Article — Dice Average Calculator
Dice Average Calculator: Expected Value and Standard Deviation for NdS Rolls
The expected average of a single S-sided die is (S + 1) / 2. For 1d20, that is 10.5; for 1d6, it is 3.5. Rolling N dice scales the average by N: 3d6 averages 10.5, and 2d20 averages 21. Modifiers shift the average but not the standard deviation. The dice average calculator returns expected value, minimum, maximum, variance, and standard deviation for any NdS roll plus an additive modifier.
Dice statistics show up in tabletop role-playing games, board games, probability classes, and Monte Carlo simulations. The formulas are exact for fair dice, where each face has equal probability 1/S.
What dice average means
The dice average, also called the expected value or expected average, is the long-run mean of repeated rolls. For 1d6, the six possible outcomes (1, 2, 3, 4, 5, 6) each occur with probability 1/6. Their weighted average is (1 + 2 + 3 + 4 + 5 + 6) / 6 = 3.5.
Any individual roll lands on an integer, but the average converges to the expected value as the number of rolls grows. The law of large numbers guarantees this: with enough trials, the observed sample mean approaches the theoretical expected value.
Dice average formula
E(1dS) = (S + 1) / 2 one dieE(NdS) = N(S + 1) / 2 N diceE(NdS + M) = N(S + 1) / 2 + M with modifierVar(1dS) = (S² − 1) / 12 variance per dieσ = √(N(S² − 1) / 12) standard deviationThe expected value formula comes directly from averaging the integers 1 through S. The variance formula follows from computing the second moment of the discrete uniform distribution.
Dice standard deviation
Standard deviation measures roll-to-roll variability. A low σ (close to 0) means rolls cluster near the average; a high σ means wild swings are common. For a d4, σ = 1.118; for a d20, σ = 5.766. Bigger dice are more volatile.
For N independent dice, variance adds: Var(NdS) = N · Var(1dS). But standard deviation grows as the square root of N, not linearly. Rolling 4d6 has σ ≈ 3.42, only twice the σ of 1d6 (1.708) even though the average is four times larger. Multiple smaller dice produce more predictable totals than one large die.
The oldest known dice are about 5,000 years old, found at the Burnt City (Shahr-e Sukhteh) archaeological site in Iran. Dice precede written history; the polyhedral dice used in modern role-playing games (d4 through d20) were standardized only in the 1970s with the release of Dungeons & Dragons. Before that, gaming dice were almost universally six-sided.
NdS notation in RPGs
Tabletop role-playing games use the notation NdS to mean "roll N dice, each with S sides." The most common dice are d4, d6, d8, d10, d12, and d20 — the five Platonic solids plus the pentagonal trapezohedron (d10). Specialty dice include the percentile d100 (often rolled as two d10s) and d2 (a coin flip).
A modifier follows the dice notation: 3d6+2 means roll three six-sided dice and add 2 to the total. The expected value is N(S+1)/2 + M = 3(7)/2 + 2 = 10.5 + 2 = 12.5. Critical: the modifier shifts the mean but does not affect the dice's standard deviation.
Worked dice average examples
Single d20 attack roll. 1d20 averages 10.5 with σ = 5.77. The probability of rolling at or above any given target N is (21 − N) / 20. A target of 15 gives probability 6/20 = 30%. The high variance means luck dominates over many small differences in modifiers.
3d6 for D&D ability scores. Average = 3 × 3.5 = 10.5. σ = √(3 × 35/12) ≈ 2.96. The distribution is bell-shaped: scores 10–11 are most likely (each about 12.5%), scores 3 and 18 each occur with probability 1/216 (0.46%). Compared to a single d20, the spread is much tighter.
2d20 with advantage. Rolling 2d20 and keeping the higher gives a different distribution than 2d20 sum. The expected maximum of two d20 rolls is about 13.83, with the boost concentrated in the middle of the range (rolls of 9–14 become much more likely; rolls of 1 and 20 stay rare).
Multiple dice and the bell curve
A single die has a flat (uniform) distribution: every face is equally likely. Adding more dice transforms the shape via the central limit theorem. By 3d6, the distribution is already clearly bell-shaped; by 10d6, it is indistinguishable from a normal curve for practical purposes.
For 2d6, the sum distribution is triangular: sum 7 has 6/36 = 16.7% probability, sums 2 and 12 only 1/36 = 2.78% each. For 3d6, the most likely sums are 10 and 11 (12.5% each), and extreme totals get rare quickly. The bell shape sharpens as more dice are added.
Using dice averages in games
Game designers use dice averages to balance encounters. A combat where each attack averages 8 damage and the enemy has 80 HP should resolve in about 10 successful attacks — assuming hit rates near 50%, that is roughly 20 combat rounds. Adjusting any input (damage, HP, hit rate) shifts the expected length and lethality.
Players often misjudge expected value because individual rolls are memorable. A streak of low rolls feels like the dice are "cursed," but the long-run average always returns. Over 100 d20 rolls, the average should land near 10.5 ± 0.6 (one standard error of the mean). Streaks longer than three rolls in either direction are not unusual.
Common dice average mistakes
Adding a constant modifier shifts every outcome by the same amount, so the spread is unchanged. The expected value moves, but variance and standard deviation stay the same. This is a frequent mistake when comparing builds: a +5 modifier adds reliability to the mean but does not reduce volatility.
- Average of 2d6 is 7, not 6 — the most common sum equals the expected value here
- Most likely sum of 2d6 is 7, with 6/36 = 16.7% probability
- Rolling 4d6 drop lowest averages about 12.24 (D&D character generation method)
- 1d20 is the highest-variance common RPG roll: σ = 5.77, three times that of 1d6
- Multiple smaller dice are more predictable than one large die of equivalent average
- Critical hits (rolling 20 on 1d20) happen 5% of the time on any single roll
Other regular slip-ups: assuming each total is equally likely with multiple dice (only single dice have uniform distributions); confusing "average" with "most likely outcome" (they often differ); and treating non-independent rolls as if they were independent (rolling with advantage or disadvantage changes the math).