Building a Comparables Set: What "a Good Deal" Actually Means
"A good deal" is a comparison, not a feeling, and every comparison needs something to compare against. Most shoppers reach for whatever's easiest — the seller's own "was" price, or one competing listing they happened to notice — and call that their reference point. A single data point isn't a reference price; it's a guess wearing a reference price's clothes, and it's exactly as easy for a seller to steer as any other number they control. A real comparables set takes a bit more effort to build and is far harder to fool.
What actually counts as a comparable
A comparable is a sold price for the same item in the same or very similar condition, not an active asking price. Active listings show what sellers hope to get, which can sit well above what buyers are actually willing to pay; sold listings show what a real transaction actually settled at. When you're building a comp set, filter out anything that's still listed and unsold, and be careful to match condition — a comp for a mint, complete item tells you very little about what a worn, incomplete one is actually worth, even if it's nominally "the same item."
How many comps you need, and what to do with them
Five comparable sold prices is a reasonable working minimum for most everyday items; fewer than that and a single unusual sale can skew your sense of the market badly. Suppose you've gathered five recent sold prices for an item you're considering: $140, $148, $155, $162, and $171. Averaged, that comp set puts the typical price at $155.20 — and that average, not any single one of the five sales, is your reference price going forward.
It's worth glancing at the spread as well as the average. This set ranges from $140 to $171, a reasonably tight band around the $155.20 average, which gives you some confidence the average is meaningful. A comp set where the five sales ranged from $80 to $300 would tell a very different story — not a stable market price, but a market where condition, completeness, or something else is driving wildly different outcomes, and averaging blindly across it would hide more than it reveals.
Worked example: a discount that doesn't clear the bar
With that $155.20 reference price in hand, take a listing showing a $220 "was" price marked down to $165. Run it through the Deal Savings & Discount Calculator with $220 as the list price, $165 as the sale price, and $155.20 as the reference price you built from the comp set: it shows $55 saved and a 25% discount off the list price — a genuinely impressive-looking markdown — but $9.80 above the comp average, and the calculator flags it plainly as not a real deal. The 25% off was measured against a list price nobody actually pays; measured against what the item genuinely sells for, this is a worse-than-typical price wearing a big discount sticker.
Worked example: when the discount holds up
Now take the same $220 list price marked down further, to $145, and run the identical comparison. This time you're paying $75 less than list — a 34.09% discount — and $10.20 below the $155.20 comp average, a real savings of about 6.6% against what the item actually trades for. The calculator marks this one a genuine deal. Notice that the list price and the shape of the offer were identical in both examples; only the actual sale price relative to the comp average changed the verdict, which is exactly the point of building the comp set in the first place — it's the only number in either example that wasn't chosen by the seller.
Adjusting for condition differences
Real comp sets rarely arrive perfectly matched. If your five sold comparables include some listings in noticeably better or worse condition than the item you're actually considering, don't average them blindly — separate them first. Group comps by condition tier if you have enough of them, and use the average for the tier that actually matches what you're bidding on. A single "great condition" comp mixed in with four "well-worn" ones will pull your reference price upward in a way that misrepresents what the item you're actually looking at is worth.
When the comps genuinely disagree
Sometimes a comp set won't cluster neatly no matter how carefully you filter it — you'll find, say, three sales in the $140s and two sales up near $200, with nothing in between. That's usually a sign the item has more than one identifiable variant hiding under one description — different editions, different included accessories, different authentication status — and averaging across all five would blur a real distinction into a meaningless midpoint. When this happens, dig one level deeper into what's actually different about the higher and lower clusters before you settle on a single reference price; the median of a genuinely single-cluster set is usually more representative than the mean, but no single statistic fixes a comp set that's secretly measuring two different things.
Keep the comp set current
Prices drift, sometimes quickly, especially for anything with active collector or resale interest. A comp set built from sales six months ago can be meaningfully stale for a fast-moving category. Rebuild or at least spot-check your comparables close to when you're actually about to bid, rather than relying on research you did weeks earlier — the five minutes it takes to refresh a comp set is cheap next to bidding on a stale reference price.
Where to actually find sold-price data
Most marketplaces let you filter search results to completed or sold listings specifically, rather than active ones — that filter is the single most useful tool for building a comp set, and it's worth learning where it lives on whichever platform you use most. For items also sold through auctions, look specifically for past-lot results or auction archives, which many auction houses and marketplaces publish. Where a platform doesn't expose sold data directly, a general web search for the item name plus "sold" will often surface listings elsewhere that do. None of this requires a paid tool or special access — it requires knowing that the filter exists and remembering to use it before you get attached to a listing.
Comps for auction items: pick one basis and stick to it
When your comparables come from past auction results rather than fixed-price sales, decide up front whether you're comparing hammer prices or true all-in totals, and use the same basis consistently across every comp and against the listing you're evaluating. Mixing a comp set of hammer prices with a target price that already includes premium and tax will make an item look far cheaper or more expensive than it actually is relative to the market. If you can find the buyer's premium and tax rules for the past auctions in your comp set, converting everything to an all-in basis with the Total Cost-to-Win Estimator gives the fairest comparison; if you can't, hammer-price-to-hammer-price is still useful as long as you're honest that it's missing the fee layer on both sides.
A comp-set checklist
Before you treat any reference price as solid enough to bid against, run through this quickly:
- Are all your comps sold prices, not active asking prices?
- Do you have at least four or five of them, not just one or two?
- Are they reasonably close in condition and completeness to the item you're evaluating — and grouped by tier if they aren't?
- Is the spread between your comps tight enough that an average is meaningful, or does it suggest two different things are being sold under one description?
- Are the comps recent enough to reflect the current market, not several months stale?
- Are you comparing on a consistent basis — hammer price to hammer price, or all-in total to all-in total, not a mix of the two?
A comp set that clears all six checks is a genuine reference price. One that fails even one or two is still better than no comp set at all, but it's worth treating the resulting number as a rougher estimate, not a precise line in the sand.
The habit, not just the math
The arithmetic here — averaging, comparing against a discount, checking the spread — is genuinely simple. The habit that's actually hard to build is doing it every time, before you get excited about a listing, rather than after a discount claim has already done its job on your judgment. For the broader set of tricks a comp set defends against — anchoring, decoy pricing, urgency framing — see spotting a real deal vs a fake discount. This post is the method for building the one number that whole approach depends on: an honest reference price, built from real sold data, not from whatever a seller chose to print with a line through it.