Loot Drop Probability Calculator
Calculates the cumulative probability of getting at least one of a desired item across multiple independent attempts at a given per-attempt drop rate. Applies to gaming loot drops, dice-roll board game outcomes, and any binomial 'at least one success' problem.
Last updated: May 2026
Compare with similar
About this calculator
The probability of getting at least one success across N independent trials, each with probability p, is the complement of the probability of zero successes: P(at least one) = 1 − (1 − p)^N. Expressed as a percentage with p in percent: probability% = (1 − (1 − dropRate/100)^attempts) × 100. Variables: dropRate (per-attempt success probability in %), attempts (number of independent tries). This is the canonical 'gambler's question' — how many rolls/attempts before I hit the result I want? — and applies equally to MMO loot drops, board game dice mechanics (rolling a 6 on at least one of N dice), trading-card pack pulls, and crit-rate streaks in RPGs. Edge cases: this formula assumes independent trials with constant probability. Loot tables that change based on prior drops ('pity timers', 'bad luck protection' in Hearthstone or Genshin Impact) violate the independence assumption — true probability is higher than the formula predicts as you accumulate attempts. Conversely, RNG seeding that loops can cluster outcomes (often perceived as 'streaks'). The formula gives the cumulative chance of at least one drop — not the expected number of drops (that's N × p) or the probability of exactly k drops (binomial PMF). For very low drop rates, the rule of thumb 'expected attempts ≈ 1/p' is a good approximation for the median; for the 95% confidence (95% chance of getting at least one), attempts ≈ −ln(0.05) / p ≈ 3 / p. Common applications: dice ('chance of rolling at least one 6 in 4 rolls' = 1 − (5/6)^4 ≈ 51.8%); MMO drops (rare boss item with 5% rate, 30 attempts → 1 − 0.95^30 ≈ 78.5%); gacha games (3-star pull at 1.5%, 40 pulls → 1 − 0.985^40 ≈ 45.4%).
How to use
Example 1: 5% drop rate, 30 attempts. Step 1: per-attempt fail rate = 1 − 0.05 = 0.95. Step 2: 30-attempt fail rate = 0.95^30 ≈ 0.2146 (21.5%). Step 3: at-least-one success = 1 − 0.2146 = 0.7854 → 78.54%. Verify: rule of thumb '3/p' gives 60 attempts for 95% confidence, so 30 attempts giving ~78% confidence is consistent with that approximation. Example 2: 1% drop rate, 100 attempts. Step 1: fail rate = 0.99. Step 2: 100-attempt fail = 0.99^100 ≈ 0.3660 (36.6%). Step 3: at-least-one = 1 − 0.366 = 0.6340 → 63.4%. Verify: 100 attempts at 1% gives ~63% chance — much lower than the naive '100 × 1% = 100%' expectation, illustrating why the at-least-one formula is essential. Expected number of successes is 1 (since 100 × 0.01 = 1), but only 63% chance of getting AT LEAST one.
Frequently asked questions
How do I calculate the chance of getting an item after multiple attempts?
Use the 'at least one success' formula: P = 1 − (1 − p)^N, where p is the per-attempt drop rate (as a decimal) and N is the number of attempts. This works because (1 − p) is the chance of NOT getting the item on a single attempt, and (1 − p)^N is the chance of failing N times in a row. One minus that gives the chance of succeeding at least once. The formula assumes each attempt is independent — earlier failures don't change the probability of later successes. For drop rates with 'pity' systems (Genshin, Hearthstone, Diablo's smart loot), the real cumulative probability is higher than this formula predicts because the per-attempt rate increases after consecutive failures.
Why does a 1% drop rate not give 100% after 100 attempts?
The naive multiplication 100 × 1% = 100% incorrectly assumes drops accumulate additively. In reality, each attempt is independent and the chance of failing 100 times in a row is 0.99^100 ≈ 36.6% — so the chance of getting at least one drop is only 63.4%. To reach 95% confidence of at least one drop, you need approximately −ln(0.05) / p = 3 / 0.01 = 300 attempts. The expected number of drops is exactly 100 × 1% = 1 (the mean), but the median number of attempts to see your first drop is around 69 (where 1 − 0.99^69 ≈ 50%). The right-skewed distribution means you're as likely to get your first drop very early as you are to wait far longer than expected — variance is enormous at low drop rates.
How does this apply to board games and tabletop dice rolls?
The exact same formula governs board-game dice probability. Rolling at least one 6 on N six-sided dice: p = 1/6 ≈ 16.67%, so N = 1 gives 16.67%, N = 4 gives 1 − (5/6)^4 ≈ 51.8% (the classic De Méré problem solved by Pascal). For two dice rolling at least one specific number, treat each die as an independent attempt. Catan resource production, Yahtzee scoring chances, Risk attack outcomes, and Settlers of Catan robber placement decisions all benefit from this math. For card-drawing problems (Magic: The Gathering, Hearthstone), the math is different because draws without replacement use hypergeometric distribution, not binomial — the formula here only applies if you shuffle the discarded card back.
What are common mistakes when calculating cumulative probabilities?
Adding probabilities instead of compounding failures — '50 attempts at 5% = 250%' is meaningless; use 1 − (1 − p)^N instead. Forgetting that p must be a decimal in the formula, not a percent (use 0.05, not 5). Confusing 'expected drops' (mean = N × p) with 'probability of at least one drop' — they describe different things. Treating successive attempts as dependent when they're independent, or vice versa. Applying the binomial formula to card draws without replacement (use hypergeometric instead). Ignoring 'pity' mechanics in modern gacha and loot games where rates increase after failures — true probability is higher than the formula predicts. Using the formula to predict 'when' you'll get a drop — it tells you the cumulative probability but not the exact attempt; the distribution of first-success attempt is geometric, heavily right-skewed.
When should I NOT use this binomial probability formula?
Games with pity timers, bad-luck protection, or guaranteed drops at certain milestones violate the constant-probability assumption — true chance is higher than the formula predicts. Card-drawing games (MTG, Hearthstone, Pokemon TCG) use draws WITHOUT replacement, requiring the hypergeometric distribution instead. Dependent events — e.g., 'roll two 6s in a row' — need conditional probability (multiplication of dependent probabilities), not the formula here. Games with conditional drop pools that change as you progress (Pokemon evolution, Diablo paragon drops) have time-varying p that this formula can't model. Server-side RNG seeding can cluster results in ways pure independence doesn't predict. Pure random number generation in software is pseudo-random; truly fair physical dice differ from RNG implementations. For exact probability of getting EXACTLY k successes (not 'at least one'), use the binomial PMF: C(N,k) × p^k × (1−p)^(N−k). Finally, for very small probabilities and many attempts (N × p < 5 and p < 0.05), the Poisson approximation is computationally easier but slightly less accurate.