Gambling guides, explained simply
Start with a question you want answered. Each guide gives you the idea, practical steps and an example. Save your progress with a free account and earn Atlas Chips as you learn.
START HERE · FREE INTRODUCTIONStart with a question you can check.
Choose a game or concept, read the worked example and follow its source. Good guidance explains the rules as well as the conclusion.
“This game has a high return” needs an exact paytable and strategy assumption before it becomes useful.
Find the answer behind the next decision.
2 detailed guides. Original explanations, worked examples and links to the source rules.
Video poker
Ultimate X, Bonus Streak and Gold: which rules apply?
Compare the three Ultimate X families, their stored state, wager costs and the right Atlas tool. Similar names do not mean interchangeable strategies.
Read the guide Video poker · 5 min9/6 vs 8/5 Jacks or Better: read the complete paytable
Compare 9/6 and 8/5 video poker, the full Jacks or Better paytable, royal-flush assumptions, total wager and expected cost.
Read the guideBrowse the original field guides & checklists
24 guides here, plus the established blackjack, video poker, poker and AP lessons within their atlases.
Research methodSlots & RTPHow slot RTP and advantage-play calculators work
A calculator can make the arithmetic precise while its inputs remain uncertain. Understand the difference between a published game return, a state-dependent estimate and a source’s entry rule.
Read guide +
- Ordinary theoretical RTP is the sum of each outcome’s probability multiplied by its gross payout, divided by the wager. For a game with stored features, a full model must also account for transitions, resets and future payouts.
- A playability calculator asks a different question: what is the expected return starting from this particular counter or board state, under a specified stopping policy? A positive estimate depends on the exact configuration and the model being correct.
- Probabilities may come from manufacturer mathematics, a fully specified analytical model, observed play or statistical estimation. Simulating many spins only tests the supplied model; it cannot recover unknown reel weights by itself. Ask for the configuration, sample size, raw data, assumptions and validation.
- SlotFarmers publishes hit-frequency and payout inputs on some guides and describes simulations on others. In the material reviewed, we did not establish access to manufacturer PAR sheets, reproducible simulation code or an independent audit. Its figures are recorded as source estimates, not certified machine RTP.
- A random bonus’s average trigger interval is not a deadline. If a model stops when a particular feature hits, it must account for other bonuses, meter growth, resets, cost and the stopping rule. Excluding a rare prize can describe a conservative scenario but is not the complete theoretical RTP.
- Atlas labels tools by function: state checkers compare your screen with a cited rule; distance tools count what remains to a cap; return models estimate expected payouts from explicit inputs. An unknown probability is left unknown rather than replaced with a precise-looking percentage.
SlotFarmers’ reviewed Wolf Run Eclipse calculator displayed 290 expected spins, 76.80% return and −67.28 betting units for a Mini chase. The net conversion is 290 × (0.768 − 1) = −67.28. That arithmetic checks out; it does not independently validate the assumed trigger rate, payouts or chase-return model.
Research methodSlots & RTPOnline persistent features: research the ownership rules
A public description of a collecting or sticky feature is a useful lead. It is not enough to establish an exploitable state.
Read guide +
- Book of 99 describes collecting 99 books for free spins. Establish what happens across sessions, denominations and account changes before modelling anything.
- 3 Liberty Eagles describes prizes that can remain on the reels until collected. That does not establish whether another player can inherit them or their conditional value.
- Record the actual configured RTP, reset rule, required stake and expected future award. A state can require substantial extra wagering before release.
- Read promotional restrictions: operators may exclude play that accumulates features under one balance and releases them under another. Do not assume a game mechanic overrides those terms.
These are manufacturer-backed feature leads. No universal “play at X” threshold or positive-EV conclusion has been established.
Research methodGetting startedYour first advantage-play research plan
An edge is a positive expectation under stated conditions. A winning session, high RTP or impressive tier card does not establish one.
Read guide +
- Identify the exact game, rules, price and state. Keep the source and its date.
- Write down the proposed benefit, expected cost, uncertainty and exit condition before playing.
- Confirm residence, location, account and promotion eligibility.
- Start with practice and independent calculations. Record losses and abandoned opportunities as well as wins.
A 99.54% game still has negative expectation before realisable rewards, errors and expenses.
Conditional advantageComps & cruisesOffer matching: turn evidence into a clear quote
A casino may review a competitor’s current offer and propose its own package. A tier match transfers status; an offer match concerns the actual room, freeplay or dining offer.
Atlas Pro guide +
Cost reductionComps & cruisesTheoretical loss, actual loss and ADT
Theoretical loss is a model of expected gaming cost. Actual loss is your realised result. Hosts may use their own estimates, time records and gaming-day definitions.
Atlas Pro guide +
Cost reductionComps & cruisesShould you chase the next tier?
A loyalty threshold is a spending target, not an investment return.
Atlas Pro guide +
Cost reductionComps & cruisesWhat a “free cruise” really includes
A casino sailing offer may cover a cabin fare while leaving substantial costs or play conditions.
Atlas Pro guide +
Cost reductionComps & cruisesCaesars: credits, tiers and destinations
Tier Credits measure status; Reward Credits fund eligible redemptions. Personal offers form a third, separate category.
Atlas Pro guide +
Conditional advantagePromotionsCompare rakeback on the right denominator
Poker cashback is usually a share of eligible rake. Slot rewards may be quoted against coin-in, points or theoretical loss. These are not interchangeable.
Read guide +
- Identify the exact base: contributed rake, weighted rake, dealt rake, loss or wagering.
- Calculate cash received after caps, expiry, redemption thresholds and exclusions.
- Subtract the game’s underlying expected cost and any extra volume needed.
- Treat “up to” as a ceiling. Verify your tier and local operator rather than using an affiliate headline.
£100 back from £1,000 eligible rake is 10% effective rakeback. It is not 10% of the money you wagered.
Conditional advantagePromotionsDeposit bonuses and free spins: read the whole offer
Face value is not cash value. Terms determine what can be withdrawn and how much must be risked.
Read guide +
- Record minimum deposit, opt-in order, eligible games, time limit and payment exclusions.
- Check whether wagering applies to bonus, deposit plus bonus, or spin winnings.
- Record maximum stake, game contribution, win caps, excluded play and withdrawal rules.
- For completion-dependent bonuses, a simple turnover × edge estimate omits busting before completion. Use it as a cost illustration, not an EV guarantee.
£20 bonus with 10× bonus-only wagering creates £200 qualifying turnover. If the game contributes 20%, actual turnover would be £1,000.
Conditional advantageMatched bettingMatched betting: cash qualifiers and free bets
A back bet and a lay bet can offset the same sporting outcome. A qualifying bet normally costs money; an eligible free bet may then create value.
Read guide +
- Read the promotion before depositing: minimum odds, markets, stake-return rule, deadline and excluded combinations.
- Match selection, event and settlement rules on the bookmaker and exchange; compare void, dead-heat and extra-time treatment.
- Calculate lay stake and liability, confirm sufficient exchange funds, then check that the lay is actually fully matched.
- Record both settlements and commission. Do not proceed with uncertain rules, unmatched exposure or a bonus you are not eligible to receive.
Use the calculator’s separate cash and stake-not-returned modes. Neither removes operator restrictions, errors, liquidity risk or settlement differences.
Conditional advantageSports & exchangesArbitrage and dutching: prices are only half the story
For mutually exclusive, exhaustive outcomes, an inverse-odds sum below one indicates a theoretical price arbitrage before costs.
Read guide +
- Verify that all bets settle on the same event definition and cover every outcome.
- Include commission, currency conversion, taxes and stake limits.
- Account for odds moving, partial fills, voids and asymmetric settlement.
- Never call an unplaced quote a locked profit. Keep a timestamped execution record.
At decimal odds 2.10 and 2.10 on the only two outcomes, 1/2.10 + 1/2.10 ≈ 0.9524. This alone does not prove you can execute both legs.
No inherent advantageTable gamesRoulette: choose the rules, not a lucky pattern
Place an inside or outside wager before the wheel resolves. Standard payouts do not compensate fully for zero pockets.
Read guide +
- Single-zero even-money play has a lower house edge than double- or triple-zero.
- La partage can halve the loss on zero for specified even-money bets. Confirm the actual rule.
- Past spins do not make red or black due on an independent fair wheel.
- A progression changes stake exposure, not the wheel probabilities. Try identical settings in the simulator.
On a single-zero wheel, an even-money bet returns 36/37 of turnover on average: approximately 97.30%.
No inherent advantageTable gamesBaccarat: Banker, Player and the tie trap
Choose a wager; fixed drawing rules decide the two hands. Card values are totalled modulo ten.
Read guide +
- In the standard eight-deck game, Banker wins pay 0.95:1 after commission; ties push Banker and Player wagers.
- Player pays 1:1 but has a slightly larger house edge than standard commission Banker.
- An 8:1 Tie bet is much more expensive in expectation.
- No-commission variants change specific payouts; they need their own analysis. Scoreboard streaks are not a proven signal.
Banker RTP is about 98.94% under the specified rules. That remains negative expectation without a separate, sufficient benefit.
No inherent advantageTable gamesCraps: learn the pass line first
On the come-out roll, 7 or 11 wins a pass-line bet; 2, 3 or 12 loses. Other totals establish a point.
Read guide +
- Once a point is set, the point must repeat before seven for the pass-line wager to win.
- An odds bet behind the line can pay true odds, but the required line bet retains its house edge.
- Compare edge per resolved wager consistently; per-roll figures measure something different.
- Avoid mixing place, come, don’t-pass and proposition payouts when calculating return.
Our simulator resolves the complete pass-line wager using independent dice rolls, not just a single come-out roll.
No inherent advantageBetting systemsMartingale, Fibonacci, D’Alembert and Paroli
Stake progressions redistribute wins and losses. They cannot turn independent negative-expectation wagers into positive expectation.
Read guide +
- Martingale doubles after a loss and resets after a win; a long loss sequence escalates exposure rapidly.
- Fibonacci advances one step after a loss and retreats two after a win. D’Alembert adds or subtracts one unit.
- Paroli increases after wins; the simulator resets after three wins or any loss.
- Finite funds and table limits interrupt recovery. A high percentage of small winning sessions can coexist with a negative mean.
With a £1 base stake, six Martingale losses cost £63 and require a £64 next wager. On commission Banker, doubling does not exactly recover all prior losses.
Research methodSlots & RTPRTP: the number, the configuration and the caveat
Return to player is a long-run expected payout divided by wagering for a specified game model. It is not a session promise.
Read guide +
- Record the exact title, provider, paytable/version, denomination and bet mode.
- Distinguish manufacturer published RTP from the configuration installed at an operator.
- High volatility changes the distribution and bankroll needs; it does not, by itself, change the stated mean.
- For AP machines, baseline RTP is separate from the extra value of a persistent state or promotional benefit.
£10,000 turnover at 97% RTP has £300 expected loss. That says nothing about the exact result of one £100 session.
Conditional advantageSlots & RTPMust-hit-by meters: a ceiling is not an entry price
A guaranteed trigger boundary narrows when an event must occur, but does not establish what it is worth to chase.
Read guide +
- Identify whether the meter belongs to the bet level, denomination, machine or linked bank.
- Confirm reset, trigger distribution, contribution rate and expected award from the actual rules.
- Include ordinary game losses while collecting progress and the full cost of every required wager.
- Treat published entry thresholds as source claims until their model and cabinet configuration are verified.
The supplied Buffalo Link screenshots are retained in its evidence tab. Their 1,250/1,550 thresholds are publisher claims, not independently proved universal break-even points.
Research methodResearch methodsHow to investigate a manufacturer manual
A feature brochure describes behaviour. A PAR sheet or configuration-specific probability model may be needed to establish expected value.
Read guide +
- Record manufacturer, software revision, jurisdiction, game class and full wager configuration.
- Find ownership and persistence rules: what survives cash-out, what resets and what changes with a bet.
- Locate actual probability and payout data. Do not infer them from artwork, animation or unrelated patents.
- Archive permitted evidence and derive a reproducible calculation. Mark missing inputs explicitly.
“Collect symbols” can describe either real persistent value or merely cosmetic progress. The wording alone is insufficient.
Conditional advantagePromotionsFreeplay, matchplay and loss rebates are different
A non-cashable stake, a coupon paired with your cash, and a refund on losses have different payoff structures.
Read guide +
- Check whether the promotional stake is returned after a win.
- For once-through slot freeplay, expected conversion depends on the exact game; completion and cashout restrictions may change it.
- For matchplay, model both your cash stake and the coupon’s payout rules.
- For loss rebates, model the loss distribution, cap, time window and redemption conditions rather than subtracting a flat percentage from every wager.
A £100 non-cashable chip that wins even money returns £100 profit on a win, not necessarily £200 withdrawable cash.
Research methodOnline & cryptoCrypto casinos: payment method is not an edge
A cryptographic fairness mechanism can address outcome verification. It does not remove a house edge or establish licensing, solvency or payout reliability.
Read guide +
- Check the actual operating entity, regulator and restrictions for both residence and physical location.
- Understand KYC, withdrawal limits, custody, network charges and asset-price exposure.
- Read the game’s payout mapping and edge as well as any seed verification instructions.
- Do not use a VPN or travel assumption to bypass excluded-resident terms.
Stake.com’s current terms exclude UK residents as well as persons located in the UK; its presence in the directory is a restriction record, not an endorsement.
Research methodCommunityWrite a session report people can actually use
Good field notes help distinguish a repeatable mechanism from an isolated result.
Read guide +
- Include game, venue/country, date, configuration and visible starting state.
- Explain the source of your decision and identify every assumption.
- Report total stakes or turnover, time, expenses and net result—not just a jackpot photo.
- Redact player IDs, QR codes, balances unrelated to the report and other people’s faces. Respect venue photography rules.
Ranks count source-linked contributions by category. They do not certify expertise, profitability or the truth of a post.
Research methodComps & cruisesBuild a casino trip around confirmed opportunities
A cluster of resorts is a useful scouting route, but it is not proof of available AP games or transferable U.S. comps.
Atlas Pro guide +
Conditional advantagePokerTournament overlays and guaranteed prize pools
A guarantee can add value if entries and fees do not fully fund the promised prize pool. Your skill, payout structure and competition still matter.
Read guide +
- Separate entry money from fees and add-ons.
- Confirm whether the guarantee can change, whether late registration remains open and what counts toward it.
- For equal-skill illustration only, divide the prize pool by paid entries and compare with the full buy-in.
- Use tournament-specific strategy and ICM; overlay is not a guarantee that a particular player is profitable.
A £10,000 pool with 80 entries costing £110 each gives £125 average gross prize per entry before skill differences and expenses.
Connect the guide to deeper research.
Go further with developed machine research, structured lessons and your own review notes.
- Additional developed guides: the state to inspect, source conditions and reset behaviour
- Supported game-specific checkers, with model assumptions and limits
- Private study notes for machine checks and questions to revisit
Checking your access…
One membership across every subject. View the benefits and planned prices; no payment is taken today.Public strategy explainers and beginner guides.
Browse the free guide library