Sign up

Quant Glossary: 64 Quantitative Finance and Backtesting Terms Explained

Explore 64 quantitative finance and backtesting terms across six categories: Risk & performance, Returns & benchmarks, Statistics & estimation, Factors & portfolios, Research & validation, and Time series. Each term has its own page with a clear definition for understanding quantitative research methods and results.

64 terms

Risk and performance terms (11)

Includes Deflated Sharpe ratio, Downside deviation, and Expected shortfall.

  • Deflated Sharpe ratio

    The deflated Sharpe ratio adjusts the interpretation of an observed Sharpe ratio for non-normal returns.

  • Downside deviation

    Downside deviation measures the size of returns below a stated target by taking a lower partial second moment.

  • Expected shortfall

    Expected shortfall at confidence level alpha is the average loss in the worst tail beyond that confidence level.

  • Kelly sizing

    Kelly sizing chooses the fraction of wealth that maximizes expected logarithmic wealth growth under a specified repeated-bet model.

  • Loss quantile

    A loss quantile at probability alpha is a value q such that the cumulative probability of loss at or below q reaches alpha.

  • Probabilistic Sharpe ratio

    The probabilistic Sharpe ratio is an approximate frequentist confidence score for a population Sharpe exceeding the benchmark.

  • Sharpe ratio

    The Sharpe ratio is the mean excess return divided by the standard deviation of excess return, using one consistent return period.

  • Sharpe sampling uncertainty

    Sharpe sampling uncertainty is the variation in an estimated Sharpe ratio across repeated samples under a specified return process.

  • Sortino ratio

    The Sortino ratio divides mean return above a stated target by downside deviation relative to that target.

  • Tail conditional expectation

    Tail conditional expectation is the expected loss in a specified upper tail.

  • Value at risk

    Value at risk at confidence level alpha is a loss quantile.

Returns and benchmarks terms (5)

Includes Annualization, Benchmark subtraction, and Excess return.

  • Annualization

    Annualization converts a statistic measured over one period to an annual scale using a stated compounding or aggregation convention.

  • Benchmark subtraction

    Benchmark subtraction forms an excess or active quantity by subtracting a specified benchmark from an observed return.

  • Excess return

    Excess return is an asset or portfolio return minus a defined reference return over the same horizon.

  • Risk-free benchmark

    A risk-free benchmark is a specified return used as the reference for time value and excess returns.

  • Transaction costs

    Transaction costs are the resources consumed when changing a position.

Statistics and estimation terms (18)

Includes Bootstrap consistency, Bootstrap sampling, and Bootstrap variance.

  • Bootstrap consistency

    Bootstrap consistency means the bootstrap distribution approximates the statistic's sampling distribution as sample size grows.

  • Bootstrap sampling

    Bootstrap sampling resamples observations from the observed sample with replacement to create many same-size pseudo-samples.

  • Bootstrap variance

    Bootstrap variance is the sample variance of a statistic computed across bootstrap replicates.

  • Central limit theorem

    The central limit theorem states that a centered and scaled sample mean approaches a normal distribution as sample size grows.

  • Convergence

    Convergence describes an estimator or sequence approaching a target as sample size, simulation count or another index grows.

  • Estimator

    An estimator is a rule that maps observed data to an estimate of an unknown parameter or functional.

  • Independent and identically distributed sampling

    Independent and identically distributed sampling means observations share a distribution and do not provide information about one another.

  • Kurtosis

    Kurtosis is the standardized fourth central moment of a distribution.

  • Law of large numbers

    The law of large numbers says that an average approaches its expected value as the number of observations grows.

  • Monte Carlo simulation

    Monte Carlo simulation estimates a quantity by repeatedly drawing outcomes from a specified probabilistic model.

  • Monte Carlo standard error

    Monte Carlo standard error measures the simulation noise in an estimate formed from a finite number of simulated outcomes.

  • Sample mean

    The sample mean is the arithmetic average of n observed values.

  • Sample variance

    Sample variance is the average squared deviation from the sample mean with a degrees-of-freedom correction of n minus 1.

  • Skewness

    Skewness measures asymmetry of a distribution around its mean using a standardized third central moment.

  • Standard errors

    A standard error is the sampling standard deviation; an estimated standard error estimates it.

  • Unbiasedness

    An estimator is unbiased for a parameter when its expected value over repeated samples equals that parameter.

  • Units and dimensional analysis

    Dimensional analysis checks that an equation combines and compares quantities with compatible units.

  • Variance calculation

    Variance is the average squared deviation from a mean, measuring dispersion in squared units.

Factors and portfolios terms (11)

Includes Boundary assumptions, Constrained optimization, and Covariance estimation.

  • Boundary assumptions

    Boundary assumptions specify the limits, domains and edge conditions under which a model or derivation is intended to hold.

  • Constrained optimization

    Constrained optimization chooses a decision that optimizes an objective while satisfying equalities, inequalities or bounds.

  • Covariance estimation

    Covariance estimation infers the joint co-movement matrix of asset returns from a finite sample or model.

  • Equal-weight portfolio (1/N)

    An equal-weight portfolio assigns the same weight to each of N included assets, so each weight is 1 divided by N.

  • Factor identification

    Factor identification uses assumptions and variation to distinguish a factor exposure or effect from competing explanations.

  • Factor investing

    Factor investing groups securities or portfolio returns by measurable characteristics or common return drivers.

  • Factor regression

    A factor regression relates an asset or portfolio excess return to one or more factor returns.

  • Fama-French factors

    Fama-French factors are traded or constructed return series used to describe systematic variation in asset returns.

  • Market impact

    Market impact is the price movement caused by an order or trading program, beyond the price that would have prevailed without that trade.

  • Mean-variance portfolio

    A mean-variance portfolio chooses weights to trade off the portfolio's mean against variance under stated constraints.

  • Objective function

    An objective function is the scalar quantity an optimization procedure minimizes or maximizes over an allowed decision set.

Research and validation terms (10)

Includes Confidence intervals, Effective number of trials, and Look-ahead bias.

  • Confidence intervals

    A confidence interval is a random interval procedure designed to cover a fixed parameter at a stated long-run frequency.

  • Effective number of trials

    The effective number of trials summarizes how many independent or weakly dependent opportunities a selection process resembles.

  • Look-ahead bias

    Look-ahead bias occurs when a backtest or analysis uses information that would not have been available at the decision time.

  • Multiple testing

    Multiple testing occurs when many hypotheses or candidate specifications are evaluated and attention is focused on selected results.

  • Out-of-sample validation

    Out-of-sample validation evaluates a model or rule on observations that were not used to select its specification or parameters.

  • Overfitting

    Overfitting occurs when a model adapts to noise or quirks in its development sample as though they were stable structure.

  • Replication versus illustration

    A replication attempts to reproduce a study's data, sample construction, method and reported result closely enough to test reproducibility.

  • Selection bias

    Selection bias occurs when inclusion of observations, assets, studies or results is related to the quantity being measured.

  • Survivorship bias

    Survivorship bias arises when observable entities remain in the sample while disappeared, failed or delisted entities are omitted.

  • Synthetic examples

    A synthetic example uses deliberately chosen inputs to illustrate a definition, calculation or failure mode.

Time series terms (9)

Includes Cointegration, Cointegration residual, and Dependence and autocorrelation.

  • Cointegration

    Two nonstationary series are cointegrated when some nonzero linear combination of them is stationary.

  • Cointegration residual

    A cointegration residual is the estimated deviation from a long-run level relation between nonstationary series.

  • Dependence and autocorrelation

    Dependence means observations are statistically related, so the joint distribution does not factor into independent pieces.

  • GARCH

    A GARCH model describes conditional variance as a function of a constant, recent squared innovations and its own recent variance.

  • Heteroskedasticity

    Heteroskedasticity means the conditional or unconditional variance of an error or response is not constant across observations or states.

  • Momentum

    Momentum is a return pattern in which assets that have recently performed relatively well or poorly are grouped by past performance.

  • Momentum crashes

    A momentum crash is a sharp adverse episode for a momentum portfolio.

  • Stationarity

    A strictly stationary process has a joint distribution unchanged by shifting time.

  • Volatility clustering

    Volatility clustering is the empirical tendency for large absolute returns to be followed by more large absolute returns.