How casino cashback actually works — the house edge maths

That last sentence is the one that trips people up, so it's worth going slowly.

The word "cashback" does a lot of damage here. It implies you lose £100 and get £5 of it back. Almost no casino programme works that way, and the ones that do usually cap it so tightly that it isn't worth pursuing.

What actually happens is closer to a rebate on your turnover. You wager. The casino records the total. It multiplies that total by the margin it holds on the games you played. That figure — your theoretical loss — is the pool everything else is calculated from.

Two players can walk away from the same session with wildly different outcomes and identical cashback. One is up 400 USDT, one is down 900, both wagered 12,000 on the same slot. Both earn exactly the same. The casino doesn't care who won; it cares how much volume went through.

This is better for you than a loss rebate, and it isn't close. A loss rebate pays nothing on your winning sessions. Turnover-based cashback pays on every session you ever play.

→For the underlying number, see What "theoretical loss" means.

There are three parties and two splits.

The casino holds a mathematical edge on every game. Over enough volume, that edge is revenue — reliable, forecastable revenue. The casino spends a chunk of it on acquiring players.

The affiliate — WagerSave, or any other — is one of those acquisition channels. When you sign up through a tracked link, the casino pays the affiliate a share of the revenue you generate. Typically 30–50% of net gaming revenue, sometimes higher for volume partners.

You get a portion of the affiliate's share paid back. That's the cashback. The affiliate keeps the rest as margin.

So the chain is: your turnover → house edge → casino revenue → affiliate commission → your cashback. Every link is a percentage of the one before it, which is why the final number looks small next to what you wagered even when the headline rate looks large.

Nothing here is subsidised. There's no promotional budget being burned. The money is a redistribution of margin that already exists, which is the reason these programmes are stable in a way that deposit bonuses aren't.

→Full breakdown: How affiliate commission funds cashback.

For any wager-based programme:

Theoretical loss = Total wagered × House edge
Cashback         = Theoretical loss × Cashback rate

Or collapsed into one line:

Cashback = Total wagered × House edge × Cashback rate

Three inputs. You control one of them completely (how much you wager), one partially (which games, and therefore which house edge), and none of the third.

Worked example

You wager 10,000 USDT on European roulette. House edge is 2.70%.

Theoretical loss = 10,000 × 0.027   = 270 USDT
Cashback at 40%  = 270 × 0.40       = 108 USDT

108 USDT back on 10,000 wagered. As a percentage of turnover, that's 1.08%.

Now the same 10,000 USDT on a slot with a 4% house edge (96% RTP):

Theoretical loss = 10,000 × 0.04    = 400 USDT
Cashback at 40%  = 400 × 0.40       = 160 USDT

Same money wagered. Same headline rate. 52 USDT difference, because the second game has a fatter margin to share from.

This is counterintuitive and it's worth sitting with: a worse game for you is a better game for your cashback. Not better overall — you're expected to lose 400 instead of 270, and getting 160 back instead of 108 doesn't close that gap. But the cashback line item genuinely goes up.

Every game contributes at its own edge. Here's what 10,000 USDT of turnover produces at a flat 40% rate:

\ Slots are frequently weighted below 100% in cashback calculations even though their edge is high — see the weighting section below.*

The right-hand column is the number that actually matters and the one nobody advertises.

→Related: House edge, explained with real numbers and Volatility vs house edge.

This is where comparing offers falls apart, because operators quote in two incompatible units and rarely say which.

Percentage of theoretical loss. "40% cashback" means 40% of the house's expected win from you. Rakeback-style. Common on crypto casinos and CS2 case sites. Rates run high — 20% to 40%+ — because the base they're applied to is small.

Percentage of turnover. "0.5% cashback" means half a percent of everything you wagered, full stop. Common on mystery box platforms and some sportsbooks. Rates look tiny because the base is enormous.

To compare them, put both in the same unit:

% of turnover = % of theoretical loss × house edge

A 40% rate on a 5% edge game returns 2% of turnover. A flat 0.5% of turnover offer beats a 10% rate on that same game. The headline numbers point the wrong way.

Always convert to return per 1,000 wagered. It's the only figure that survives contact with a different operator's marketing.

Offer A wins. Offer D has a respectable-sounding rate and returns a sixth of what A does, because low-edge games have almost nothing to share.

→Every WagerSave partner converted to this unit: Best crypto casino cashback rates — compared.

Most programmes apply a weighting multiplier per game category before the cashback rate. It reads like fine print. It isn't.

Typical structure:

Applied to the formula:

Cashback = Wagered × House edge × Weighting × Rate

Blackjack at a 0.5% edge, weighted at 10%, on a 40% rate, over 10,000 wagered:

10,000 × 0.005 × 0.10 × 0.40 = 2 USDT

Two. On ten thousand wagered.

Low-edge table games are already poor cashback earners; weighting is what takes them to effectively zero. If you're a blackjack player, cashback is not a meaningful part of your economics and you shouldn't restructure anything around it. If you play slots, it's the single largest rebate available to you.

Check the weighting table before the rate. A 20% rate with no weighting beats a 40% rate that weights your games at 25%.

Rates vary by vertical because the underlying commission structures differ.

Above the top of any range, be suspicious. Cashback can't exceed what the affiliate is paid, and the affiliate isn't paid the full house edge. An operator advertising a rate that implies a negative margin is either quoting in a different unit, capping it somewhere you haven't read yet, or not planning to pay.

→Quarterly medians by vertical: The Crypto Cashback Index — Q3 2026.

It does not make you a winner. That needs saying plainly.

Take the 96% RTP slot. You wager 10,000, you're expected to lose 400. Cashback at 40% returns 160. You're still expected to be down 240. The effective RTP moved from 96% to 97.6% — a real improvement, and still under 100%.

Cashback narrows the gap. Nothing available on any of these sites closes it. Anyone telling you a rebate makes gambling profitable is either selling something or hasn't done the arithmetic.

What it is genuinely useful for: if you were going to play anyway, playing through a tracked link costs you nothing and returns a percentage of margin that would otherwise stay with the casino. That's the entire proposition. It's a reasonable one. It isn't an edge.