Questions, answered plainly

Open buy orders divided by current listings. It answers one question: how many buyers are queued behind every item actually for sale.

A tall wall is expensive to fake, because those orders are real money committed at a real price. Above 50x the site raises a pump alert — at that point the price has a floor under it and very little supply above it.

buy order volume ÷ offer volume · alert above 50x

The share of listed supply that sold in a day. It's the fastest way to tell a busy item from a stagnant one, and it's on every row of the screener.

Four zones: under 5% is dead, 5–15% organic, 20–50% a squeeze, above 50% the market has broken away from its supply entirely.

sold today ÷ offer volume × 100

Open buy orders climbed at least 20% between the two snapshots in your comparison window. An order is money committed at a price, so a queue filling up while supply sits still is accumulation you can read before it reaches the price.

It's the loosest of the three and the easiest to over-read: an item with a handful of orders to begin with clears 20% on a few buyers. It needs at least five orders in the earlier snapshot for that reason, and it's worth confirming against the buy wall.

(buy orders now − buy orders then) ÷ buy orders then ≥ 20%

Every skin has its own float range — a Redline can never be below 0.10, so a 0.17 Redline is far better than 0.17 sounds. The adjusted float rescales the item's float to where it sits inside its own range, so two different skins become comparable.

This is why the calculator's wear bands don't match raw float bands. A 0.26 float inside a 0.25–0.75 range is an adjusted 0.02 — Factory New territory for that skin.

(item float − min float) ÷ (max float − min float) Open the float calculator →

The ratio on its own isn't the alert. A pump alert also wants turnover above 15%, offer volume lower than it was at the start of the comparison window, and a top buy order within 15% of the asking price — a wall nobody is bidding near doesn't count.

The screener always shows every raw metric regardless, so sort by buy wall there if that's what you're hunting.

Every three hours. Each figure on every page belongs to a specific snapshot, so you can always see which refresh a number came from.

Nothing here is live. A spike that starts and ends inside a three-hour window will show up as a single step on the next refresh.

Steam's public market data, plus lowest-offer prices from the larger third-party markets for the cross-market table. No account connection of any kind — no login, no API key, no inventory access, nothing to install.

Almost always because coverage grew. When the tracker starts following more items, every total moves with it, and that isn't a market event.

That's what the per-item charts on the overview page are for: dividing by items tracked makes one day comparable to another with different coverage. Check those before reading a jump as real.

See the per-item charts →

Every Δ badge in the table is measured between the two snapshots you pick there. Widen it and you're comparing against a week ago rather than the previous refresh, which usually makes trends clearer and short spikes disappear.

Each possible outcome's price is multiplied by its probability and by your fee multiplier, then summed. Cost is the ten inputs at their current prices; profit is EV minus cost.

Filler slots dilute it: each filler adds a 10% chance of an outcome worth nothing, which is why cheap fills often look profitable on paper and aren't.

EV = Σ (probability × outcome price × fee multiplier)

It sets how much of your cost the worst possible outcome is allowed to fall below. At 0 the requirement is absolute: even the cheapest outcome must cover the contract, so any contract with a losing branch is rejected.

Raise it to accept contracts where the bad branch loses you a defined percentage in exchange for a better average.

Without it, outcomes are matched at the same wear as your inputs. With it, the output wear is computed from the average float of the ten inputs, then only the wear variants that actually exist are kept — and repriced accordingly.

Some contracts improve dramatically. Others vanish, because the wear the maths produces has no variant for that skin.

Starred items are saved to your account and drive the starred-only filter on the screener. It's the intended way to use a table this size: star what you care about, then filter down to them instead of re-reading the whole thing.

No. Every signal is a description of what already happened in public data. None of it forecasts a price, none of it accounts for your risk, and skins are a volatile, thinly traded market with no protections.

Treat a signal as a reason to look closer, never as a reason to buy.

Some more easily than others. Sales volume can be inflated by wash trading between accounts, which is exactly what the fakeout dashboard is for — hollow volume moves the sales line without moving the price.

Buy walls are the hardest to fake, because the orders are real money at a real price. That's why the buy wall carries the loudest alert on the site.

No. It means the conditions that often precede a move are present. Plenty of squeezes decay quietly, and a buy wall can be pulled at any moment.

The three signals are most useful when they agree — supply draining, orders piling up, price still flat.

Free while in beta, with no paid tier today. There's no referral arrangement with any market the site quotes, and no plan to add one.