AQIF key terms

A measure of how spread out a set of returns is around its average, computed as the square root of the variance. It is the most widely used proxy for total risk in finance.

The number of standard deviations a value lies from the mean of its distribution. Investors use it to compare observations measured on different scales and to standardise signals before combining them.

A statistic measuring how two variables move together: positive when they tend to rise and fall in tandem, negative when they move in opposite directions. It underpins correlation and portfolio variance calculations.

A technique that draws many random scenarios from assumed probability distributions to approximate the range of possible outcomes, such as portfolio returns, funding shortfalls or derivative payoffs.

The probability, assuming the null hypothesis is true, of observing a result at least as extreme as the one obtained. Small values suggest the finding is unlikely to be pure chance.

The correlation of a time series with its own lagged values. Positive autocorrelation means each value tends to resemble the previous one, which matters for trend signals and risk forecasts.

A condition in which the variance of a variable, or of regression errors, changes across observations or over time. It can invalidate standard errors and distort statistical inference.

The gap between the highest price a buyer will pay and the lowest price a seller will accept. It is a key gauge of market liquidity and an implicit cost of trading.

A measure of a bond's price sensitivity to interest-rate changes, expressed roughly in years. Higher duration means larger price swings when yields move.

A plot of interest rates across maturities for bonds of similar credit quality. Its shape, whether upward-sloping, flat or inverted, conveys expectations about growth, inflation and future policy rates.

A standardised agreement to buy or sell an asset at a set price on a future date, cleared through an exchange. Positions are marked to market daily against posted margin.

The extra yield a bond offers over a comparable government security, compensating investors for default and liquidity risk. Widening spreads generally signal rising credit stress.

Fitting a model so closely to the training data that it captures noise rather than genuine structure, producing poor performance on new, unseen data.

A penalty added to a model's objective that discourages large coefficients, trading a little training accuracy for better generalisation. Common forms are L1 (lasso) and L2 (ridge) penalties.

A resampling procedure that splits data into folds, trains on some and validates on others, then averages the results to estimate out-of-sample performance more reliably.

The tension between a model too simple to capture real patterns (high bias) and one so flexible that it fits noise (high variance). Total error is minimised near a balance point between them.

A ranking of how much each input variable contributes to a model's predictions. It aids interpretation and variable selection but can mislead when inputs are strongly correlated with one another.

A dimensionality-reduction method that transforms correlated variables into a smaller set of uncorrelated components, ordered by the amount of variance each one explains.

Excess return above the risk-free rate divided by return volatility. It expresses risk-adjusted performance as the reward earned per unit of total risk taken.

An estimate of the loss a portfolio is unlikely to exceed over a given horizon at a stated confidence level, such as a one-day 95% VaR.

The sensitivity of an asset's returns to movements in a benchmark. A beta above one implies the asset has historically amplified market moves; below one implies it has dampened them.

The largest peak-to-trough decline in a portfolio's value over a period, usually quoted as a percentage. It shows the worst loss an investor would have had to endure.

The standard deviation of the difference between a portfolio's returns and those of its benchmark. Low tracking error indicates closely index-like behaviour; high values indicate active deviation.

A rule-generated value or condition indicating whether to buy, sell or hold an asset, typically derived from price, fundamental or alternative data.

The tendency for assets that have performed well recently to keep outperforming, and for recent laggards to keep lagging, over horizons of roughly months to a year.

The tendency of a price or spread to move back toward a long-run average after deviating from it, a pattern exploited by contrarian trading strategies.

The rate at which a portfolio's holdings are replaced over a period. High turnover raises trading costs and can erode an otherwise attractive strategy's net returns.

Scaling position sizes so that the portfolio's expected volatility stays near a chosen level, cutting exposure when markets are turbulent and raising it when calm.

An error in which a backtest uses information that would not have been available at the time of the simulated decision, inflating the apparent performance of the strategy.

Distortion caused by analysing only assets or funds that still exist, since failures and delistings are excluded, making historical results look better than they really were.

Searching many strategies or parameter settings on the same dataset until one looks good by chance, producing results that are unlikely to repeat out of sample.

A validation approach that repeatedly trains on a rolling window of past data and tests on the immediately following period, mimicking how a strategy would actually be deployed.

Evaluation of a model or strategy on data that was not used in its construction, providing an honest check of whether its performance generalises.

The obligation to act solely in the client's best interest, placing the client's interests ahead of the firm's or the professional's own when managing money.

Global Investment Performance Standards: a voluntary, ethics-based set of rules for calculating and presenting investment performance so that results are comparable across firms.

Placing orders with no intention of executing them in order to create a false impression of supply or demand. It is a form of market manipulation prohibited in most jurisdictions.

Information that could reasonably affect a security's price and is not yet publicly available. Trading on it, or tipping others who do, is generally illegal insider trading.

The risk of loss or harm from decisions based on a model that is wrong, misapplied, or built on flawed assumptions, poor data or outdated calibration.