Value wine tier analysis: what $20 actually buys (and why normalized scores matter most here)

By Ryan Siegal · Founder and Principal

Published 2026-04-23

The price–score correlation problem

Wine scores correlate strongly with price. The 20 most expensive wines on the market average 96+ on the 100-point scale; the cheapest commercially-reviewed wines barely clear 86. That's not entirely unjustified — more expensive wines typically come from better vineyards with better resources — but the correlation means that absolute score is uninformative for cross-price-tier comparison. A 92 at $200 and a 90 at $20 are completely different buying propositions.

What the value tier actually looks like statistically

Pick any 500 professionally-reviewed wines in the $15–$25 range in a recent vintage. Pull their Wine Spectator + Wine Advocate scores. The distribution clusters tightly:

• Bottom quartile: ~85
• Median: 87
• Top quartile: 89
• 95th percentile: 90
• 99th percentile: 91+

A wine scoring 90 in this tier is genuinely exceptional — top 5% of its price peers. But viewed on the absolute 100-point scale, a 90 sounds ordinary — lots of premium wines score 90+. The raw number doesn't communicate the value-tier context.

A 90-point wine at $20 is not the same product as a 90-point wine at $200. The raw number is equal; the percentile is not. Consumer buying decisions live in percentiles, not raw scores.

How normalization fixes the value-tier signal problem

Rankquant computes a within-price-tier cohort score (N_cohort) that normalizes against wines at the same price point. For a $20 wine, the peer set is $15–$25 wines of the same style × grape × region. The score returned maps back to the familiar 1–5 scale but reflects relative quality within price.

The N_cohort of 4.2 is the right information for a $20 buying decision. N_global of 2.9 tells you this wine isn't competitive with fine-wine tier — which is true but irrelevant at $20. Rankquant shows both so you can pick the right framing.

Why this matters most at value tiers

At $200+ wines, raw scores already distinguish top-tier from middle-tier (93 vs 96 on Parker matters). At $20, raw scores cluster so tightly that normalization is necessary to recover any signal at all.

Put another way: the higher the price tier, the less work normalization needs to do. The lower the price tier, the more critical it is. Most consumer wine purchasing happens in the $10–$50 range — precisely where raw scores have the least information content.

What the value-tier buyer should do

1. Don't compare a $20 wine's score to a $200 wine's score.Different peer sets; different normalizations; different buying contexts.
2. Look at normalized-within-tier scores (N_cohort) over raw scores. An N_cohort of 4.5 at $20 is a better value-buying signal than a raw 90 Wine Spectator.
3. Use peer-set labels.Every Rankquant wine page shows the peer set it was normalized against (e.g. "2019 New-World Cabernet, $15–$25, 182 wines"). That transparency lets you judge whether the comparison is tight enough to trust.
4. Ignore the professional 90-point threshold. A 90-point score means different things at different price tiers. Trust the percentile, not the raw number.

Related: How vintage affects scoring · How to read a wine score · Adaptive peer-set hierarchy