Huobi Futures

Huobi Futures historical data for all it's instruments is available since 2019-11-19.

​

Downloadable CSV files

Historical CSV datasets for the first day of each month are available to download without API key. See downloadable CSV files documentation.

​

API Access and data format

Historical data format is the same as provided by real-time Huobi Futures WebSocket API with addition of local timestamps. If you'd like to work with normalized data format instead (same format for each exchange) see downloadable CSV files or official client libs that perform data normalization client-side.

# pip install tardis-client
import asyncio
from tardis_client import TardisClient, Channel
tardis_client = TardisClient(api_key="YOUR_API_KEY")
​
async def replay():
  # replay method returns Async Generator
  messages = tardis_client.replay(
    exchange="huobi-dm",
    from_date="2020-02-01",
    to_date="2020-02-02",
    filters=[Channel(name="depth", symbols=["BTC_CW"])]
  )
​
  # messages as provided by Huobi Futures real-time stream
  async for local_timestamp, message in messages:
    print(message)
​
​
asyncio.run(replay())

// npm install tardis-dev
const { replay } = require('tardis-dev');
​
async function run() {
  try {
    const messages = replay({
      exchange: 'huobi-dm',
      from: '2020-02-01',
      to: '2020-02-02',
      filters: [{ channel: 'depth', symbols: ['BTC_CW'] }],
      apiKey: 'YOUR_API_KEY'
    });
​
    // messages as provided by Huobi Futures real-time stream
    for await (const { localTimestamp, message } of messages) {
      console.log(localTimestamp, message);
    }
  } catch (e) {
    console.error(e);
  }
}
​
run();

curl -g 'https://api.tardis.dev/v1/data-feeds/huobi-dm?from=2020-02-01&filters=[{"channel":"depth","symbols":["BTC_CW"]}]&offset=0'

curl -g 'localhost:8000/replay?options={"exchange":"huobi-dm","filters":[{"channel":"depth","symbols":["BTC_CW"]}],"from":"2020-02-01","to":"2020-02-02"}'

​

Captured real-time channels

• ​depth During data collection integrity of order book incremental updates is being validated using sequence numbers provided by Huobi Futures real-time feed (version field) - in case of detecting missed message WebSocket connection is being restarted. See also details below regarding depth channel data collection details.

• ​trade​

• ​detail​

• ​bbo - available since 2020-06-24

• ​open_interest - generated channel, available since 2020-06-24 Since Huobi Futures does not offer currently real-time WebSocket open interest channel, we simulate it by fetching that info from REST API (https://huobiapi.github.io/docs/dm/v1/en/#get-contract-open-interest-information) every 4-6 seconds for each instrument. Such messages are marked with "ch":"market.<symbol>.open_interest" and "generated":true fields and data field has the same format as REST API response data.

• ​basis (index price updates) - available since 2020-06-24

• ​liquidation_orders - available since 2020-06-24

• ​contract_info - available since 2020-06-24

​

​

Up until 2020-01-31 depth channel was collected with step0 aggregation level (no aggregation) which produces full order book snapshots for each book change which is very inefficient to store. To circumvent this issue we stored only initial book snapshots and then incremental updates instead - incremental updates were calculated by diffing two subsequent book snapshots and provided in the same format as other depth messages, except having additional update: true flag set as in snippet below. Update with amount (second value in array) set to 0 means such level should be deleted, otherwise price level should be updated with new amount value.

{
  "ch": "market.ETC_CW.depth.step0",
  "ts": 1572652860024,
  "update": true,
  "tick": {
    "mrid": 24660973159,
    "id": 1572652860,
    "bids": [
      [
        4.869,
        1175
      ]
    ],
    "asks": [],
    "ts": 1572652860013,
    "version": 1572652860,
    "ch": "market.ETC_CW.depth.step0"
  }
}

On 2020-01-31 we've switched to depth.size_150.high_freq channel instead when collecting data and which natively provides incremental order book updates without workarounds described above.

Unfortunately it means that when requesting data for depth channel it may return slightly different format depending for which time period request was made. It's only slightly different and boils down to the way order book update messages are marked vs order book snapshots. In depth.size_150.high_freq order book message has event field always present with value update or snapshot, for example:

{
  "ch": "market.LTC_CW.depth.size_150.high_freq",
  "tick": {
   ...
    "event": "update",
   ...
  }
}

For messages before 2020-01-31 we've used depth.step0 channel for collecting order book data which means order book update message has flag update set to true, if it's a snapshot it doesn't have that flag at all, for example:

{
  "ch": "market.ETC_CW.depth.step0",
  "update": true,
  "tick": {
   ...
  }
}

All other fields are are the same (tick.bids and tick.asks etc).

Please feel free to contact us if it's confusing in any way.

We also provide normalization layer that handles those differences transparently via our client libs.

​

Market data collection details

​Market data collection infrastructure for Huobi Futures since 2020-06-19 is located in GCP asia-northeast1 (Tokyo, Japan), before that it was located in GCP europe-west2 region (London, UK). Real-time market data is captured via multiple WebSocket connections.

​

​

​