Last year I chaired the Classic Wine Rosé blind tasting. As panel chair, I was duly emailed audited spreadsheets of scores of the five judges, retail prices, star ratings and tasting notes by the scarily efficient Celia Gilloway. 94 wines were entered and tasted blind. The published results were 13 four star wines and one 4.5 star stunna from Pulpit Rock, the 2012 vintage made from Pinotage. But in this era of big data with the NSA spending billions of US taxpayers money to track Celia, surely publisher Dominc Ntsele (below) can do more than present the top six wines with certificates, even if its at the Mount Nelson ballroom?
So putting on my applied mathematics cap (a real one from Wits, not a UCT business school visiting professor sans qualifications scam) I thought I’d apply some statistics. After all, probability theory is the language of science and the scientific paradigm of scores out of 100 and chemical analyses is the most popular way of appraising wine. In fact not only is probability theory the language, it also supplies the logic.
94 wines were entered but five had no prices and so were excluded. The 89 remaining were priced from R21 to R120 and were scored blind between 11.7 to 16.8 (average scores out of 20). The two data sets are negatively correlated:
which does not surprise me at all as many co-op wines were submitted and for SA co-ops, incredible quality (high score) is available at a cheap price. But more of that later. The negative correlation means that price is a poor indication of quality as judged by our panel. If the boss is coming for lunch and she drinks rosé, buying the most expensive bottle on the supermarket shelf will not do it.
Using stats, I transformed the RRPs into predicted prices which correlate remarkably well with the scores. What I’m saying is that a producer may ask R120 for her rosé, but after a blind taste, I’d be prepared to pay R80 (say).
This predicted price [PP] is a valuable beast. For starters, we can easily see from the above that there are two populations of rosé: those with a PP less than R58.63 and nine more expensive ones. These nine wines are our First Growth Rosés and are listed below:
Pulpit Rock Brink Family Pinotage Rosé 2012
Solms – Delta Lekkerwijn Rosé 2011
Dieu Donne Rosé 2011
Darling Cellars Classic range Merlot Rosé 2012
Jordan Rosé 2012
Amani Poppy Blush Rosé 2011
De Meye Shiraz Rosé 2011
Lord’s The Wicked Maiden Rosé 2012
Kanu Merlot Rosé NV
Something for the Cape Vintner Classification to consider. Let blind tastings decide the first growth brands. The number nine came from a broken stick regression of predicted prices
Predicted prices may be used as a proxy for scores as they’re in units of Rands which means something to all but the most plutocratic wine consumer. A far more useful aid to shoppers than telling her “this is a 4.5 star Rosé.” Huh? In particular, a value can be easily determined by dividing the predicted price by the RRP. Values larger than one are bargains, of which there are 46 as opposed to rip-offs, of which there are 43.
The ten biggest bargains of the tasting (plus RRP, PP and Score) are:
Darling Cellars Classic range Merlot Rosé 2012 R24.00 R84.65 16.50
Pulpit Rock Brink Family Pinotage Rosé 2012 R30.00 R96.77 16.80
Live a Little Rather Revealing Rosé 2012 R21.00 R52.05 15.40
Dieu Donne Rosé 2011 R38.00 R90.22 16.60
Tangled Tree Moscato Rosé 2012 (Van Loveren) R25.00 R55.12 15.40
Wellington “We Call This Dancing Rosé” 2012 R27.01 R58.84 15.60
Koelenhof Pinotage Rosé 2012 R29.50 R56.40 15.40
Amani Poppy Blush Rosé 2011 R40.00 R72.74 16.40
Kanu Merlot Rosé NV R35.00 R62.64 16.00
Rooiberg Pinotage Rosé 2012 R28.00 R50.02 15.20
So the star of the tasting was a co-op wine: the Darling Cellars Merlot Rosé 2012 with R85 value for R24 with Pulpit Rock in second place.