How CoinValueDB Calculates Coin Values

Raw transaction data is noisy. A single coin can sell for wildly different prices depending on the seller, the listing quality, the buyer's urgency, and dozens of other factors. Our data pipeline is designed to cut through that noise and arrive at reliable, representative values.

1. Identification and Matching

Each transaction is matched to a specific coin in our database using a combination of title parsing, item specifics extraction, and pattern matching. We identify the year, mint mark, denomination, variety, and (where stated) the grade. Transactions that cannot be confidently matched are excluded.

2. Grade Classification

For certified coins, the grade is taken directly from the listing (e.g., "PCGS MS-65"). For raw coins, we use a combination of seller descriptions, photo analysis context clues, and pricing signals to estimate a grade range. Raw coin sales are weighted accordingly to reflect the uncertainty in their grade assignment.

3. Outlier Removal

We filter out transactions that are statistical outliers, such as coins sold well below market (possible errors, lot sales, or damaged coins not disclosed) or well above market (possible shill bidding or misidentified coins). This is done using interquartile range (IQR) filtering within each grade bucket.

4. Aggregation

After filtering, we compute the median sale price for each grade. The median is preferred over the mean because it is more resistant to outliers and better represents what a typical buyer pays. We also track the number of observations (sample size) so users can assess the confidence level of each value.

5. Blending

For the final displayed value, we blend eBay transaction data with PCGS and NGC price guide figures. When we have a large sample of recent eBay sales (10+ in the last 90 days), the eBay median receives the highest weight. When eBay data is sparse, we lean more heavily on PCGS/NGC guide prices. This blending approach ensures every coin has a value estimate, even those that rarely appear for sale.

The condition of a coin is the single most important factor in determining its value. The numismatic industry uses the Sheldon scale, which runs from 1 (barely identifiable) to 70 (perfect). Even a single grade point can mean hundreds or thousands of dollars for scarce coins.

For a detailed explanation of each grade level and what to look for, see our Coin Grading Guide.

Our price tables show values at standard grade points: G-4 (Good), VG-8 (Very Good), F-12 (Fine), VF-20 (Very Fine), EF-40 (Extremely Fine), AU-50 through AU-58 (About Uncirculated), and MS-60 through MS-67 (Mint State). These represent the most commonly traded grade levels and give collectors a clear picture of how value scales with condition.

If you compare our values to the PCGS Price Guide, NGC Price Guide, or the Greysheet, you may notice differences. This is expected and reflects fundamental differences in methodology:

• Transaction-based vs. estimate-based: Traditional guides often set prices based on expert opinion and dealer input. Our values are computed from actual completed sales.
• Broader market coverage: We include eBay sales alongside auction house results, capturing a wider range of transactions including raw coins and lower-grade examples that traditional guides may not emphasize.
• Faster response to market shifts: Because we update daily from eBay, our prices react more quickly to supply and demand changes than guides that update monthly or quarterly.
• Certified vs. raw premiums: PCGS and NGC guide prices reflect certified coin values. Raw coins typically sell for less, and our aggregate includes both, which can pull our figures lower for some coins.

Neither approach is "right" or "wrong." Each serves a different purpose. We believe showing real market prices gives collectors the most actionable information when buying, selling, or insuring their coins.

We take data quality seriously. Our pipeline includes multiple layers of validation, from automated sanity checks to manual review of statistical anomalies. Key measures include:

• Sample size transparency: We show the number of observations behind each value so you can assess confidence.
• Source attribution: Where possible, we indicate whether a value is based primarily on eBay transactions, PCGS guide data, or auction results.
• Continuous improvement: We regularly refine our parsing algorithms, grade classification models, and outlier detection to improve accuracy over time.
• Historical tracking: We maintain price history charts so you can see how values have changed over time, making trends visible rather than hidden.

If you believe a value is incorrect or have data that could improve our coverage, we welcome feedback through our support page.