Does a 5% game drop become certain after 20 tries?

Twenty attempts, still no drop

A rare item has a 5% chance of dropping when you defeat an enemy. After twenty victories, it is tempting to calculate 5 × 20 = 100% and expect certainty. But with independent attempts and an unchanged drop probability, the chance of at least one drop within twenty attempts is about 64.2%. About 35.8% remains on the no-drop side.

These are calculations for an invented game, not findings about any commercial title. We have fixed the probability at 5%, made attempts independent and excluded hidden changes. A real guarantee or changing probability would require a different model.

The interesting question is not simply who gets lucky. It is how the system designs the wait. Three rules can share the same 5% first attempt and produce very different experiences afterwards.

Count “no drop” before counting “at least one”

The probability of no drop in one attempt is 0.95. Under independence, no drop in two attempts has probability 0.95 × 0.95. After twenty, it is 0.95 raised to the twentieth power, about 0.3585. Subtract the no-drop outcome from all possible outcomes and the probability of at least one drop remains: 1 − 0.95ⁿ.

Adding 5% twenty times does not calculate that probability. Outcomes with multiple drops overlap when you simply add the separate chances. However, 20 × 0.05 = 1 does give the expected number of drops over twenty attempts. An average of one item is not a promise that everybody receives one.

Imagine a hundred players each making twenty attempts. Some could receive several items and some none. The 64.2% probability helps describe the expected proportion succeeding under the model; it does not require exactly 64 successful players in each group of a hundred.

Does the system owe you success after nineteen failures?

If attempts are independent and the probability stays fixed, the next attempt still has a 5% chance after nineteen failures. The system has not accumulated a debt to the player. Our knowledge of past results has changed; the rule for the next draw has not.

Compare two questions: “Will there be a drop within twenty attempts?” and “Given nineteen failures, will the next attempt succeed?” The first includes twenty still-uncertain opportunities when asked at the start. The second concerns one remaining opportunity after the earlier outcomes are known.

This conclusion follows from independence, not from the word “random” alone. A designer can keep track of misses and make later attempts more favourable. Without checking the actual rules, we cannot apply the independent model to every game.

A twentieth-attempt guarantee cuts off the long wait

In the second invented system, the first nineteen attempts use independent 5% draws. Anyone still without a drop receives one on attempt twenty. The cumulative success probabilities match the independent model through attempt nineteen. At twenty, everyone still waiting succeeds.

The guarantee here covers the wait for the first item only. We have not specified reset rules for continued play or shared guarantees across different items. A real “guarantee” needs those details before it can be evaluated.

This rule does not distribute wins evenly from the beginning. It retains the possibility of an early drop and places a ceiling on the longest wait. Move the diagram from nineteen to twenty to see the jump. It represents a rule change at a boundary, not a sudden change in a player’s personal luck.

Drawing without replacement changes the situation after each miss

The third system shuffles twenty cards, with one winning card, so that all orders are equally likely. Draw one card at a time without putting any back. The first chance is 1/20, or 5%. After a known miss, nineteen cards remain with one winner, so the next chance is 1/19.

Here every failed draw changes the contents of the bag. The chance of finding the winner within n draws is n/20 for n up to twenty: 50% within ten draws and 100% within twenty. Both this system and the guarantee reach certainty at twenty, but they take different routes.

Before seeing any cards, the winner has a 5% chance of occupying each position. After observing misses, the next-draw probability changes. There is no contradiction: an unconditional probability and a probability conditioned on the results so far use different information.

Fairness depends on which experience you want to protect

Independent draws preserve the same chance at every attempt. A guarantee limits a long unlucky streak. Drawing without replacement fixes the number of winners inside one set. These systems protect different conditions rather than expressing one identical kind of fairness.

Their expected waits for the first item also differ: 20 attempts for the independent model, about 12.83 for the twenty-attempt guarantee and 10.5 for the bag. In this setup, the guarantee changes the mean as well as the longest possible wait. The calculations are 1/0.05; the sum of 0.95⁰ through 0.95¹⁹; and the average of positions 1 through 20, respectively.

That does not establish one best rule for every game. The meaning of a miss, the length of an attempt and the way players compare experiences can all influence the design choice. The model lets us name the trade-off instead of hiding it inside the word “fair”.

Our perspective: read the rule behind the percentage

The point of this exhibit is not to keep drawing until a satisfying result appears. It is to notice what stays fixed and what changes. A displayed 5% needs a rule behind it: independence, removal from a finite set or a guarantee at a particular boundary.

Separate the chance on one attempt, the chance of success by a chosen attempt and the longest possible wait. An unpleasant result can then lead to a clearer question: is it variation allowed by the model, or did we assume the wrong model?

These calculations do not certify a commercial service or suggest that continuing to play is worthwhile. They offer a way to connect a small probability on a screen with the experience designed around it.

An editorial perspective This section offers comparisons and questions based on facts in the sources. It does not claim a shared underlying principle or present new research results.