Skip to content

Distribution reference ​

OpenRiskSim ships 33 distributions: 24 continuous, 8 discrete (including custom) and the fixed value. This page is generated automatically from the same registry the app uses (packages/distributions/src/registry.ts), so names, parameter keys, defaults and descriptions match the gallery of the Define assumption dialog exactly.

To decide which one to use, see Choosing a distribution.

Parameter keys

The Key column is the parameter's internal name (e.g. stdDev or mode). It is what you'll see if you inspect the model saved in the .xlsx, and what developers use. The interface shows the Parameter label.

Mean and variance formulas ​

Notation: a = minimum, b = maximum, m = most likely (mode), Γ = gamma function, γ ≈ 0.5772 (Euler–Mascheroni constant). Every distribution also supports truncation, which changes the mean and variance (OpenRiskSim recomputes them numerically).

DistributionSupportMeanVariance
Normal (mean μ, stdDev σ)(−∞, ∞)μσ²
Lognormal (mean, stdDev of the variable)(0, ∞)meanstdDev²
Uniform[a, b](a + b)/2(b − a)²/12
Triangular[a, b](a + m + b)/3(a² + m² + b² − am − ab − mb)/18
PERT[a, b](a + 4m + b)/6(μ − a)(b − μ)/7
Beta (alpha α, beta β, on [a, b])[a, b]a + (b − a)·α/(α + β)(b − a)²·αβ / [(α + β)²(α + β + 1)]
Gamma (shape k, scale θ)(0, ∞)kθkθ²
Exponential (rate λ)[0, ∞)1/λ1/λ²
Weibull (shape k, scale λ, location c)[c, ∞)c + λ·Γ(1 + 1/k)λ²·[Γ(1 + 2/k) − Γ(1 + 1/k)²]
Logistic (mean μ, scale s)(−∞, ∞)μs²π²/3
Student's t (df ν, mean μ, scale s)(−∞, ∞)μ (if ν > 1)s²·ν/(ν − 2) (if ν > 2)
Chi-square (df k)[0, ∞)k2k
F (df1 d₁, df2 d₂)[0, ∞)d₂/(d₂ − 2) (if d₂ > 2)2d₂²(d₁ + d₂ − 2) / [d₁(d₂ − 2)²(d₂ − 4)] (if d₂ > 4)
Cauchy(−∞, ∞)undefinedundefined
Gumbel (maximum) (location μ, scale β)(−∞, ∞)μ + γβπ²β²/6
Fréchet (shape α, scale s, location m)(m, ∞)m + s·Γ(1 − 1/α) (if α > 1)s²·[Γ(1 − 2/α) − Γ(1 − 1/α)²] (if α > 2)
Pareto (shape α, scale xₘ)[xₘ, ∞)α·xₘ/(α − 1) (if α > 1)xₘ²·α / [(α − 1)²(α − 2)] (if α > 2)
Laplace (location μ, scale b)(−∞, ∞)μ2b²
Rayleigh (scale σ)[0, ∞)σ·√(π/2)(4 − π)σ²/2
Erlang (k, rate λ)[0, ∞)k/λk/λ²
Arcsine[a, b](a + b)/2(b − a)²/8
Cosine[a, b](a + b)/2(b − a)²·(1/4 − 2/π²)
Power function (alpha α)[a, b]a + (b − a)·α/(α + 1)(b − a)²·α / [(α + 1)²(α + 2)]
Trapezoidal (min, mode1, mode2, max)[a, b]computed numericallycomputed numerically
Bernoulli (p){0, 1}pp(1 − p)
Binomial (n, p){0, …, n}npnp(1 − p)
Poisson (lambda λ){0, 1, 2, …}λλ
Geometric (p, failures before the first success){0, 1, 2, …}(1 − p)/p(1 − p)/p²
Negative binomial (r, p, failures before the r-th success){0, 1, 2, …}r(1 − p)/pr(1 − p)/p²
Hypergeometric (population N, successes K, draws n){max(0, n+K−N), …, min(n, K)}nK/Nn(K/N)(1 − K/N)(N − n)/(N − 1)
Discrete uniform{a, …, b}(a + b)/2[(b − a + 1)² − 1]/12
Customthe given valuesweighted meanweighted variance
Fixed value (value){value}value0

Constraints between parameters

Besides each parameter's range, relationships are validated: minimum < maximum, minimum ≤ most likely ≤ maximum (Triangular, PERT), mode1 ≤ mode2 (Trapezoidal), successes and draws ≤ population (Hypergeometric), etc. The dialog shows the error next to the field.

The Lognormal uses the variable's mean and standard deviation

In OpenRiskSim (as in Risk Simulator) the Lognormal's parameters are the mean and standard deviation of the variable itself, not of its logarithm (as in Excel's LOGNORM.DIST). For a price with mean $26 and sd $3, type 26 and 3 directly.

Full catalogue ​

33 distributions

Normal

normalcontinuous

Symmetric bell-shaped distribution defined by its mean and standard deviation.

KeyParameterDefaultValid values
meanMean100any number
stdDevStandard deviation10> 0

Typical use: Measurement errors, natural phenomena and quantities that add up many small independent effects (central limit theorem).

Lognormal

lognormalcontinuous

Positive, right-skewed distribution whose logarithm is normal. Parameters are the mean and standard deviation of the variable itself (not of its logarithm).

KeyParameterDefaultValid values
meanMean100> 0
stdDevStandard deviation30> 0

Typical use: Prices, costs, incomes, real-estate values and other quantities that cannot be negative and grow multiplicatively.

Uniform

uniformcontinuous

Every value between the minimum and the maximum is equally likely.

KeyParameterDefaultValid values
minMinimum80any number
maxMaximum120any number

Typical use: When only the range is known and no value is more plausible than another.

Triangular

triangularcontinuous

Defined by a minimum, a most likely value and a maximum; the density is a triangle.

KeyParameterDefaultValid values
minMinimum80any number
modeMost likely100any number
maxMaximum130any number

Typical use: Expert estimates (pessimistic / most likely / optimistic) of costs, durations or sales when data are scarce.

PERT

pertcontinuous

Smooth beta-shaped alternative to the triangular distribution that gives more weight to the most likely value; mean = (min + 4·mode + max)/6.

KeyParameterDefaultValid values
minMinimum80any number
modeMost likely100any number
maxMaximum130any number

Typical use: Task durations and costs in project management (three-point estimates).

Beta

betacontinuous

Very flexible distribution on a bounded interval [min, max] (by default [0, 1]) controlled by two shape parameters.

KeyParameterDefaultValid values
alphaAlpha (shape 1)2> 0
betaBeta (shape 2)5> 0
minMinimum0any number
maxMaximum1any number

Typical use: Proportions, percentages, probabilities and rates bounded between two limits.

Gamma

gammacontinuous

Positive, right-skewed distribution with shape and scale parameters; mean = shape·scale.

KeyParameterDefaultValid values
shapeShape2> 0
scaleScale10> 0

Typical use: Waiting times, insurance claim sizes, rainfall amounts and other positive skewed quantities.

Exponential

exponentialcontinuous

Time between events of a Poisson process with a constant rate; memoryless. Mean = 1/rate.

KeyParameterDefaultValid values
rateRate0.1> 0

Typical use: Time between arrivals, failures of components with a constant failure rate, service times.

Weibull

weibullcontinuous

Flexible distribution for lifetimes; shape < 1 means decreasing failure rate, = 1 exponential, > 1 wear-out.

KeyParameterDefaultValid values
shapeShape2> 0
scaleScale100> 0
locationLocation0any number

Typical use: Reliability and lifetime of equipment, wind speeds, time to failure.

Logistic

logisticcontinuous

Symmetric bell-shaped distribution similar to the normal but with heavier tails.

KeyParameterDefaultValid values
meanMean100any number
scaleScale5> 0

Typical use: Growth processes, demographic and market-share models, alternative to the normal with more extreme values.

Student's t

studentTcontinuous

Symmetric distribution with heavier tails than the normal; with location (mean) and scale. Approaches the normal as the degrees of freedom grow.

KeyParameterDefaultValid values
dfDegrees of freedom5> 0
meanMean0any number
scaleScale1> 0

Typical use: Financial returns with fat tails, small-sample uncertainty.

Chi-square

chiSquarecontinuous

Sum of squares of independent standard normal variables; mean = degrees of freedom.

KeyParameterDefaultValid values
dfDegrees of freedom5> 0

Typical use: Variances, goodness-of-fit and hypothesis testing.

F (Fisher–Snedecor)

fcontinuous

Ratio of two independent chi-square variables, each divided by its degrees of freedom.

KeyParameterDefaultValid values
df1Numerator degrees of freedom5> 0
df2Denominator degrees of freedom10> 0

Typical use: Ratios of variances, analysis of variance (ANOVA).

Cauchy

cauchycontinuous

Symmetric distribution with extremely heavy tails; its mean and variance are undefined.

KeyParameterDefaultValid values
locationLocation0any number
scaleScale1> 0

Typical use: Stress-testing models with very extreme outcomes; resonance phenomena in physics.

Gumbel (maximum)

gumbelcontinuous

Extreme value distribution of the maximum of many observations; right-skewed.

KeyParameterDefaultValid values
locationLocation100any number
scaleScale10> 0

Typical use: Maximum floods, peak loads, largest daily losses, extreme weather.

Fréchet

frechetcontinuous

Heavy-tailed extreme value distribution (type II), bounded below by the location.

KeyParameterDefaultValid values
shapeShape3> 0
scaleScale100> 0
locationLocation0any number

Typical use: Extreme financial losses, maximum rainfall, catastrophic insurance claims.

Pareto

paretocontinuous

Power-law distribution starting at the scale (minimum value); small shapes produce very heavy tails.

KeyParameterDefaultValid values
shapeShape3> 0
scaleScale (minimum)100> 0

Typical use: Income and wealth distribution, city sizes, large insurance losses (80/20 rule).

Laplace (double exponential)

laplacecontinuous

Symmetric distribution with a sharp peak at the location and exponential tails.

KeyParameterDefaultValid values
locationLocation100any number
scaleScale10> 0

Typical use: Daily price changes, forecast errors, differences between two exponential variables.

Rayleigh

rayleighcontinuous

Magnitude of a two-dimensional vector whose components are independent normals.

KeyParameterDefaultValid values
scaleScale10> 0

Typical use: Wind speeds, wave heights, signal amplitudes.

Erlang

erlangcontinuous

Sum of k independent exponential times with the same rate (gamma with integer shape).

KeyParameterDefaultValid values
kShape (k)3≥ 1, integer
rateRate0.5> 0

Typical use: Total time to complete k sequential tasks, queueing and call-centre models.

Arcsine

arcsinecontinuous

U-shaped distribution on [min, max]: values near the extremes are the most likely.

KeyParameterDefaultValid values
minMinimum0any number
maxMaximum1any number

Typical use: Proportion of time a random walk stays positive; variables that tend to the extremes.

Cosine

cosinecontinuous

Symmetric bell-shaped distribution on a bounded interval, with a cosine density.

KeyParameterDefaultValid values
minMinimum0any number
maxMaximum100any number

Typical use: Bounded symmetric quantities, an alternative to the triangular or the truncated normal.

Power function

powerFunctioncontinuous

Distribution on [min, max] with F(x) = ((x − min)/(max − min))^α; a beta with β = 1.

KeyParameterDefaultValid values
alphaAlpha (shape)2> 0
minMinimum0any number
maxMaximum100any number

Typical use: Bounded quantities skewed towards one end, such as utilisation rates or yields.

Trapezoidal

trapezoidalcontinuous

Density rises linearly from the minimum to the first mode, stays flat up to the second mode and falls to the maximum.

KeyParameterDefaultValid values
minMinimum0any number
mode1Lower mode30any number
mode2Upper mode70any number
maxMaximum100any number

Typical use: Expert estimates where a whole range of values is considered equally likely.

Bernoulli (yes/no)

bernoullidiscrete

Takes the value 1 with probability p and 0 otherwise.

KeyParameterDefaultValid values
pProbability of success0.5[0, 1]

Typical use: Whether an event happens: a risk materialises, a contract is won, a test passes.

Binomial

binomialdiscrete

Number of successes in n independent trials with the same probability of success p.

KeyParameterDefaultValid values
nNumber of trials10≥ 1, integer
pProbability of success0.5[0, 1]

Typical use: Defective units in a batch, customers who buy out of n contacted, approved projects.

Poisson

poissondiscrete

Number of events in a fixed interval when they occur independently at a constant average rate λ.

KeyParameterDefaultValid values
lambdaMean number of events (λ)5> 0

Typical use: Customer arrivals per hour, accidents per month, defects per unit, calls received.

Geometric

geometricdiscrete

Number of failures before the first success in independent trials with probability p.

KeyParameterDefaultValid values
pProbability of success0.2(0, 1]

Typical use: Attempts needed before closing a sale, wells drilled before finding oil.

Negative binomial

negativeBinomialdiscrete

Number of failures before the r-th success in independent trials with probability p.

KeyParameterDefaultValid values
rNumber of successes (r)5> 0
pProbability of success0.5(0, 1]

Typical use: Over-dispersed counts (variance greater than the mean), calls needed to reach r sales.

Hypergeometric

hypergeometricdiscrete

Number of successes in a sample drawn without replacement from a finite population.

KeyParameterDefaultValid values
populationPopulation size50≥ 1, integer
successesSuccesses in the population15≥ 0, integer
drawsSample size (draws)10≥ 0, integer

Typical use: Quality inspection of lots, audits, sampling from a finite list.

Discrete uniform

discreteUniformdiscrete

Every integer between the minimum and the maximum is equally likely.

KeyParameterDefaultValid values
minMinimum1integer
maxMaximum6integer

Typical use: Dice, random choice among numbered options.

Custom

customdiscrete

Defined by your own data: a list of values with weights (discrete) or, without weights, an empirical continuous distribution that interpolates the data.

It has no numeric parameters: it is defined by a list of values and, optionally, their probabilities (weights). Without weights it is a continuous empirical distribution (interpolating the data).

Typical use: Historical data, scenarios with subjective probabilities, any shape not covered by the standard families.

Fixed value

fixeddiscrete

A constant: always takes the same value.

KeyParameterDefaultValid values
valueValue100any number

Typical use: Temporarily removing the uncertainty of an input without deleting the assumption.

Released under the GPL-3.0 license · © 2026 Fernando Daza and contributors