Data Sources
This page shows what powers the DepthSignal platform: the market data we collect, how we process it, and how the combined intelligence surface stays grounded in visible inputs. You can inspect the exchange coverage, derived context, and compliance commitments without guessing where the numbers come from. The same feeds support chart context, pressure context, and organized technical-analysis context.
Derivatives coverage is documented. Availability depends on enabled analytics, plan scope and current source data; collecting inputs alone does not establish customer access.
Customer API-key access is documented. Issuance requires enabled API access and an eligible account; signup or plan selection does not issue a key.
Exchange Coverage
We collect public market data from the following exchanges and normalize it into one shared trader platform.
Live = currently connected and reporting. Supported = in the supported registry but not confirmed reporting right now. Treat its metrics as unverified until Live. Feed freshness per exchange is reported on /status.
Centralized Exchanges
| Exchange | Data Type | Connection | Status |
|---|---|---|---|
| Binance | Spot | REST | Supported |
| Bitfinex | Spot | REST | Supported |
| Bitstamp | Spot | REST | Supported |
| Bybit | Spot | REST | Supported |
| Coinbase | Spot | REST | Supported |
| Gate.io | Spot | REST | Supported |
| Gemini | Spot | REST | Supported |
| HTX | Spot | REST | Supported |
| Kraken | Spot | REST | Supported |
| KuCoin | Spot | REST | Supported |
| MEXC | Spot | REST | Supported |
| OKX | Spot | REST | Supported |
Futures Venues
| Exchange | Data Type | Connection | Status |
|---|---|---|---|
| No futures venues are reported in this coverage snapshot. | |||
Decentralized Exchanges (CLOB)
| Exchange | Data Type | Connection | Status |
|---|---|---|---|
| No decentralized exchanges are reported in this coverage snapshot. | |||
What We Collect
All data is collected exclusively via public, unauthenticated API endpoints. We do not use private endpoints, authenticated sessions, or any data that requires exchange account credentials.
Public orderbook snapshots
Public bid and ask depth around the current price. Shows visible supply and demand without redistributing raw books.
Public trade streams
Aggregated trade data used to compute Cumulative Volume Delta (CVD), taker buy/sell ratio, and positioning context.
Public liquidation feeds
Public liquidation data from supported venues, used to compute cascade stress and long/short imbalance.
Public funding rates and open interest
Derivatives data from Binance, Bybit, and OKX. Used to compute funding consensus, open interest changes, and long/short ratios.
Derived chart and context
Normalized venue price series support chart context, market-pressure summaries, and organized technical-analysis context.
Collection frequency
Data freshness varies by venue and transport. Where freshness metadata is available, use it to distinguish current, stale, and unknown inputs before interpreting derived values.
How We Process It
Raw public data goes through proprietary computation to produce 49 market-context features. The examples below show the kind of processing we apply.
Raw inputs are transformed into readable market context, pressure summaries, and structured TA context. Internal design and computation details are not disclosed.
The platform derives chart structure, pressure context, and TA context from the same normalized market feed so visible price action stays aligned.
Zone Memory Model
Support and resistance zones are computed and time-decayed using a proprietary zone memory model so traders can read structure without sorting raw feed noise. Specific parameters and decay functions are not disclosed.
Cross-Exchange Analysis
The current feed count is unavailable because the status endpoint did not provide a valid freshness summary. Exchange coverage above is a separate catalog view. Cross-exchange features help traders compare pressure and structure across markets, including flow relationships, depth divergence, and large-participant agreement.
5 Composite Intelligence Reads
Our highest-level outputs combine multiple analysis features into simple public-facing reads:
- Directional Pressure: Composite market-pressure read derived from multiple orderbook measurements.
- Liquidity Risk: Proprietary composite measuring current market liquidity conditions.
- Large-Order Activity: Proprietary composite tracking large-participant positioning.
- Flow Momentum: Proprietary composite tracking the acceleration of buying and selling activity.
- Cross-Exchange Agreement: Proprietary composite measuring agreement across supported venues.
What We Sell
We turn raw market data into readable market context.
DepthSignal feature responses contain scalar values or null: a single computed number per feature, per exchange. For example, a directional-pressure read of 0.42 or a price-impact read of 0.0034. A value of null means the metric could not be computed for that observation (warmup or missing evidence); it is not a zero reading. List-structured endpoints (support/resistance zones, historical snapshots) are documented separately.
Our API responses do not contain:
- Raw bid/ask arrays or orderbook depth snapshots
- Raw trade records or tick-level data
- Raw liquidation event streams
- Raw funding rate histories
Every value returned by our API is the result of proprietary computation applied to public data. The raw inputs are never exposed, forwarded, or redistributed in any form.
Example API response structure
{
"symbol": "BTC/USDT",
"features": {
"binance": {
"bid_ask_spread": 0.00012,
"depth_imbalance": 0.15,
"order_flow_imbalance_proxy": 0.42,
"price_impact": 0.0034,
"spoof_score": 0.08
}
}
}Note: Feature values are scalars (single numbers) or null when unavailable. Multi-level metrics such as Level OFI are published as separate scalar keys (level_ofi_1…level_ofi_5). No arrays of bids, asks, trades, or raw market data are present. Feed freshness is reported separately by /v1/status.
Our Commitment to Exchanges
We respect exchange Terms of Service
Our data collection practices are designed to comply with the public API terms of each exchange we connect to. We monitor rate limits, adhere to usage policies, and adjust our collection patterns as terms are updated.
We do not redistribute raw data
No raw orderbook snapshots, trade records, or market data streams are stored for redistribution or made available through our API. All outputs are the result of meaningful proprietary transformation.
We transform data through proprietary computation
Our 49 analysis features are the result of substantial engineering work and proprietary algorithms. The distance between raw input and computed output is intentional and irreversible.
Per-exchange compliance controls
Our architecture supports disabling data collection from any individual exchange instantly, without affecting the rest of the system. If an exchange contacts us with concerns, we can respond immediately.
If you represent an exchange and have questions about our data practices, please contact us at compliance@depthsignal.io. We aim to maintain a cooperative relationship with every venue we connect to and will work with you to address any concerns.
Data Freshness
The /v1/status endpoint reports per-venue connection and freshness fields. Its versioned summary reports counts and venue names; per-venue ages come from exchanges_live.
"market_data": {
"exchanges_live": {
"binance": {
"connected": true,
"last_data_age_s": 2,
"stale": false,
"freshness_status": "current"
}
},
"data_freshness": {
"exchange_feeds": {
"contract_version": 1,
"current_count": 1,
"stale_count": 0,
"unknown_count": 0,
"current_exchanges": ["binance"],
"stale_exchanges": [],
"unknown_exchanges": []
}
}
}Freshness status: The producer classifies each feed as current, stale, or unknown. A missing age is unknown, not evidence of a current feed.
Current feed count: A feed counts as current only when its status is current, it is connected, it is not stale, and its age is a finite non-negative number. If the endpoint or versioned summary is unavailable or invalid, the count is unknown.
Symbol Coverage
DepthSignal covers 52 customer-facing USDT pairs, including:
Coverage varies by exchange and data type. Use the /v1/symbols endpoint for real-time symbol availability, including which data types (orderbook, liquidations, derivatives, spot-futures, CVD, positioning, and large trades) are available for each pair.
Market-Context Analytics
Beyond standard analysis features, DepthSignal computes higher-order market-context analytics from public market data using proprietary statistical models. The public surface describes inputs, transformed context, and limitations; internal training and model structure are not disclosed. These outputs are informational only and are not trading signals, price predictions, or advice.
Proprietary Regime Classification (describing whether the market is trending, ranging, or volatile)
Our proprietary model classifies current market conditions into discrete regimes. Internal classification design and training details are not disclosed.
Statistical Drift Detection (noticing when market behaviour changes structurally)
Proprietary drift detection identifies structural changes in market microstructure in real time. Specific statistical methods are not disclosed.
Ensemble Consensus (combining internal components into one market-pressure view)
Multiple internal model components may contribute to a market-pressure consensus score. Model count, architecture, and aggregation methodology are proprietary.
Method Boundaries: Not Financial Advice
All market-context analytics produce purely computed statistical outputs. They are not instructions to trade, investment advice, or recommendations. Models are updated as part of internal maintenance, not as a public promise of prediction or performance.
This page reflects our current data practices as of October 2026. Our data collection and processing methods may evolve as we add exchanges, expand coverage, and refine our analytics. For questions about our data practices, contact compliance@depthsignal.io.