

An Agent's record is not your personal sportsbook statement. If you took a different price, changed the stake or skipped a pick, your result can differ. This guide's ROI formula is a manual review method, not a claim that every WagerProof screen uses this exact denominator.

Copy the weekly review sheet

Open the sports betting ROI review sheet. Copy it into your notes and fill it in for one fixed period. It is a plain-text worksheet, not a calculator or an automatic export.

For each selection, save the event, exact market and line, price, sportsbook, timestamp, stake, result and net profit or loss. Keep the research cutoff and original reason in the same record. If a model or Agent supplied the idea, record its name and the settings you used.

Do that before the event. Do not rewrite the reason after you know the score.

At the end of the period:

1. Reconcile each settled result with the ticket or published pick you are evaluating. Do not mix your own tickets with an Agent's hypothetical stakes.
2. Count wins, losses and pushes. List pending bets, voids and missing records separately.
3. Add included stake and net profit. Calculate ROI from those matching totals.
4. Compare the result with the previous period using the same rules.
5. Note any changed settings, missing data or price differences that prevent a fair comparison.

If the included stake is zero, ROI is unavailable. It is not 0%.

Missing tickets are a different problem. If you cannot reconcile the full period, label the record incomplete and show exactly which tickets and amounts the calculation includes. A percentage for the rows you found is not a complete account of the month.

What a positive ROI does and does not prove

A positive ROI means the defined sample made money under the stated accounting rules. It does not establish what will happen next or prove that the research has a lasting advantage.

There is no universal number of picks that makes a record trustworthy. Price range, stake sizes, overlapping games and repeated exposure to the same team all affect what the sample can tell you. Ten related picks are not ten separate tests of an idea.

Keep closing line value beside results when you have a comparable closing price. It measures a different part of the process. If you save probability estimates, review how those estimates held up over time too. Neither replaces the cash result.

When you change an Agent's settings or a System's filters, date the change in your notes. Preserve the earlier record. A new approach should not silently inherit only the attractive parts of the old one.

The point of tracking is to see what happened clearly, including when the answer is disappointing. Do not raise stakes to repair a percentage. Set time and money limits before researching, and step away if reviewing the record starts turning into chasing losses. Our responsible research guide has a practical starting point.

About the author

Chris Habib

Chris builds WagerProof and writes practical guides about sports research, product methodology, and responsible use of betting data. He focuses on inspectable workflows rather than outcome promises.

Expertise: sports analytics product design, probability and odds interpretation, research workflows.

Have a correction or a newer first-party source? Email the editorial team.