Kraken

Kraken offers spot and derivatives trading across a wide range of digital assets. This integration connects to Kraken Pro and supports live market data and order execution for Kraken Spot and Kraken Derivatives (Futures).

Overview

The adapter is implemented in Rust with Python bindings and does not require an external Kraken client library. Each data or execution configuration selects a Spot or Futures client through its product_type.

The main Python components are:

• KrakenDataClientConfig and KrakenExecClientConfig: Live client configuration.
• KrakenDataClientFactory and KrakenExecutionClientFactory: Factories used by the trading node builder.
• KrakenSpotHttpClient and KrakenFuturesHttpClient: Lower‑level HTTP access for direct requests.
• KrakenSpotWebSocketClient and KrakenFuturesWebSocketClient: Lower‑level WebSocket access.

Examples

• Python examples
• Rust examples

Kraken documentation

Kraken provides detailed documentation for users:

• Kraken API documentation
• Kraken Spot REST API
• Kraken Derivatives API

Refer to the Kraken documentation in conjunction with this NautilusTrader integration guide.

Products

The adapter supports these product categories:

Bar streaming

Supported intervals

The Kraken adapter supports real-time bar (OHLC) streaming for Spot markets via WebSocket. The following intervals are available:

Bar emission latency

Kraken's Spot WebSocket OHLC channel updates the current, incomplete bar on trade events. It does not provide a field that marks a bar as closed.

During normal streaming, the adapter buffers the current bar and emits it after receiving an update with a new interval_begin. The delay therefore depends on the first trade in the next interval and is not bounded to one bar period when a market has no trades. When the WebSocket message handler stops, the adapter flushes its buffered bars, including a current bar that may still be incomplete.

The adapter uses buffering instead of timer‑based emission because:

• Timer‑based emission could miss the final update before the bar closes.
• Kraken's updates are not guaranteed to arrive at exact interval boundaries.

This favors the latest venue update at the cost of latency.

Symbology

Spot symbol normalization

Kraken uses different Bitcoin symbol conventions across their APIs:

Kraken also uses XDG for Dogecoin in some Spot responses. The adapter normalizes it to DOGE, including in quote currency symbols.

Spot markets

NautilusTrader uses normalized, slash‑separated symbols for Kraken Spot instruments. The adapter translates them to Kraken's native format internally.

Instrument ID format:

InstrumentId.from_str("BTC/USD.KRAKEN")  # Spot BTC/USD
InstrumentId.from_str("ETH/USD.KRAKEN")  # Spot ETH/USD
InstrumentId.from_str("SOL/USD.KRAKEN")  # Spot SOL/USD
InstrumentId.from_str("BTC/USDT.KRAKEN")  # Spot BTC/USDT
InstrumentId.from_str("ETH/BTC.KRAKEN")  # Spot ETH/BTC (normalized from ETH/XBT)

Futures markets

Kraken Futures instruments use a specific naming convention with prefixes:

• PI_ - Perpetual Inverse contracts (e.g., PI_XBTUSD)
• PF_ - Perpetual Fixed‑margin contracts (e.g., PF_XBTUSD)
• PV_ - Perpetual Vanilla contracts (e.g., PV_XRPXBT)
• FI_ - Fixed maturity Inverse contracts (e.g., FI_XBTUSD_230929)
• FF_ - Flex futures contracts

Instrument ID format:

InstrumentId.from_str("PI_XBTUSD.KRAKEN")  # Perpetual inverse BTC
InstrumentId.from_str("PI_ETHUSD.KRAKEN")  # Perpetual inverse ETH
InstrumentId.from_str("PF_XBTUSD.KRAKEN")  # Perpetual fixed-margin BTC

Data capability

Subscriptions (real-time)

Requests (historical)

L3 order book (market-by-order)

Kraken exposes Spot per-order book data via the WebSocket v2 level3 channel at wss://ws-l3.kraken.com/v2. This gives venue order IDs, per-order quantities, and true incremental events (add, modify, delete). The adapter hashes each venue order ID into the u64 BookOrder.order_id field used by NautilusTrader.

Prerequisites

L3 subscriptions require Spot API credentials because Kraken's level3 channel is authenticated. Pass them to KrakenDataClientConfig:

from nautilus_trader.adapters.kraken import KrakenDataClientConfig

config = KrakenDataClientConfig(
    api_key="YOUR_KEY",
    api_secret="YOUR_SECRET",
)

Then subscribe with book_type=BookType.L3_MBO:

from nautilus_trader.model import BookType

await client.subscribe_book_deltas(
    instrument_id=instrument_id,
    book_type=BookType.L3_MBO,
    depth=1000,  # valid: 10, 100, 1000
)

Valid depths are 10, 100, and 1000. A depth of 0 uses 1000.

CRC32 checksum validation

By default, the adapter validates the CRC32 checksum on each L3 snapshot and update when Kraken provides one. On mismatch, it emits a Clear delta, clears local L3 state, refreshes the auth token, and resubscribes so Kraken sends a fresh snapshot. To disable validation for benchmarking:

config = KrakenDataClientConfig(
    api_key="...",
    api_secret="...",
    validate_l3_checksum=False,
)

Storage recommendations

OrderBookDelta already carries order_id: u64 in its Arrow schema, so L3 data is stored identically to L2 in the ParquetDataCatalog. L3 generates significantly more events per instrument than L2. Recommended settings:

• Lower chunk size (e.g. chunk_size=50_000) for faster parallel reads.
• Enable zstd compression in catalog config.
• Use per-instrument path partitioning (enabled by default).

Orders capability

Order types

Time in force

Execution instructions

Trigger types

Conditional orders (stop, take-profit, trailing stop) support a trigger price reference on Spot:

Batch operations

Position management

Order querying

Contingent orders

Order routing (Spot)

The Spot execution client routes order submission, modification, cancellation, and batch cancellation through Kraken's authenticated WebSocket v2 trade channel by default. It falls back to REST when the WebSocket is inactive. Set use_ws_trade=False on KrakenExecClientConfig to route these operations through REST.

Order shapes routed via REST

Kraken's Spot WebSocket v2 add_order method supports these shapes, but the adapter routes them through REST:

Mixed‑symbol order lists also use REST because Kraken's WebSocket batch_add request requires one shared symbol. Unsupported trigger references fall back to the REST path, which rejects them locally before sending a request to Kraken.

The per-call params={"use_ws_trade": False} override forces a single command through REST regardless of the configured default. Set it on SubmitOrder, ModifyOrder, CancelOrder, SubmitOrderList, or BatchCancelOrders.

WebSocket request timeout

When a WebSocket round‑trip exceeds ws_request_timeout_secs (default 5), the venue outcome may still be unknown. The dispatcher handles each operation as follows:

• Submit and batch add: emits OrderRejected for each affected order, then sends a best‑effort compensating cancel over the same WebSocket.
• Modify: emits OrderModifyRejected.
• Cancel: emits no rejection event, logs the timeout, and awaits reconciliation.

A delayed venue acceptance can race with the local rejection. WebSocket order updates or the live execution reconciliation engine (open_check_interval_secs) recover divergent state.

WebSocket order-routing options

KrakenExecClientConfig exposes:

Reconciliation

The Kraken adapter provides reconciliation capabilities for both Spot and Futures markets, allowing traders to synchronize their local state with the exchange state at startup or during operation.

Spot reconciliation

Order status reports:

• Open orders: Fetches all currently active orders.
• Closed orders: Fetches historical orders with pagination support.
• Time-bounded queries: Supports filtering by start/end timestamps.

Fill reports:

• Trade history: Fetches execution history with pagination.
• Time-bounded queries: Supports filtering by start/end timestamps.
• All fill types: Market, limit, and conditional order fills.

Margin position reports (when spot_account_type=Margin):

• Open positions: Fetched from POST /0/private/OpenPositions and aggregated by (pair, side) into PositionStatusReport entries.
• Synthetic FLAT cleanup: If the local cache has an open spot margin position that no longer appears on the venue (Kraken omits closed positions from OpenPositions), the adapter emits a synthetic FLAT report on the next position-check tick so the engine reconciles to closed.
• Margin balances: POST /0/private/TradeBalance is called alongside the account‑state refresh; used margin populates MarginBalance.initial, while equity and free margin populate the summary balance (see Spot margin trading).

Futures reconciliation

Order status reports:

• Open orders: Fetches all currently active futures orders.
• Historical orders: Fetches closed and filled orders when open_only=False.
• Order events: Full order lifecycle history via /api/history/v2/orders endpoint.

Fill reports:

• Fill history: Fetches all execution reports.
• Time filtering: Client-side filtering by start/end timestamps (parses RFC3339 timestamps).
• All fill types: Maker and taker fills with fee information.

Position status reports:

• Open positions: Fetches all active futures positions.
• Real-time data: Includes unrealized funding, average price, and position size.

Spot position reports (cash mode)

In cash mode, the Kraken adapter can optionally report wallet balances as position status reports for spot instruments. This feature is disabled by default and must be explicitly enabled via configuration. Margin-mode accounts should leave it disabled and rely on OpenPositions instead (see Spot margin trading).

How it works:

• When enabled, wallet balances are converted to PositionStatusReport objects.
• Positive balances are reported as LONG positions.
• Only instruments matching the configured quote currency are reported (default: USDT).
• This prevents duplicate reports when the same asset is available with multiple quote currencies (e.g., BTC/USD, BTC/USDT, BTC/EUR).

Configuration:

from nautilus_trader.adapters.kraken import KrakenExecClientConfig
from nautilus_trader.model import AccountId
from nautilus_trader.model import TraderId


exec_config = KrakenExecClientConfig(
    trader_id=TraderId.from_str("TRADER-001"),
    account_id=AccountId.from_str("KRAKEN-001"),
    api_key="YOUR_API_KEY",
    api_secret="YOUR_API_SECRET",
    use_spot_position_reports=True,
    spot_positions_quote_currency="USDT",  # Default
)

Spot margin trading

Kraken Spot supports leveraged trading on selected pairs. Per-pair availability and the valid leverage tiers are advertised by Kraken on the instruments endpoint as AssetPairInfo.leverage_buy and leverage_sell; the adapter caches these at instrument-load time and validates the requested tier before order submission. Margin trading is enabled per-execution-client via spot_account_type, with per-order leverage params.

Configuration

from nautilus_trader.adapters.kraken import KrakenExecClientConfig
from nautilus_trader.model import AccountId
from nautilus_trader.model import AccountType
from nautilus_trader.model import TraderId


exec_config = KrakenExecClientConfig(
    trader_id=TraderId.from_str("TRADER-001"),
    account_id=AccountId.from_str("KRAKEN-001"),
    api_key="YOUR_API_KEY",
    api_secret="YOUR_API_SECRET",
    spot_account_type=AccountType.MARGIN,
    default_leverage=3,  # Optional config-level default
    margin_balance_asset="ZGBP",  # Optional summary-display asset
)

margin_balance_asset controls only the denomination of the account-summary metrics returned by Kraken's TradeBalance endpoint (equity, free margin, used margin, etc.). Per-position figures from OpenPositions are always in the traded pair's quote currency.

Per-order leverage

Override the configured default on a single order via params:

order = strategy.order_factory.limit(
    instrument_id=BTC_USD,
    order_side=OrderSide.BUY,
    quantity=Quantity.from_str("0.01"),
    price=Price.from_str("50000.00"),
    params={"leverage": 5},
)

The adapter validates the requested tier against AssetPairInfo.leverage_buy / leverage_sell for the pair before submitting; an invalid tier produces an OrderDenied event and never hits the venue.

Reduce-only

Margin orders can carry reduce_only=True so they reduce an existing position without opening a larger opposite position. The adapter denies cash orders with reduce_only before sending them to Kraken.

Account state

When spot_account_type=Margin, the execution client calls Kraken's TradeBalance endpoint during account refreshes. The live account state uses:

• Equity (e) and free margin (mf) for the balance denominated by margin_balance_asset.
• Used margin (m) for MarginBalance.initial. Maintenance margin is zero because Kraken does not return a separate maintenance‑margin amount.

The lower‑level KrakenSpotHttpClient methods request_margin_metrics() and request_account_state_with_metrics() return the full TradeBalance metrics dictionary for direct consumers. The live execution client does not attach that dictionary to AccountState.info.

Position reconciliation

Open spot margin positions are surfaced via POST /0/private/OpenPositions on each position_check_interval_secs tick. Closed positions on the venue that still appear open in the local cache are reconciled to FLAT on the next sweep. This path is independent of use_spot_position_reports (which is wallet-derived, cash-mode-only).

Funding rates

The adapter receives funding rate data from the Futures ticker WebSocket feed, which provides relative_funding_rate and next_funding_rate_time for perpetual futures.

The interval field on FundingRateUpdate is None for Kraken because the ticker feed does not include a funding interval field and the Kraken API documentation does not specify a fixed funding period.

Rate limiting

Each Kraken HTTP client applies an adapter‑side request throttle. The default is five requests per second and max_requests_per_second can override it. This is a request‑count throttle, not a complete model of Kraken's endpoint costs or account‑tier budgets.

Kraken applies different venue limits to Spot and Futures:

• Spot REST rate limits use a tier‑dependent call counter. Ledger and trade history calls add 2, most other REST calls add 1, and order management uses a separate trading limiter.
• Derivatives rate limits use endpoint costs and separate budgets for /derivatives and /history paths.

The current Spot REST call‑counter limits are:

If the adapter's fixed request rate is too high for the endpoint mix and account tier, Kraken can still reject or throttle requests.

Reconciliation interval guidance

The execution engine's open_check_interval_secs and position_check_interval_secs settings create sustained private REST API load. Short intervals can exhaust Kraken's venue budgets even when the adapter stays below its configured requests‑per‑second throttle.

Use conservative intervals as a starting point, especially for a Spot Starter account:

exec_engine = LiveExecEngineConfig(
    reconciliation=True,
    open_check_interval_secs=30.0,  # Conservative Spot Starter-tier starting point
    position_check_interval_secs=120.0,
)

Tune these values for the account tier, enabled reconciliation checks, and other clients using the same API key. If Kraken returns EAPI:Rate limit exceeded, increase the intervals or reduce max_requests_per_second.

Configuration

The product type for each client is specified via the product_type option.

Data client configuration options

Execution client configuration options

For spot margin, default_leverage applies when an order has no per-order leverage param. margin_balance_asset only changes the TradeBalance summary denomination; per-position figures remain in the pair's quote currency.

Demo environment setup

To test with Kraken Futures demo (paper trading):

1. Sign up at Kraken Futures demo and generate API credentials.
2. Set environment variables with your demo credentials:
   • KRAKEN_FUTURES_DEMO_API_KEY
   • KRAKEN_FUTURES_DEMO_API_SECRET
3. Read the credentials and pass them to KrakenExecClientConfig, then set environment=KrakenEnvironment.DEMO and product_type=KrakenProductType.FUTURES.

The Python examples show the complete demo and live LiveNode configurations.

Production configuration

Use KrakenDataClientConfig with KrakenDataClientFactory, and use KrakenExecClientConfig with KrakenExecutionClientFactory. The Python examples show the complete LiveNode.builder(...) configuration for data and execution clients.

API credentials

Live‑node configuration objects do not read credential environment variables automatically. Pass api_key and api_secret explicitly to KrakenExecClientConfig and, for Spot L3 data, to KrakenDataClientConfig. Public market data does not require credentials.

The lower‑level Python HTTP and WebSocket clients load the following variables when their credential arguments are omitted. Rust applications can use KrakenCredential::from_env_spot() or KrakenCredential::from_env_futures(demo) to load them before constructing live‑node configs.

Authentication errors are reported when a private client connects or performs a private operation. Required permissions depend on the requested data or trading operation.

Contributing