Prerequisites
Cryptohopper MCP configured in an MCP client — see the setup overview.
Available on all tiers. Orderbook access is not tier-gated beyond exchange coverage — see subscription tiers.
A basic understanding of orderbook structure (bids, asks, depth) — see a practical guide to crypto orderbook data.
Liquidity-wall analysis is most informative on liquid pairs on major exchanges. On thin books, every level looks unusual.
Setup steps
Pick your pair and exchange
Choose a liquid pair on a major crypto exchange. Thin markets produce noisy results.
Issue the core wall-detection prompt
Use the template below. Adjust the pair, exchange, and depth to match your setup.
Review the output
A well-tuned scan returns zero walls (common on calm books) or 1–3 flagged levels. Dozens of flags means the threshold is too loose.
Tune thresholds
Defaults (5× neighbour average, within 2% of mid) are a good starting point. In less liquid markets, loosen to 3×. For execution-relevant walls only, tighten to within 1% of mid.
Check for persistence
A single snapshot cannot distinguish real walls from spoofs. Re-fetch the orderbook after 60 seconds and check whether each flagged wall is still present at the same price and size.
Escalate to cross-exchange
A wall appearing at the same level on multiple exchanges is far more meaningful than a single-venue wall. Fetch the same pair from 2–3 exchanges and compare placements.
Schedule for ongoing monitoring (optional)
Scanning every 10–15 minutes for walls forming or breaking gives an early-warning system. See how to schedule Cryptohopper MCP workflows.
Core prompt template
Using the Cryptohopper MCP, pull the orderbook for BTC/USDT on Binance.
Scan for liquidity walls:
- For each level within 2% of mid-price, compare its size to the average size of its 10 nearest neighbours.
- Flag any level where the size is 5× or more than the neighbour average.
- Report, for each flagged wall:
- Which side (bid / ask)
- Price level
- Size at level (in base asset and in USDT equivalent)
- Size as a multiple of neighbour average
- Distance from mid-price (basis points)
Present walls sorted by distance from mid (closest first).
Explain what each wall likely means (support / resistance).
Persistence check prompt
After 60 seconds, re-fetch the same orderbook. For each wall flagged in the first scan, report whether it is still present at the same price with roughly the same size, or has moved / vanished.
Interpreting what you find
Pattern | Likely meaning |
Large ask wall slightly above mid | Short-term resistance; rejection likely until the wall thins or absorbs |
Large bid wall slightly below mid | Short-term support; buyers defending the level |
Wall that shrinks as price approaches | Real interest — orders being filled or pulled as needed |
Wall static for hours | May be a market-maker's resting edge; may be a spoof — uncertain |
Wall that vanishes when price touches it | Spoof — ignore |
Symmetric walls on both sides at equal distance | Range-bound regime; price likely to oscillate between them |
Ask the agent to categorise each wall and suggest what to watch for — it can make these distinctions with the right prompt.
Cost profile
Action | Call units |
Single orderbook snapshot | 1 |
Persistence check (2 snapshots, 60s apart) | 2 |
Multi-exchange (3 venues + persistence check) | 6 |
10 pairs scanned every 15 min | 40/hr · 640/16-hr day · 3,200/week |
Very cheap at scale — a persistent wall monitor is feasible even on Pioneer's 6,000 calls/week. See rate limits explained.
Troubleshooting
Every scan flags dozens of levels
Threshold is too loose, or the book is genuinely sparse. Tighten the ratio (5× → 8×), narrow the distance from mid (2% → 1%), or pick a more liquid pair. A good wall detector should flag rarely.
No walls ever flag on liquid pairs
Counterintuitive but common — on very liquid pairs like BTC/USDT on Binance, the book is smooth and real walls are rare. Loosen the ratio (5× → 3×) or accept that walls are genuinely rare and signal lives elsewhere.
A flagged wall vanishes during the persistence check
It was a spoof or was filled. Either way, do not treat it as durable support or resistance. Real walls persist across at least a few minutes.
Walls cluster at round numbers that aren't meaningful
Most orderbooks show size clumping at $100 and $1,000 intervals because humans pick round numbers. Flag these only if the size is much larger than typical round-number clustering.
You want to compare walls across venues
Add to the prompt: "For each flagged wall, also fetch the orderbook from Coinbase and Kraken and report whether a similar wall exists at the same price on those venues." Cross-venue confirmation is one of the strongest liquidity signals available.
EXCHANGE_NOT_SUPPORTED on the venue you want
Your tier doesn't include that exchange. Pioneer covers three venues; Explorer covers eight. See supported exchanges.
Walls form during news events and you miss them
Walls form and vanish fast during news. Scheduled 15-minute scans miss 30-second formations. Pair wall detection with news monitoring — see how to build a news-driven research workflow.