← Return to Master Citadel

The Loot Box Vault: Probability & RNG

Unmask the mathematics behind game drop rates, Random Number Generators (RNG), Expected Value matrices, and team composition combinatorics!

Step 1: The Briefing

Don't Get Scammed by RNG! Unlocking Statistical Edge

Every time you crack a mystery chest or roll for a 5-star character in a gacha game, a pseudo-random number generator (PRNG) rolls between \(0.000\) and \(1.000\). The game feels magical or "cursed", but it is strictly controlled by three mathematical weapons:

1. Theoretical Probability P(E):

For equally likely outcomes, \(P = \frac{\text{Favorable}}{\text{Total}}\). A fair six-sided die has one favourable outcome for rolling a 6, so the probability is \(1/6 \approx 16.7\%\). Weighted outcomes require their actual probabilities.

2. Expected Value E(X):

The long-run average payout per chest! Multiply each reward by its probability: \(E(X) = \sum x_i \cdot P(x_i)\). If \(E(X) < \text{Cost}\), the chest is a statistical trap!

3. The Gambler's Fallacy:

"I lost 10 times in a row, so I am DUE for a win!" False! Coin flips and independent dice rolls have zero memory. Every roll is born fresh.

Step 2: The Lab Bench

Gacha Drop Rate & Law of Large Numbers Lab

Open chests and watch experimental frequencies converge to theoretical probability!

Chests Opened
0
Gold Spent
0 G
Loot Value Won
0 G
Net Balance
0 G
Theoretical E(X)
78.5 G / chest
Observed Average Payout
0.0 G / chest
Latest Drop:
None Yet
Step 3: Sparring Ring

Expected Value Battle Drill

A carnival wheel costs 10 Gold to spin. It has an 80% chance of paying 0 Gold, and a 20% chance of paying 40 Gold. What is the Expected Value \(E(X)\), and should a rational gamer play?

DIFFICULTY:
Step 4: The Arcade Arena Round 1/8: Novice Dice
Vault Lock Timer
STATISTICAL CIPHER PROTOCOL:
Rolling a fair 6-sided die. What is the exact probability of rolling a number greater than 4?
Evaluate probabilities and statistical expectation!
Step-by-step lesson

Probability: outcomes, independence and expected value

Before you start: Use fractions, percentages and multiplication, and count small sets of possibilities.

Count outcomes only when they are equally likely

A fair six-sided die has six equally likely outcomes. Two are greater than 4, so P(result > 4) = 2/6 = 1/3. Counting favourable outcomes over total outcomes requires equal likelihood. For a weighted die, use the actual probabilities instead. Probabilities lie between 0 and 1, and all mutually exclusive, exhaustive outcomes have probabilities summing to 1.

Independent events and the complement

For a fair coin flipped twice, HH, HT, TH and TT are equally likely. Three include at least one head, giving 3/4. Alternatively, the only failure is TT, whose probability is (1/2)(1/2) = 1/4. Therefore P(at least one head) = 1 − 1/4 = 3/4.

If each independent attempt succeeds with probability 0.1, three attempts all fail with probability 0.9³ = 0.729. At least one succeeds with probability 0.271, or 27.1%, not 30%. After earlier failures, the next attempt still has probability 0.1. A rule that changes the chance after failures makes the trials dependent and needs a different calculation.

Expected value is a weighted average

A fictional chest pays 20 points with probability 0.75 and 100 points with probability 0.25. Its expected payout is 20 × 0.75 + 100 × 0.25 = 15 + 25 = 40 points. If opening costs 50 points, expected net change is 40 − 50 = −10 points. No single opening pays 40 here: expectation describes an average, not a promised reward. A small simulation can differ substantially from that average.

When order or replacement changes the count

Choosing two different letters from A, B, C, D for an ordered code gives 4 × 3 = 12 possibilities. Choosing a two-person team from four people gives 12/2 = 6, because each pair was counted in both orders. Drawing two red counters without replacement from 3 red and 2 blue has probability (3/5)(2/4) = 3/10. The second fraction changes because one counter has been removed.

Try it yourself

A fictional reward pays 0 points with probability 0.6 and 50 points with probability 0.4. Find the expected payout and the expected net change if an attempt costs 25 points.

Show answer and explanation

Expected payout = 0 × 0.6 + 50 × 0.4 = 20 points. Expected net change = 20 − 25 = −5 points per attempt. This does not predict the outcome of one attempt.

Use the activity to check your understanding

Predict a theoretical probability or average before running the simulation. Compare repeated trials without treating a streak as a guarantee.

Try the next activity: Chart Raider →

Sources and further reading