eBay vs auction Rolex prices: how the two markets compare in our data

If you look up a Rolex on eBay and then at a recent auction, the numbers often don't match. That is not a mistake in either. They are different markets with different buyers, fees and record-keeping, and in our sold data they settle at different levels.

This post shows the gap reference by reference, explains what each figure includes, and describes how we combine both in one price range. It describes market data, not a recommendation on where to buy or sell, and is not financial advice.

Our two sold-price sources

• eBay sold listings: 87,061 completed sales over the last three years. eBay's Authenticity Guarantee applies to eligible watches above a price threshold, with partner authenticators inspecting them before they reach the buyer, per eBay's own page.
• Auction results: 20,973 results, some going back to the late 1990s. The auction channel covers traditional houses such as Phillips, Sotheby's and Antiquorum, regional European houses, and online auction platforms such as Grailzee. We record each result as the hammer price plus the buyer's premium where one applies, the amount the buyer actually paid.

A third record, the forum archive of 44,070 posts marked sold, is not part of this comparison. Those figures are the listed price when the post was marked sold, not completed sales, and they never feed our numbers.

The gap, reference by reference

Each published reference page shows the eBay median and the auction median over the last twelve months, where each has at least three sales. Here they are side by side for references with plenty of both.

Across every published reference that shows both medians, 209 in this export, the eBay median is below the auction median for 170 of them, or 81%. Read the six rows above against that share: each reference can sit with the majority or against it.

The gap is a raw comparison. Neither median is adjusted for box and papers, condition or the mix of watches that happened to sell, so part of any gap can come from what sold in each channel rather than from the channel itself.

Where eBay comes out higher

The majority pattern has exceptions, and they are not limited to references with few auction sales. The four below had the eBay median above the auction median when we wrote this post. Compare their columns now, because the order can flip from one export to the next.

When a reference has only a few auction results in twelve months, those few set its auction median, and one strong or weak lot can put it above or below eBay. Check the reference you care about rather than assume the general rule.

Vintage: where auction carries more weight

For vintage references the balance shifts. Auction results make up a larger share of their sales in our data, and the auction record goes back decades.

Compare the two columns: for vintage pieces the medians can sit far apart, in either direction. Vintage pieces vary far more from one watch to the next than modern ones: original dials, case condition and provenance all matter. The data can't tell us how much of the gap comes from differences between the watches sold in each channel, which is one reason vintage ranges tend to carry lower confidence labels.

What each figure includes

The two numbers measure slightly different things, which matters when you compare them.

The buyer's premium is the clearest structural difference. Auction houses publish their rates. Phillips' rate sheet, effective April 2026, sets its watch-auction premium as a share of the hammer price, charged on top of it, in tiers. A buyer at auction pays the hammer price plus that premium; the recorded result includes both. We can't separate how much of the eBay to auction gap comes from the premium and how much from the watches themselves, so we don't claim a cause.

How we combine them in one range

Our headline range is a pre-owned watch with box and papers, sold privately. Both sources feed it.

1. Box and papers are normalised within each channel. We measured how watch-only and partial sets sell against full sets separately on eBay and at auction, and adjust each sale by its own channel's ratio. On eBay, a watch-only sale averages 13% below a full set in our measurements.
2. The channel gap is not removed. We also measured the auction to eBay ratio for full sets. When we tested applying that ratio, it didn't improve accuracy on the tuning year and made errors on held-out auction sales worse. So auction sales keep their own level.
3. eBay carries most of the weight by volume. With many more eBay sales, a busy reference's range is mostly set by eBay. A rare vintage reference may rest mainly on auction results.

The method, including the backtest behind each decision, is on our methodology page.

What this means if you are selling

• Look at both medians. For a Submariner or Datejust, the reference page shows both where they exist.
• Count the fees. The recorded prices are what buyers paid. What a seller keeps is lower in both channels, after eBay's fees or the auction house's seller commission.
• Think about time. An auction runs on the house's calendar and pays out after the sale. eBay can be faster but puts the work of listing, shipping and handling returns on you.
• Get offers too. Dealers price off the same market. Our sell page lets you request offers and compare them with the ranges.

What this means if you are buying

• Price the premium in. At auction, the hammer price is not what you pay. Read the house's published rate sheet before bidding and add the premium and any taxes.
• Use the median, not one result. A single eBay or auction sale can sit well away from the middle. The medians and the range on the reference page smooth that out.
• Check what came with the watch. A low eBay result can be a watch without box or papers, or in rough condition. Our range assumes a full set.

Summary

In our sold data, the eBay median is below the auction median for 170 of the 209 published references that show both, but the order varies by reference. Auction figures include the buyer's premium; eBay figures are completed sale prices. We use both, keep each at its own level, and show you the medians side by side on every reference page that has them.