Liv-ex’s ‘fair value’ methodology
Liv-ex uses regression analysis to map the relationship between prices and critic scores. Taking Margaux as an example below, scores (Wine Advocate) and prices of the last 11 vintages are plotted against each other to create a trend line.
Liv-ex’s ‘fair value’ methodology accounts for the fact that the relationship between price and critic score for most of Bordeaux’s fine wine is not linear. Beyond a certain score, the relationship between price and score is typically exponential. This means that additional points have an increasingly large impact on price. However, additional points for lower scoring wines have little or no impact on price.
The R-squared value is used to measure how close data points are to the fitted regression line. R-squared is always between 0 and 100%. In general, the higher the R-squared, the better the regression line (or ‘trend line’) fits the data. For example, the R-squared in the Margaux model below is 92.2% which represents an extremely close fit. In statistical terms, the R-squared measures how much of the variation in price can be explained by variations in score.
For models that have a high R-squared – generally a value above 50% – the trend line can be used to estimate a ‘fair value’ for a particular wine given its score. The model can also be used to identify potential pricing anomalies in the secondary market. Vintages that lie above the trend line are assumed to be trading at a premium to ‘fair value’, while those below it at a discount. In the example below, the 96-point Margaux 2012 appears to be priced significantly below the trend line and might therefore offer value to buyers.
Similarly, the trend line can be used to approximate a Market Price for an En Primeur wine once it has been scored. In the example below, if Neal Martin awards the 2016 Margaux (En Primeur) a score of 96-98 points, then the trend line suggests a fair value of £4,800 (using the mid-point score of 97).
Margaux: ‘fair value’ analysis
However, buying En Primeur is riskier than buying back vintages in bottle. Therefore, it is expected that En Primeur would be priced at a discount to the trend line, on average. Of course, there will be exceptions to this. For example, a 96-point wine in an exceptional vintage may warrant a ‘vintage premium’ to the price of the same 96-point wine in an average vintage, as this report demonstrates later. In these cases, ‘fair value’ represents a valuable and transparent starting point for subsequent negotiations over that premium.
For some wines such as the second wines of the First Growths, age has a greater influence on price than score does. This may be because buyers are seeking exposure to a particular brand and are prepared to pay a premium based on age.
The charts below illustrate that prices for Carruades Lafite are more closely correlated to their age (88.9%) than their score (22.3%), suggesting that age is the key influencer of price.
Carruades Lafite shows very little correlation between price and score
Carruades Lafite prices are highly correlated to age
 Regression analysis is a statistical process for estimating the relationships among variables, in this case between the Market Price (dependent variable) and Wine Advocate critic score (independent variable).