By Zavian Thornell, iGaming mathematician and expected-value researcher
I spent three weeks running a parametric expected-value model across 800 casino bonus configurations. My conclusion is uncomfortable for the industry and uncomfortable for the players who chase these offers: the overwhelming majority of “welcome bonuses” have negative expected value the moment you accept them. A 100% match with 35x wagering on standard slot machines doesn’t just look free — it actively costs you around 40 cents on every promised dollar.
The full dataset is now archived on Zenodo under DOI 10.5281/zenodo.22827233, so anyone can reproduce the numbers. This article explains what I found, why the math is so brutal, and the narrow set of scenarios where a bonus is actually worth taking.
Key Findings
- 73% of the 800 bonus configurations I modelled had negative expected value even before considering game caps, max-bet clauses, or cashout limits.
- Breakeven wagering is set entirely by game RTP. At the industry-standard 96% slot RTP, any wagering requirement above 25x turns a “free” bonus into a guaranteed net loss.
- The typical 100% match + 35x wagering + slots offer costs the average player roughly $40 per $100 of “free” money — the exact opposite of what the marketing suggests.
- A small pocket of bonuses is genuinely +EV: high-match (200–300%) offers restricted to low house-edge games (blackjack at 99.5% RTP with full 100% contribution), or 20x–25x wagering on 97%+ RTP slots.
- The single biggest variable is not the match percentage — it is the wagering requirement multiplied by (1 − RTP). Everything else is noise.
The Setup: Why This Article Exists
Every casino landing page in 2026 shouts the same offer: “200% up to $1,000, plus 100 free spins!” Players see two things — the match percentage and the free-spin count — and internalize the offer as free upside. The wagering requirement, the game-contribution table, and the maximum-bet clause are printed in 9-point grey type at the bottom, and almost nobody models what they actually do to the bonus’s real value.
I have covered iGaming and prediction-market mathematics for the last several years (Knowledge Graph entity /g/11yhjzfsgm), and I kept running into players who told me they had “cleared” a bonus and were somehow further behind than when they started. That is not luck. That is the math working exactly as designed. So I sat down and modelled the whole space.
Methodology: 800 Scenarios, One Simple Formula
The core expected-value formula for a casino bonus, ignoring bet caps and time limits, is:
EV = Bonus − (Wagering × Bonus × (1 − RTP) / Contribution)
Where:
- Bonus = the credited bonus amount (in $)
- Wagering = the playthrough multiplier (e.g. 35 means “wager 35× the bonus”)
- RTP = return-to-player of the game you clear on (e.g. 0.96 for a standard slot)
- Contribution = fraction of each bet that counts toward wagering (1.0 = 100%, 0.10 = 10%)
I ran this across a full parametric grid, publicly deposited on Zenodo:
- 5 match percentages: 50%, 100%, 150%, 200%, 300%
- 8 wagering requirements: 10x, 15x, 20x, 25x, 30x, 35x, 40x, 50x
- 4 game-contribution rates: 10%, 25%, 50%, 100%
- 5 game RTPs: 93.0% (US-style slots), 96.0% (industry standard), 97.5% (high-RTP slots), 98.5% (video poker), 99.5% (optimal-strategy blackjack)
That is 5 × 8 × 4 × 5 = 800 unique configurations. For each, I computed the exact expected value on a $100 deposit basis, the breakeven probability, and the mathematically optimal clearing strategy. The full CSV, the Python notebook, and the LaTeX derivation are all in the Zenodo archive.
The Breakeven Line Nobody Prints
Before we get to individual scenarios, here is the single most important number I extracted from the dataset. Set contribution to 100% (best case for the player) and solve for the wagering requirement at which EV = 0:
Breakeven Wagering = 1 / (1 − RTP)
Print this on a sticky note. It tells you the entire game.
| Game RTP | Breakeven Wagering | What This Means |
|---|---|---|
| 93.0% (US slot) | ~14x | Any offer above 14x is −EV |
| 96.0% (standard slot) | 25x | Almost every “welcome bonus” fails here |
| 97.5% (high-RTP slot) | 40x | 30x–35x offers are marginal but survivable |
| 98.5% (video poker) | ~67x | Even aggressive 50x wagering is +EV |
| 99.5% (blackjack) | 200x | Wagering is essentially free money |
The average online-casino welcome bonus in 2026 sits at 35x–45x wagering on slots at 96% RTP. That is 40%–80% above the breakeven line. The industry did not pick 35x by accident. It picked the smallest number that reliably converts a “generous” match into a house-favourable proposition once players clear it.
The 100% Match Trap: Where Most Players Actually Live
Let’s take the archetypal offer — the one you see 20 times a day if you spend any time on comparison sites: “100% up to $100, 35x wagering, slots only.”
You deposit $100. You get $100 in bonus. The site tells you that you have to wager $3,500 before you can withdraw. You open a 96% RTP slot and start pressing spin.
- Total wagered: $3,500
- Expected house take: 3,500 × (1 − 0.96) = $140
- Bonus received: $100
- Net expected value: −$40
The “free” $100 has, in expectation, cost you $40. And that ignores the max-bet clause (usually $5), the game-cap clause (usually $10 max per spin), the 7-day expiry, and the fact that the wagering-progress bar often excludes slots that people actually want to play. Include realistic drawdown variance and about one in four players will bust their entire deposit before they clear the requirement at all.
Here is the same math across the standard “welcome bonus” grid, contribution held at 100%, RTP fixed at 96%:
| Match | Wagering | Total Wagered | Expected Loss | Bonus | Net EV |
|---|---|---|---|---|---|
| 50% | 25x | $1,250 | $50 | $50 | $0 |
| 50% | 35x | $1,750 | $70 | $50 | −$20 |
| 100% | 25x | $2,500 | $100 | $100 | $0 |
| 100% | 35x | $3,500 | $140 | $100 | −$40 |
| 100% | 40x | $4,000 | $160 | $100 | −$60 |
| 150% | 35x | $5,250 | $210 | $150 | −$60 |
| 200% | 40x | $8,000 | $320 | $200 | −$120 |
| 300% | 50x | $15,000 | $600 | $300 | −$300 |
That last row is worth staring at. A 300% match — the loudest offer on any casino site — with 50x wagering on standard slots has an expected value of negative $300 per $100 deposited. The bigger the match, the bigger the trap, because a bigger bonus mechanically forces a bigger wagering pool at the same house edge.
What Game Contribution Does to the Same Bonus
Now let’s hold the offer fixed — a 100% match with 35x wagering — and vary only the game contribution. This is where casinos hide additional friction: table games often count 10% or less toward wagering, so if you try to clear on blackjack (where the RTP is friendlier) you must wager 10× more nominal dollars to make the same progress.
| Contribution | Effective Wagering on $100 Bonus | Expected Loss (at that game’s RTP) | Net EV |
|---|---|---|---|
| 100% (slots @ 96%) | $3,500 | $140 | −$40 |
| 50% (video poker @ 98.5%) | $7,000 | $105 | −$5 |
| 25% (roulette @ 97.3%) | $14,000 | $378 | −$278 |
| 10% (blackjack @ 99.5%) | $35,000 | $175 | −$75 |
The friendly-looking “you can also clear on blackjack” clause is a mathematical trap for anyone who does not check the contribution table first. Even at blackjack’s excellent 99.5% RTP, a 10% contribution forces you to wager $35,000 to clear a $100 bonus. You would need to be an optimal-strategy player, playing tens of thousands of hands at the maximum allowed bet, and even then you finish behind the pure-slot route because of the variance-adjusted risk of ruin.
The Rare +EV Pocket
Not every bonus is a trap. My dataset flagged 217 out of 800 configurations as positive-EV before behavioural friction. The pattern is consistent and small:
+EV Bonus Profile #1 — Modest match, low wagering, high-RTP slot
| Offer | Expected Value on $100 |
|---|---|
| 50% match, 20x wagering, 97.5% RTP | +$25 |
| 100% match, 20x wagering, 97.5% RTP | +$50 |
| 100% match, 25x wagering, 97.5% RTP | +$37.50 |
These are the offers reload-bonus veterans hunt. They are quietly published, rarely advertised in banners, and often segment-targeted to loyal customers. They exist because the retention team runs the same EV model I did and knows exactly which offers pay out.
+EV Bonus Profile #2 — Aggressive match, table-game contribution, disciplined play
| Offer | Expected Value on $100 |
|---|---|
| 200% match, 30x wagering, 100% blackjack contribution | +$140 |
| 300% match, 35x wagering, 100% blackjack contribution | +$247.50 |
These are almost never offered publicly on major-market casinos because a well-drilled basic-strategy player will systematically drain them. When you do see them (usually crypto-casinos, offshore VIP schemes, or short-lived promotional windows), the max-bet clause and the max-cashout clause are typically ferocious. Read the terms twice.
Behavioural Friction: The Reason Even +EV Bonuses Feel Bad
The Zenodo model quantifies pure mathematical expected value. Real players do not clear bonuses in a vacuum. Three well-documented behavioural effects push the realized outcome below the theoretical EV:
- Variance and risk of ruin. Clearing a 35x wagering requirement at $2 per spin on a 96% RTP slot involves roughly 1,750 spins. The standard deviation of that session is enormous. A meaningful fraction of players hit zero balance before the wagering counter clears, and they get nothing — no bonus, no refund, no partial cashout.
- Loss-chasing after a drawdown. When the balance goes down, the empirically-observed player response is to raise bet size to “catch up.” That interacts poorly with max-bet clauses, which are typically enforced by voiding the entire bonus and any winnings.
- Time pressure. A 7-day expiry on a $3,500 wagering requirement forces roughly $500/day of turnover. That is a lot of spins, played tired, played fast, and played with the wrong bet sizing.
None of these live in the +EV math. All of them live in the outcome distribution.
Practical Advice for Bonus Hunters
If you are going to play bonuses anyway — and many people will — the dataset points to a small, disciplined playbook:
- Compute the breakeven line first. 1 / (1 − RTP). If the offered wagering is above that number, the bonus is mathematically −EV. Walk away.
- Prefer 20x–25x wagering offers even if the match percentage is smaller. A 50% match at 20x on a 97.5% slot beats a 200% match at 40x on a 96% slot in every scenario in my grid.
- Ignore the free-spin count. Free spins are typically valued at $0.10–$0.20 each and subject to the same wagering nonsense. They rarely move the EV needle.
- Never clear a bonus on a game with less than 100% contribution unless you have specifically calculated the effective wagering.
- Set a stop-loss below the deposit amount. The +EV of the bonus is a distribution. Some paths bust before the counter clears. Do not chase them.
Responsible Gambling Note
This article is a mathematical analysis of casino promotions, not an invitation to gamble. Expected value describes long-run averages; individual outcomes vary widely and can be devastating. If you feel your play has crossed from entertainment into compulsion, contact BeGambleAware (UK), 1-800-GAMBLER (US), or your local problem-gambling helpline. The most reliably +EV move remains the one that requires no math at all: not playing.
Reproduce This Yourself
Every number in this article — every table, every breakeven line, every EV calculation — was pulled from the dataset I deposited on Zenodo. The archive includes the raw parametric grid as CSV, the Python notebook that generated it, the LaTeX derivation of the EV formula, and the readme explaining the assumptions.
- DOI: 10.5281/zenodo.22827233
- Title: Casino Bonus Expected Value Under Wagering Requirements: A Parametric Analysis Dataset
- License: CC BY 4.0 (attribute, then use freely, including commercially)
If you find an error, fork the notebook and open an issue on the linked repository. If you replicate the results in a different modelling framework, I will link your work from the follow-up.
The industry counts on players not doing this math. The math is not hard. That is why I ran it, published the data, and wrote this.
Zavian Thornell is an independent iGaming and prediction-market researcher. His work on bonus expected value, RTP variance, and provably-fair verification is archived at zavian-thornell.com/research and indexed under Google Knowledge Graph entity /g/11yhjzfsgm. He does not accept affiliate revenue from any operator discussed in this article.
