QuantLake
GitHub
Documentation

Data types

The data types QuantLake collects, their columns and how values are stored.

Data type names and columns follow the Databento schemas. Data types marked with * are QuantLake extensions for Tardis.dev data that has no Databento equivalent. The same data type has the same columns, whichever provider it comes from.

Data types

Data typeProviderWhat it holds
tradesTardis.dev, DatabentoEvery trade
mbp-1Tardis.dev (quotes), DatabentoBest bid and ask after each change
mbp-10DatabentoTop 10 price levels after each change
mbp-5, mbp-25 *Tardis.dev (book_snapshot_5, book_snapshot_25)Top 5 or 25 price levels
book_ticker *Tardis.devThe exchange's own best bid and ask feed. Same columns as mbp-1
incremental_book_L2 *Tardis.devUpdates to every price level of the order book
derivative_ticker *Tardis.devFunding rate, open interest, last, index and mark price
liquidations *Tardis.devLiquidations
mboDatabentoEvery order book event, order by order (L3)
tbboDatabentoEach trade, with the best bid and ask just before it
bbo-1s, bbo-1mDatabentoBest bid and ask, every second or minute
ohlcv-1s, ohlcv-1m, ohlcv-1h, ohlcv-1dDatabentoOpen, high, low, close and volume bars
definitionDatabentoInstrument definitions, such as tick size, expiry and legs
statisticsDatabentoOfficial statistics, such as settlement price and open interest
statusDatabentoTrading status, such as halts and pre-open
imbalanceDatabentoAuction imbalance
cmbp-1, cbbo-1s, cbbo-1m, tcbboDatabentoBest bid and ask across several venues

Common columns

  • ts_recv: when the provider received the message. Each daily file is split by this time.
  • ts_event: the time set by the exchange. OHLCV bars have only this time, and their daily files are split by it.
  • publisher_id, instrument_id: Databento IDs. Null for Tardis.dev data.
  • raw_symbol: the actual symbol of each row. This differs from the symbol folder for grouped and continuous symbols, such as ES.c.0.

Order book and trade columns

These columns are in trades, mbp-*, tbbo and book_ticker.

  • action, side, depth, price, size, flags, ts_in_delta, sequence: the event that changed the book. For trades, this is the trade itself, and side is the taker side.
  • bid_px_00, ask_px_00, bid_sz_00, ask_sz_00, bid_ct_00, ask_ct_00: the book after the event. 00 is the best level, 01 the next, and so on.
  • trade_id (trades only): the trade ID from Tardis.dev. Null for Databento data.

Extension columns

  • incremental_book_L2: side, price, size (the new size at that price level; 0 removes the level) and flags (32 marks rows from an initial snapshot). Skip the rows before the first row with flags 32.
  • derivative_ticker: ts_funding, funding_rate, predicted_funding_rate, open_interest, last_price, index_price, mark_price.
  • liquidations: side (B means a short position was liquidated, A means a long position was liquidated), price, size, trade_id.

For the other Databento data types, the columns are the same as in the Databento schemas.

Values

  • Timestamps are TIMESTAMP(NANOS, UTC). Tardis.dev times, which are in microseconds, are stored without loss.
  • Prices are 64-bit floating point numbers. Databento's fixed-point prices are converted to normal prices, and undefined prices become null.
  • Sizes are 64-bit floating point numbers, because crypto amounts can be fractional.
  • Side is B (bid or buy), A (ask or sell) or N (none).
  • Columns that a provider doesn't have are null. For example, Tardis.dev rows have no instrument_id or sequence.

Tardis.dev column names

If you have used Tardis.dev CSV files, the old column names map as follows:

Tardis.devQuantLake
local_timestampts_recv
timestampts_event
amountsize
idtrade_id
asks[0].priceask_px_00
is_snapshotflags 32

For the full definition of each column, see the Databento schemas and the Tardis.dev data types.