Loot Drop Probability Calculator

Introduction: what a loot drop chance actually promises

Video games hand out their rarest rewards through loot drops: a boss dies, a chest opens, a quest completes, and the game rolls a hidden die to decide whether the item you want appears. Developers publish or leak a drop rate, the probability that a single roll produces the item. A legendary sword at one percent means one roll in a hundred, on average, will yield it.

What players actually want to know is different: if I run this boss a hundred times, will I get it? That question is not answered by the drop rate on its own. It is answered by the cumulative probability of at least one success across many independent rolls, and the answer is almost always lower than intuition suggests. A one percent drop over one hundred kills is not a certainty and not even close to one; it is about 63.4 percent. This calculator computes that number exactly, along with the grind lengths that correspond to 50, 90 and 99 percent confidence, so you can decide whether a hunt is worth your evening.

The at-least-one formula for loot drop probability

Each attempt is modelled as an independent Bernoulli trial with success probability . Counting successes across such trials gives a binomial distribution, and the quantity we need is one minus the binomial probability of zero successes. Because the probability of failing every single time is , the complement rule gives:

The term that gets subtracted is the survival function: the chance that after attempts you are still empty-handed. Writing it on its own makes the shape of a grind obvious, because it decays geometrically rather than reaching zero:

Two consequences follow immediately. First, is never exactly zero for any finite number of attempts, so no amount of grinding guarantees a purely random drop. Second, the curve is steep early and flat late: the first fifty attempts buy you far more probability than the second fifty. That is why doubling a grind rarely doubles your chances.

This page evaluates the formula with -expm1(n · log1p(-p)) rather than the literal 1 - Math.pow(1 - p, n). The two are algebraically identical, but for very small drop rates the naive version loses significant digits when it subtracts a number extremely close to one, while log1p and expm1 are designed to stay accurate near zero. For a 0.001 percent drop over ten attempts the difference shows up in the fourth significant figure.

How to use the loot drop probability calculator

Enter the drop rate per attempt as a percentage and the number of attempts you plan to make, then select Calculate the odds. Enter 1 for a one percent drop, 0.05 for a one-in-two-thousand drop, and use whole numbers for attempts. The result panel reports:

• the chance of at least one drop across your planned attempts;
• the chance of finishing the run empty-handed, which is the same figure subtracted from one;
• the expected number of attempts, the long-run average of one divided by the drop rate;
• the attempts needed to reach 50, 90 and 99 percent cumulative confidence;
• a plotted cumulative-probability curve with your attempt count marked.

The comparison table below is refilled on every calculation so you can see the same drop rate evaluated at three common grind lengths. Everything runs locally in your browser; nothing you type is transmitted anywhere.

Worked example: a 1% mount over 100 boss kills

Take the classic case. A mount drops at one percent per kill, and you commit to one hundred kills. Write the drop rate as a decimal, , so a single failure has probability 0.99. One hundred consecutive failures have probability

so the chance of at least one mount is , or 63.40%. Roughly 37 hunters in every 100 finish those hundred kills with nothing to show for it. That is the number this calculator exists to make visible: one hundred attempts at a one-in-a-hundred chance is a coin flip with a small bonus, not a guarantee.

Push the same grind further and the shape of the curve becomes clear. At 69 kills you cross 50 percent. At 230 kills you cross 90 percent. At 459 kills you cross 99 percent. Getting from 90 to 99 percent costs 229 extra kills, almost exactly as many as the entire journey from zero to 90 percent. The last slice of certainty is always the most expensive part of a grind.

Expected attempts, median attempts and confidence targets

Three different numbers get called "how long will this take", and they disagree. The expected number of attempts until the first success follows the geometric distribution and equals

which is 100 attempts for a one percent drop. The median is smaller: half of all hunters finish inside 69 attempts, because the geometric distribution has a long right tail that drags the mean above the midpoint. And the attempts required to reach an arbitrary confidence target come from inverting the cumulative formula:

Set and you get the 90 percent figure; set it to 0.5 and you recover the median. Because the logarithms are both negative the ratio is positive, and because the result is generally fractional the calculator rounds up to the next whole attempt. Notice that makes the numerator infinite, which is the formal statement that certainty is unreachable.

Reference table: how long common drop rates really take

Every figure below was computed with the formulas on this page and rounded up to whole attempts. Read the gap between the average column and the 90 percent column as the price of confidence.

The pattern in the table is scale invariance. Halving the drop rate roughly doubles every column, because for small the logarithm in the denominator behaves like . A useful mental shortcut falls out of this: the attempts needed for 90 percent confidence are about 2.3 divided by the drop rate, and for 50 percent about 0.69 divided by the drop rate.

Drop tables, multiple items and pity timers

Real games rarely present a single clean probability. Three complications matter most.

Weighted drop tables. Many bosses first roll whether anything drops, then roll which entry of a table it is. If a boss drops loot 40 percent of the time and your item holds a quarter of the table weight, its effective per-kill rate is 0.40 × 0.25 = 10 percent, not 25 percent. Convert to the effective per-attempt probability before using this calculator.

Multiple independent items. If you need two different items that roll independently on the same kill, the chance of completing the set in attempts is the product of the two individual cumulative chances:

For a 3 percent item and a 5 percent item over 60 kills that is 0.839 × 0.954 ≈ 80 percent, noticeably worse than either item alone. Sets are always slower than their slowest piece suggests.

Pity timers. Modern gacha and looter systems often make the per-attempt probability rise once you pass a threshold, and sometimes guarantee the drop outright at a hard cap. The rate is then a function of the attempt index, and the survival function becomes a running product rather than a power:

With a hard cap the product hits zero exactly, which is the one situation where a grind really is guaranteed. The calculator above assumes a constant rate and therefore reports a floor, not a ceiling, for any game with bad-luck protection. The Drop Table Oracle game further down lets you feel the difference: its pity bosses ramp the rate near the cap, and the simulated curve visibly bends upward where a flat-rate curve would keep crawling.

Reading variance without blaming the game

Over hundreds of attempts an observed frequency should approach the true rate, a consequence of the law of large numbers, but "hundreds" is doing a lot of work in that sentence. At a one percent rate, one hundred kills produce zero drops about 37 percent of the time and two or more drops about 26 percent of the time. Neither outcome is evidence that a rate is misreported. A useful discipline is to ask how surprising your result would be if the published rate were correct: if the answer is "it happens to more than one player in twenty", the data is not telling you anything.

The same logic works in reverse when you feel unlucky. If you have gone 300 kills without a one percent drop, the probability of a streak that long or longer is 0.99 raised to the 300th power, about 4.9 percent. Roughly one hunter in twenty is having your evening. It is genuinely bad luck, and it is also completely ordinary.

Practising with the Drop Table Oracle game

Beneath the calculator sits Drop Table Oracle, an interactive prediction game that runs the same arithmetic. Each boss states a per-kill drop rate and a planned number of kills, and you place a marker on a probability arc to predict the chance of at least one drop before any dice are rolled. Locking the prediction starts a live simulation: your own kills scroll across a tape, gold marks land where drops occurred, and a cohort of 240 simulated hunters builds an empirical cumulative curve that converges on the theoretical curve drawn from the formula above.

Later bosses add the complications from the previous section. Twin-drop bosses ask for the chance of completing a two-item set, pity bosses ramp their rate toward a guaranteed cap, and oracle bosses invert the question entirely and ask how many kills are needed before the chance of at least one drop first reaches 90 percent. A bad-luck meter after every roll places your own run inside the distribution, so you can see how often a result that feels broken is simply the tail of a perfectly fair curve.

Limitations and assumptions behind these loot drop odds

This calculator assumes every attempt is independent and shares one fixed drop rate, so it does not model the systems many modern games layer on top of pure randomness. Pity timers, bad-luck protection, and rates that rise with each failure all make your true odds better than this baseline, while drop tables that first roll whether anything drops and then which item can make your target rarer than its headline rate. It also assumes your entered drop rate is accurate; official rates are sometimes undisclosed or affected by difficulty, party size, instance modifiers or limited-time events. Group loot rules, personal versus shared loot, and per-account daily caps can all break the independence assumption outright. Treat the result as a clean statistical baseline for a single independent drop, then adjust for the specific rules of your game.

Loot drop probability questions answered

Sources: the at-least-one result is the complement of the binomial probability of zero successes, defined in the NIST/SEMATECH e-Handbook of Statistical Methods, §1.3.6.6.18 Binomial Distribution; the expected-attempts and confidence-target formulas follow from the geometric distribution, whose mean is 1/p, and the underlying trial model is the binomial distribution. The numerically stable evaluation uses Math.expm1 and Math.log1p rather than a direct power. Actual in-game rates depend on each developer's published drop tables and any pity mechanics. Last reviewed August 2026.