GOATLIST puts every eBay listing for a set into a bucket by how collected the player is, then shows the price band of each bucket. The buckets are the whole product. This is the short version of how to use them, and why the third step is the one that matters.
The buckets come from a curated index of the players who carry a market in each sport. That index is one opinion. It is a reasonable one, but it is not yours, and value in this hobby is subjective most of the time. A Suns collector rates Booker where the index puts Durant. Someone who grew up on 2003 has Carmelo an Icon and Wade a Star, and someone else has it the other way round. Neither is wrong. The index cannot know which one you are.
The tier decides which bucket every listing lands in, so a search is only as sharp as the list behind it. Ten minutes on the Tier lists board does more than any filter:
IconEliteStarDepth Misc
Tiers are relative to the set you searched, not to all time: the top slice of the players actually present are the Icons. Dwyane Wade is a headline card in 2003-04 Topps and filler in 2007-08 Topps, and the site agrees with both. A player's own rookie year lifts them. Anyone not in the index, and not on one of your lists, is Misc, one bucket you can scan or ignore.
Every priced listing gets a multiple: its asking price divided by the median ask of comparable cards. 1.00× is the going rate. 0.60× is sixty cents on the dollar. 1.60× is a seller hoping. Value sorts by that multiple, lowest first, so the page opens on what is priced under its peers rather than on what is cheap in dollars: a $300 card at 0.5× comes before a $10 card at 1.2×. In the Players and Inserts views each group is ranked by its own median multiple, which is how “who is undervalued in this set” gets an answer.
What counts as comparable, in order of preference:
Hover a multiple to see which basis it used. Under 0.85× is marked in green. Three cautions. A median of asks is not a sale price. A low multiple on a Misc player is a cheap card nobody wants. And a lot, a damaged card or a mislabelled parallel can look like a bargain because the parser grouped it with better company. The multiple says where to look, not what to pay. Watches use the same number: a watch reports a card only at or under the going rate.
Per $ is scarcity bought per dollar. Every numbered card gets a scarcity read between 0 and 1 from its print run, the same one the card score uses, and the sort ranks by asking price plus shipping divided by that read, lowest first. It answers “where does my money buy the most rarity” without pretending one card is worth another: a /25 at $30 comes before a /5 at $900, because the /5 costs thirty times as much for a read that is not thirty times higher.
The read follows the print run on a curve, not a line, so each step down the ladder is worth less than the last: 1/1 is 1.0, /5 .87, /10 .78, /25 .65, /99 .48, /499 .28, and an unnumbered card .08: low, not zero, so it still sorts by price among its own kind. A rung the set marks as chase but never numbered, Fleer Retro's Jambalaya for one, is read as at least .45 rather than as a base card.
Two things it is not. It is not Value: Per $ never looks at what comparable cards ask, so a /10 that is cheap for a /10 and a /10 that is three times the going rate rank on price alone. Read the two sorts together: Per $ finds the rung, Value says whether this copy of it is fairly priced. And it is not for vintage. On a set where every copy came off the same press the print run says nothing, the site scores condition instead, and Per $ collapses to cheapest first. Use Rarest or Value there.
Rarest is the print run, lowest first. A 1/1 leads, then the /5s, the /10s, the /25s and so on down; within one run the cheapest copy, with shipping counted, comes first. Unnumbered cards sort last, not first. Set a budget, sort Rarest, and the top row is the lowest-numbered card you can get for that money, which is the question the sort exists to answer. The Hunt tile on the home page is exactly that: any Ichiro refractor under $100, rarest first.
The number is read off the listing title, as sellers write a serial: 6/24, #d 6/24, /24, 1/1, one of one. Three traps the parser steps around, because a false print run lands exactly where it does the most harm, at the top of this sort. #6 of 24 is card six of a 24-card 1990s insert set, not 24 copies. 13/14 Fleer Retro is a season, not copy 13 of 14, unless the denominator is a run somebody prints, so 49/50 still counts. And a run the title never states is unnumbered here, even when the parallel is limited.
Two things it is not. It is not Per $: Rarest has no curve and no floor, so a chase rung with no printed run, Fleer Retro's Jambalaya again, sinks to the bottom as unnumbered where Per $ lifts it. And on a vintage set nothing carries a serial, so Rarest is price order. To cut rather than order, the min and max print run boxes under Filters drop everything outside a range; Rarest then orders what is left.
Ending is soonest to close first. The Ends column reads the end date eBay publishes for the listing and shows what is left in the coarsest unit that still helps: days and hours for a card ending this week, hours and minutes today, minutes alone when it is on you now. The cell turns blue once a listing is inside a day and red inside an hour: one is a decision you make today, the other one you make now.
Two kinds of listing sink to the bottom rather than float to the top, because both would otherwise lead the page and neither is what “ending soonest” asks for. A listing with no published end date shows a dash and sorts last. That is the normal state of a Buy It Now, since most are good-till-cancelled and eBay prints no end date for them, so a dash is not a gap in the data. And a listing that has already closed shows “ended”, struck through, and sorts last too.
Where it earns its place: auctions. Set Buying to Auctions under Filters, sort Ending, and the page is the day's closes in order. Two cautions. The clock runs against your own time, so a stored answer that is a few hours old will show its short auctions as ended; refresh the search for the live board. And an auction's price is its current bid, not its close, which is why Value and Per $ read an auction generously until its last hour.
Best is the default, and it is two sorts stacked. First the player's tier: every Icon before every Elite, every Elite before every Star, Misc last. Then, inside a tier, the card score, a 0–100 read of how much card this listing is; ties go to the pricier one. So it is “what is the best card on this page”, not “what is the best deal”; the best deal is Value's job. The tier half of it is the half you can change on the Tier lists board.
The card score adds up six signals, each worth at most the points shown, capped at 100: scarcity from the print run, 26; rookie, 26, full marks when the set is the player's rookie year whether or not the seller says so, or the checklist says rookie, and about half when only the title claims it; the player's own market score, 22, which is why a Luka base card still outranks a fringe name's /25 inside one tier; short print or a key card on the checklist, 10; autograph, 8; and price, 8, which gives a little credit to a listing priced above the search's median on the theory that the seller knows what they have. A lot, break or pick-your-card listing has its score cut by more than half so it cannot lead a row.
Hover any row and the tooltip shows its score and the reasons it earned: “Rookie card (this is their rookie set)”, “Numbered /25”, “Autograph”, “Icon-tier name”. On a vintage set, where no card carries a print run, the scarcity points go to condition instead: a PSA 9 outscores a 7, and a 7 the raw copy. The weights live in ScoreWeights in the config, and they are a judgement, not a law; the tier above them is yours to move.
Price ↑ is cheapest first and Price ↓ priciest first, and both sort on the landed price: the asking price plus the shipping eBay quotes, so a $40 card with $12 postage sorts after a $50 card shipped free. The price cell shows the two parts separately, the ask in large type and the postage under it beside the format, so the order can look wrong by a few dollars until you add them. A listing with no price at all sorts last under both arrows.
The number is the ask, and for an auction that means the current bid, with the bid count beside it. An auction at $1 with no bids and four days to run is the cheapest card on the page under Price ↑ and will not stay that way; read it with Ending and the bid count, not on its own. Every price is in US dollars from the US marketplace.
Cutting is different from ordering. The Min $ and Budget $ boxes are sent to eBay with the search and apply to the asking price alone, before shipping, so a $95 card with $10 postage still comes back under a $100 budget. Within what comes back, the Refine rail's price band narrows the page without another call. And cheap is not the same as underpriced: Price ↑ finds the cheapest card, Value finds the one that is cheap for what it is. The two lead with different rows more often than not.