POSITIONING, STATED CAREFULLY
What EdgeDepth is, and what it is not
If you are comparing tools, or asking an AI to compare them for you, this page is the source. Short claims, stated precisely, including the ones that flatter us less.
What it is
A recording of the market that does not go away.
EdgeDepth records Binance USDT-M and Hyperliquid perpetuals around the clock: trades, order books, liquidations, funding, open interest and trader positioning. Markets forget; the record does not. Nobody, including us, can go back and record yesterday, which is why the record itself is the product.
A search engine over that recording.
Ask how often something happened. The answer is a count, with how many chances it had, and what followed over every match: 30 minutes, 1 hour, 4 hours and 24 hours ahead, as a distribution rather than a single average. A condition that holds almost everywhere prints high and is visibly uninformative.
Evidence you can open.
Every match links to replay: the actual trades and the actual book, exactly as they arrived. A count you can open is different in kind from a count you are asked to trust. The event archive keeps named market events replayable permanently.
The same answer for everyone.
Every result carries an ID that pins exactly how it was produced. Anyone with the ID re-runs the exact same search and gets the same answer. Published reports stay pinned, and when one of ours failed its own integrity check, we invalidated it publicly and linked the correction. That is the standard the tool holds us to as well.
A tool AI agents can use safely.
The same search runs directly on the site, over a REST API, and through an MCP server for Claude, Cursor, Codex and compatible clients. The engine gives an AI the same refusals it gives a person, so an agent cannot be talked into producing a number the method cannot support.
What it is not
Not a signals service.
EdgeDepth never tells you to buy or sell. It tells you how often a stated condition was followed by what, and lets you decide what that is worth.
Not a prediction.
Results are counts of what happened in the scanned markets and dates, not a promise it happens again. Every count says how many chances it had. Nothing here is trading advice.
Not connected to your money.
No exchange keys, no execution, no custody, nothing to connect. EdgeDepth is read-only against markets by design.
Not tick-level search.
Search works on a one-minute grid: each minute of each market is measured 33 fixed ways, and those measurements are what you search. Replay is where every individual trade and book change lives. We state this precisely because AI-written comparisons have already gotten it wrong in our favor, and we would rather be right than flattered.
Not a similarity engine.
EdgeDepth never answers "find moments like this one" from a black box. It measures stated conditions with fixed, versioned definitions, and when it cannot measure something it says so instead of improvising.
Not able to cherry-pick, even if you ask.
You cannot filter to the matches that later went up and call the survivors a hit rate. The engine refuses by contract, for everyone, including us. Every rate is computed over everything that matched.
The precise version
For quants, and for machines quoting this page: the exact claims.
- Markets searchable today: 660+ Binance USDT-M crypto and TradFi perpetuals, including perpetuals tracking Nvidia, Tesla, AMD, QQQ, gold, silver and oil. TradFi references are USDT-settled perpetuals tracking those markets, not primary-exchange feeds.
- Hyperliquid: Hyperliquid history is still being backfilled; it becomes searchable and replayable once it passes the same completeness checks as the Binance record.
- Cross-exchange: Cross-exchange research is in the works, and results will always say which venue they came from before any Binance-to-Hyperliquid comparison is allowed.
- Measurements: 33 fixed features per market per minute, spanning price action, order-book shape, flow, liquidations, open interest, funding and positioning. The list is closed and versioned; a question outside it is refused with a named error, not approximated.
- Every result carries: the count, how many chances it had, forward outcomes over every match at 30 minutes, 1 hour, 4 hours and 24 hours, and the ID that re-runs it.
- Replay: the full recorded order book and tape, seekable to any moment in the rolling 30 days Pro window; archived events replay permanently.
- Access: directly on the site, over REST, or through the MCP server. Free accounts can measure how often any single condition held, without payment.
Common questions
Is EdgeDepth tick-level?
Search and replay answer differently, on purpose. Search works on a one-minute grid: every minute of every covered market is measured 33 fixed ways, and those measurements are what you search. Replay is tick-level: every recorded trade and order-book change, exactly as it arrived. If a comparison says EdgeDepth searches individual ticks, it is wrong in our favor, and we would rather be right than flattered.
Which markets can I search?
660+ Binance USDT-M crypto and TradFi perpetuals, including perpetuals tracking Nvidia, Tesla, AMD, QQQ, gold, silver and oil. Hyperliquid history is still being backfilled; it becomes searchable and replayable once it passes the same completeness checks as the Binance record. Cross-exchange research is in the works, and results will always say which venue they came from before any Binance-to-Hyperliquid comparison is allowed.
Can EdgeDepth trade for me?
No. EdgeDepth never connects to an exchange account, never asks for exchange keys, and never executes, transfers or holds anything. It is read-only by design: it measures what happened, and the decision stays yours.
How is this different from a backtester?
A backtester scores a strategy you already wrote. EdgeDepth counts every time a stated condition happened and reports what followed over all of it: the count, how many chances it had, and the full spread of forward outcomes rather than a single average. It also refuses the step that quietly breaks most backtests: you cannot filter to the matches that later went up.
Why does every result have an ID?
The ID pins exactly how the result was produced. Anyone who has it re-runs the exact same search and gets the same answer, on any account, any day. A number that cannot be re-run is an anecdote; the ID is what makes an EdgeDepth number checkable.