Losses disguised as wins: how often does a slot celebrate a spin that lost money?

What is a loss disguised as a win?

Modern slots bet on many lines or ways at once. Say you stake 20 cents across 20 lines and one line pays 15 cents. The reels light up, the counter rolls and the game plays its win sound. You are 5 cents down. Researchers call this a loss disguised as a win, or LDW.

In a 2010 study, Mike Dixon and colleagues at the University of Waterloo measured 40 novice players' skin conductance and heart rate. The response to these disguised losses was about as large as the response to real wins, and larger than the response to plain losses. Dixon et al. (2010), Addiction →

A 2013 follow-up by the same group found that the more of these outcomes people saw in 200 spins, the more often they believed they had won. Asked to think out loud while playing, most described them as wins. Jensen et al. (2013), International Gambling Studies →

Great Britain acted on it. Since 31 October 2021, online slots licensed by the Gambling Commission must not celebrate a return that is less than or equal to the total stake, with sounds or visuals linked to winning. Gambling Commission: the rule and its reasoning →

What nobody publishes is how often it happens. Providers don't report it, and you can't tell from playing. We can count it, because our lab records the result of every spin it plays.

How common are they?

On average, 29 in every 100 spins paid something, and 16 of those 29 paid less than the stake. Put another way: 53% of all paying spins were losses. Game to game this varies a lot: the typical (median) game sits at 57%, from 0% on Lucky Lucky to 80% on Fortune Tree Wild. Every figure comes from 65 slots and about 44 million spins.

Almost all of them are ordinary spins: 99.9% of these losing wins came on spins that did not trigger free spins.

Take Gates of Olympus: 29 in 100 spins pay something, but 16 of those pay less than the bet and only 12 come out ahead. 55% of its "wins" are losses.

Do games that pay more often disguise more losses?

Yes, clearly. Games that pay on more spins make up much of that frequency with payouts smaller than the stake. Across games, the correlation between how often a game pays and the share of those payouts that lose money is 0.54.

This matters because how often a game pays is the thing you feel. A game that lights up every third spin feels generous. Much of that is the disguise.

What a real session looks like

Averages hide what a session feels like, so we cut each game's recorded spins, in the order they were played, into stretches of 500 spins: about the length of an evening. No simulation; these are 88,682 real stretches.

In the typical stretch of 500 spins, 148 spins paid something. 83 of those were losses and only 56 were profits. So there were more celebrated losses than real wins, and that was true in 97% of all stretches on the typical game. The typical stretch returned 90% of what was staked, and 64% of stretches ended below 100%.

Even free spins can lose

Free spins are the biggest celebration a slot has: the transition, the music, the counter. Of 199,978 free-spin rounds in our records, 11% paid back less than 10x the stake, including the spin that triggered them. 1% paid back less than the stake itself.

That doesn't make free spins worthless. For most games they carry a large share of the total payback. But the feature arriving and the feature paying are two different events.

How to read a game with this in mind

• Watch the balance, not the celebration. The only number that says whether a spin won is the balance after it, compared with the balance before.
• A "win" counter is not a profit counter. If a game shows how much you won this session, check whether it subtracts what you staked. Great Britain requires a net position on screen for this reason.
• How often a game pays says little about how much it returns. Two games at the same RTP can differ hugely in how often they pay; the difference is often made up of payouts below the stake.
• Count your own session in money, not wins. Set the amount and time before you start and stop there, whatever the screen celebrated.

How slots work → · Do slot patterns work? →

What this study does not show

• We measured payouts, not animations. Whether a particular game plays a win sound for a payout below the stake depends on its interface and on where it is licensed; games built for Great Britain should not.
• All spins are from the providers' official free demos. FA CHAI and PG Soft demos paid back far more than their official RTP in our lab, so their games are left out of the headline figures (their share of paying spins below the stake was 46%).
• Games our lab flags as a suspicious measurement are also left out: Gems of Zeus, Joker King.
• Every spin was played at one fixed stake. Wins are counted as multiples of the stake, so the share does not change with the stake size.
• A spin and its free spins count as one result. A spin that triggers free spins and pays less than the stake in total is counted as a disguised loss.

Demo vs real money → · Our corrections log →

Data, download and how to cite

The per-game figures behind this page, 65 slots plus the games left out and why, are free to download and reuse under CC BY 4.0. Please link to this page when you use them.

Download the per-game data (CSV) →

OddsLab (2026). Losses disguised as wins: counted across millions of slot demo spins, 2026-10-04 to 2026-10-27. https://oddslab.pro/guides/losses-disguised-as-wins/

Every day we publish a SHA-256 fingerprint of that day's raw spin records, so they cannot be quietly changed afterwards. Researchers or journalists who want the raw records can ask, and we match them against the fingerprint. Verify the data →