The Price and Quality of Wine, 2014

The makers of the Vivino app, which allows wine lovers to rate and share reviews of wines, produced their Top 100 lists for 2014. According to their website, over 13 million users rated over 3 million wines. Based on those reviews, they produced lists of the Top 100 reds, whites, sparkling wines, and “Under $50” wines. The lists included the average price reported by their users (another feature included in the app). Naturally, I thought it would be interesting to see how well prices correlated with the quality rankings.

I started with the Under $50 category because, seriously, if I’m going to buy a bottle of wine it’s going to be under $50 unless I’m hosting a Nobel Laureate wine connoisseur, or I’m out for dinner at a nice restaurant on someone else’s dime. Besides, there are WAY too many good wines under $50 to spend more than that for most purposes. Using the ranking score (from 1 to 100), the reported prices have a positive correlation, as one would expect (higher ranked wines have higher prices; lower ranked wines have lower prices), but it’s a pretty weak relationship (0.2075).* This suggests that while a higher price wine may be higher quality, don’t count on it. That ought to make you feel better about grabbing that less expensive bottle for your neighbor’s New Year’s party. The average price of wines on the list was $35.03, which is still higher than you might buy for an evening at home, but the cheapest wine was just $11 (Wild Rock’s The Infamous Goose Sauvignon Blanc Marlborough 2013 at #97) and only one wine hit the $50 cap (Pago De CarraovejasRibera del Duero Crianza Tinto 2009 at #31).

I was going to stop there, but decided to look at the Top 100 red wines as well. Since there was no cap on the prices, one might expect some very expensive, highly rated wines to push the expected correlation. However, it’s quite the opposite. The correlation coefficient between rank and price is a mere 0.0528, which means virtually NO relationship between quality ranking and price. Of course, the standard deviation was much larger relative to the average price (std dev = 803.5; average = $549.30) than it was for the lower priced wines (std dev = 10.53, average = $35.03). The highest priced wine on the list was $5,455 (yes, that’s right, Pétrus’ Pomerol 1982 at #36) while the lowest was just $81 (Concha y Toro’s Don Melchor Cabernet Sauvignon 2009 at #82). So if you’re looking at wines in the $100+ range, there is a good chance that relative prices tell you next-to-nothing about the quality in the bottle.

One thing that may affect the weaker relationship is the context in which the most expensive wines are likely purchased. I would suspect a good percentage of these were purchased in restaurants, where mark-ups can be quite high and varied across establishments. And one wouldn’t expect a large number of the highest priced wines to be purchased, even among the 13 million Vivino users, so there may a good deal of variance in reported prices and quality that is masked in the reporting of averages. Perhaps the good people at Vivino would be willing to share more of the data for a more thorough analysis.

I opted not to take the time to do the exercise for the Top Whites or Top Sparkling Wines. For one, I generally prefer reds. I would hypothesize that the correlation is probably just as low for the whites and sparkling, but I suspect the variance in price would be lower for each of those than for the reds, since reds are generally better for aging and therefore may have some appreciated (or potential) time value built-in that the whites may not. If you decide to check it out for yourself, please post a follow-up in the comments!

So you want to make sure you’re getting a good bottle of wine at a good price? Crowd-sourcing quality using apps like Vivino (or Untappd for craft brews) is likely a much more reliable source than just relying on price. Of course, a knowledgeable friend or local wine seller wouldn’t hurt either.

* Note: I edited the post to make the correlations more intuitive (higher quality, higher price positive correlations) rather than the negative numbers that resulted from the actual ordinal rank score. I also added in the names of mentioned wines along with a link to the wine’s profile on Vivino.com).