A wall is a single order-book level holding an outsized resting limit order - far bigger than the levels around it. A bid wall (a large buy order) acts as support: price tends to bounce off it. An ask wall (a large sell order) acts as resistance: price stalls under it. Walls matter because they're visible, committed liquidity - unlike a moving average or a drawn line, a wall is real money someone parked at a price. But some walls are bait (spoofing) and some are exchange-data garbage, so reading them is a skill. Below: what a wall is, how it moves price, how to tell a real wall from a fake one, why walls matter for trading and arbitrage, and how Finder surfaces them across every exchange.

What a wall is

Every order book is a stack of price levels, each with a resting size. Most levels are small and similar. A wall is the level where one order dwarfs the rest - say $800k of bids at one price when the surrounding levels hold $20–40k each.

ASKS · resistance mid ≈ $800k bid wall BIDS · support resting size →
Most levels rest $20–40k; one bid parks ~$800k — a wall. Sellers must clear it before price can fall through, so it reads as support.
  • Bid wall = a big buy order below the current price → support. To push price down through it, sellers must fill the entire wall first.
  • Ask wall = a big sell order above the current price → resistance. To push price up through it, buyers must consume the whole wall.

The key property isn't just absolute size - it's dominance: how much of that side's visible depth the level holds. A $500k level is a wall on a thin alt (where it's 60% of the book) but noise on BTC (where it's 0.5%).

How a wall moves price

A wall is a battle line:

  • It holds → bounce. Price approaches, can't break through the resting size, and reverses. Traders lean on the wall: buy just above a bid wall (support), sell just below an ask wall (resistance).
  • It gets eaten → breakout. If aggressive flow consumes the entire wall, the level that was support flips to resistance (and vice versa). A wall being filled is a real, high-volume event.
  • It gets pulled → vacuum. If the owner cancels the wall before price reaches it, the support/resistance vanishes and price can run into the gap. This is where spoofing lives (below).

Watch what a wall does, not just that it exists. A wall that absorbs repeated hits and stays is genuine support. A wall that vanishes the moment price approaches was never support - it was a signal someone wanted you to see.

Real wall vs fake

Three things masquerade as walls. Learn to filter them:

  • Spoofing. A large order placed to be seen, then pulled before it can fill. It manufactures fake support/resistance to nudge other traders, then disappears. Tell: it sits far from the touch and cancels as price approaches, never actually absorbing flow.
  • Mis-scaled / garbage levels. Exchange data is messy. A contract-multiplier bug or a stale feed can print a level with absurd notional (a "$2-trillion" bid). Naively, that ranks as the #1 wall everywhere. Tell: the notional is physically implausible for the asset.
  • Stale books. An illiquid instrument that stopped ticking still shows its last levels. A "wall" on a dead book isn't live liquidity.

What a real wall looks like: large in absolute terms, dominant over its side's depth, near enough to the price to be actionable, coherent notional (not mis-scaled), and it persists across time / absorbs flow rather than vanishing on approach.

Walls in trading and arbitrage

  • Support / resistance levels. The most direct use - place entries/exits around durable walls, or wait for a wall to break before chasing a move.
  • Executable liquidity. A wall tells you how much size you can actually trade at a price. For a large entry/exit, a bid wall is where you can offload without slippage - crucial for sizing.
  • Arbitrage depth. A spread is only as real as the liquidity behind it. A 3% cross-exchange spread with a thin book closes the instant you size up. A spread backed by a real wall is executable at size. Reading walls is how you separate a tradable spread from a one-lot mirage - the same instinct behind a good spread screener.
  • Reversal signals. A wall appearing, getting eaten, or being pulled are all order-flow events that precede moves.

How Finder detects walls

Finder scans the whole order book of every instrument across 20+ exchanges and flags a level as a wall only when it passes a hybrid test:

  • Absolute floor - the notional clears a per-tier minimum (stablecoins / majors / alts have different floors), so a "wall" is always meaningful money.
  • Dominance - the level holds a large share of its side's visible depth (≈30%+), not just a big number on a deep book.
  • Coherence cap - mis-scaled garbage levels (absurd notional) are dropped before ranking, so a contract-multiplier bug can't rank #1.
  • Distance + depth rank - each wall reports how far it sits from mid (in %) and its position in the book (#3 = third level from best), so you can filter near, actionable walls from deep ones.
  • First-seen tracking - a freshly-appeared wall is flagged "new" (a stronger signal than a long-standing one), and the snapshot refreshes every ~3 seconds, so a pulled wall disappears from the board within seconds.

Walls are then ranked by a composite score (size × dominance × proximity) so the most actionable ones surface first. It's all in the walls section of the web dashboard, filterable by side, exchange, tier and distance.

Example: a bid wall as support

One side of an ASTER/USDT book, mid around $1.252:

best bid:     $1.2500     ~$18k
  …small levels…
BID WALL:     $1.2400     680,000 ASTER ≈ $843,000
              = 41% of visible bid-side depth
              = 0.8% below mid  ·  level #6 in the book

Read it: $843k of committed buy interest 0.8% under price, dominating 41% of the bid side. That's genuine support - to break $1.24, sellers must clear the whole wall. For a trader it's an entry zone. For an arbitrageur it's the executable depth that says a spread into this venue can be sized up to roughly that liquidity before slippage eats it. If that wall vanishes as price drifts down to it, the support was spoofed - and Finder's ~3s refresh would show it dropping off the board.

FAQ - order-book walls

What is a wall in an order book?

A single price level holding a much larger resting limit order than the levels around it. A buy-side wall is support, a sell-side wall is resistance. What makes it a wall is dominance over its side's depth, not just absolute size.

What's the difference between a buy wall and a sell wall?

A buy (bid) wall is a large buy order below price - support, price tends to bounce off it. A sell (ask) wall is a large sell order above price - resistance, price stalls under it.

Are order-book walls reliable?

Only the real ones. Spoof walls (placed to be seen, then pulled), mis-scaled garbage levels and stale-book "walls" all masquerade as liquidity. A real wall is dominant, coherent, near price and persists / absorbs flow rather than vanishing on approach.

How do walls matter for arbitrage?

A spread is only executable if there's liquidity behind it. A wall shows how much size you can actually trade at a price, so it separates a real, sizable spread from a thin one-lot mirage that closes the moment you scale up.

How does Finder find walls?

It scans the whole book on every instrument across 20+ exchanges, keeps only levels that clear an absolute floor and dominate their side's depth, drops mis-scaled garbage, ranks by size × dominance × proximity, and refreshes every ~3s so pulled walls disappear quickly.

This is not investment advice. A wall is committed liquidity at a moment in time, not a guarantee - it can be eaten, pulled or spoofed. Treat a single wall as one input, confirm it persists and absorbs flow, and never assume a visible level is executable without checking it's real.

Related: the pillar Crypto arbitrage guide, cross-exchange arbitrage and the crypto spread screener. Live order-book walls across every exchange - ranked, with side / exchange / distance filters - are in the walls section of the web dashboard.