eBay Average Sold Price: Why You Should Use the Median Instead

Resellers already do this in their heads. "If there is 1 outlier i scrap that" is the median-minus-outliers rule, done by eye. Here's the version you can put in code.

#Why the average lies

eBay sold data has a long right tail. For almost any search you'll find:

• Bundles and lots ("console + 12 games")
• Sealed or graded copies priced like a different product
• Best Offers shown at list price, which inflate the number (details)
• For-parts sales dragging the low end down

The mean adds everything up, so those few sales move it a lot. The median is just the middle sale, so they barely matter.

#The rule we use

1. Drop known junk: lots, for-parts, box-only listings, and Best Offers whose real price is hidden.
2. Drop statistical outliers with the IQR rule. Q1 is the 25th percentile, Q3 the 75th, IQR = Q3 − Q1. Drop anything below Q1 − 1.5×IQR or above Q3 + 1.5×IQR.
3. Take the median of what's left.
4. Report the count alongside it. Ten clean sales is a price; three is a hint.

This is exactly what the stats block in a SaleHistory response does. You also get the raw items if you'd rather apply your own rules.

#In Python

import statistics

def fair_price(prices):
    s = sorted(prices)
    if len(s) < 4:
        return statistics.median(s) if s else None
    q1, _, q3 = statistics.quantiles(s, n=4)
    iqr = q3 - q1
    kept = [p for p in s if q1 - 1.5 * iqr <= p <= q3 + 1.5 * iqr]
    return statistics.median(kept), len(kept)

prices = [96, 104, 110, 115, 118.5, 121, 124, 130, 139, 145, 389.99, 899]
print(fair_price(prices))   # (119.75, 10)
print(statistics.mean(prices))  # 207.6 — the average is wildly off

#In a spreadsheet

If your sold prices are in B2:B50:

Q1:      =QUARTILE(B2:B50, 1)
Q3:      =QUARTILE(B2:B50, 3)
Fair:    =MEDIAN(FILTER(B2:B50, B2:B50>=Q1-1.5*(Q3-Q1), B2:B50<=Q3+1.5*(Q3-Q1)))

Or skip this entirely with our =SOLDMEDIAN() formula.

#When the average is fine

If you're estimating total revenue from many items, like a whole estate, the average of clean prices is fine, since the extremes even out across hundreds of sales. For pricing one item, use the median.

Seeing a price that's way off? It may not be a real sale at all: cancelled sales, revised prices and shill bids. And if eBay's own suggestion disagrees with your median, here's why.