> ## Documentation Index > Fetch the complete documentation index at: https://docs.probalytics.io/llms.txt > Use this file to discover all available pages before exploring further. # Fills (Trades) > The fills dataset shape, coverage, and venue-specific trade normalization rules ## Shape A fill is one normalized trade execution. It links to a market and one of that market's outcomes. | Field | Type | Values and meaning | | --------------------- | ----------------- | -------------------------------------------------------------------------------------------------------------------------------------------- | | `id` | `UUID` | Stable Probalytics fill ID. Generated deterministically from the venue trade or chain event. | | `market_id` | `UUID` | Probalytics market ID. Join this to `markets.id`. | | `market_platform_id` | `string` | Venue-native market identifier copied from the linked market. | | `platform` | `enum` | `POLYMARKET` or `KALSHI`. | | `platform_id` | `string` | Venue-native trade, order, or execution ID. The exact source differs by product. | | `outcome` | `Outcome` | Traded outcome containing `id` (UUID), `platform_id` (string), `name` (string), and zero-based `index` (integer). | | `size` | `decimal` | Executed quantity. Prediction markets use contracts/shares; perpetual and margin products use venue instrument units. | | `price` | `decimal` | Execution price for the stored `outcome`. Prediction-market prices are normally between `0` and `1`; perpetual prices are instrument prices. | | `normalized_price` | `decimal` | Prediction markets: execution price expressed as the price of outcome index `0`. Perpetual and margin products: same value as `price`. | | `taker_side` | `enum` | `BUY` or `SELL`. The exact mapping is product-specific and documented below. | | `taker_cash_flow` | `decimal` | Positive gross/notional value assigned to the taker. It is not signed by trade direction. | | `maker_cash_flow` | `decimal` | Positive gross/notional value assigned to the maker. It is not signed by trade direction. | | `taker_id` | `string \| null` | Taker wallet or participant ID when exposed by the source. Currently populated for Polymarket prediction-market fills. | | `maker_id` | `string \| null` | Maker wallet or participant ID when exposed by the source. Currently populated for Polymarket prediction-market fills. | | `fee` | `decimal` | Source fee amount when available. `0` is also used when the source does not expose fees. | | `source_block_number` | `integer \| null` | Chain block containing the execution. Present for Polymarket prediction-market fills. | | `source_tx_hash` | `string \| null` | Chain transaction hash when available. | | `source_log_index` | `integer \| null` | Event position within the chain transaction. | | `timestamp` | `datetime` | Source execution time in UTC. | | `indexed_at` | `datetime` | Time Probalytics wrote the fill. | | `metadata` | `object` | Product-specific annotations. Values are strings, numbers, or booleans. Usually empty for prediction-market fills. | ## Coverage by venue and product | Venue | Product | Fills | Coverage starts | | ---------- | ------------------------ | ------------: | --------------- | | Polymarket | Prediction-market trades | 1,171,916,395 | 2022-11-21 | | Polymarket | Perpetual trades | 9,580,526 | 2026-05-06 | | Kalshi | Prediction-market trades | 1,360,159,691 | 2022-01-01 | | Kalshi | Margin trades | 25,225,233 | 2026-06-03 | Counts as of 2026-08-24 17:59 UTC. Coverage starts use the earliest source `timestamp`. Prediction and perpetual/margin rows are separated using the linked market's `market_type`. Fills are indexed continuously, so these counts change throughout the day. Query the data warehouse for current values: ```sql theme={null} WITH market_id IN ( SELECT id FROM markets FINAL WHERE market_type = 'PERPETUAL' ) AS is_perpetual SELECT platform, if(is_perpetual, 'PERPETUAL', 'PREDICTION') AS product, count() AS fills, min(timestamp) AS earliest_fill FROM fills GROUP BY platform, product ORDER BY platform, product; ``` ## Polymarket prediction-market trades ### Source These fills come from `OrderFilled` events emitted on Polygon by the supported CTF Exchange and Neg Risk Exchange contracts, including their V2 versions. Legacy executions can emit two related `OrderFilled` logs. Probalytics retains the maker-focused representation instead of publishing both as separate trades. ### Field mapping | Polymarket source | Probalytics field | | ---------------------------------------------------- | ------------------------------------ | | Transaction hash + log index + taker address | `id` | | Outcome token's linked market | `market_id`, `market_platform_id` | | Order hash | `platform_id` | | Outcome token ID | `outcome` | | Outcome-token amount ÷ `1,000,000` | `size` | | USDC amount ÷ outcome-token amount | `price` | | `price` for outcome index `0`; otherwise `1 - price` | `normalized_price` | | Taker pays USDC | `taker_side = 'BUY'` | | Taker supplies outcome tokens | `taker_side = 'SELL'` | | USDC amount ÷ `1,000,000` | `taker_cash_flow`, `maker_cash_flow` | | Taker and maker wallet addresses | `taker_id`, `maker_id` | | Event fee ÷ `1,000,000` | `fee` | | Block number, transaction hash, log index | Source provenance fields | | Block timestamp | `timestamp` | Outcome index `0` is the reverse-alphabetically first normalized market outcome. For a typical Yes/No market, Yes is index `0` and No is index `1`. ## Polymarket perpetual trades Polymarket perpetual trades come from the perpetual product feed and link to markets with `market_type = 'PERPETUAL'`. | Polymarket perpetual source | Probalytics field | | ------------------------------------- | ----------------------------------------------------------- | | Instrument ID | `market_id`, `market_platform_id` through the linked market | | Trade ID | `id`, `platform_id` | | Synthetic Long outcome | `outcome` | | Quantity | `size` | | Instrument price | `price`, `normalized_price` | | `long` | `taker_side = 'BUY'` | | `short` | `taker_side = 'SELL'` | | `size × price` | `taker_cash_flow`, `maker_cash_flow` | | Source timestamp in milliseconds | `timestamp` | | Source transaction hash, when present | `metadata.tx_hash` | Participant IDs are `null`, and `fee` is `0` because the current source does not expose those values. ## Kalshi prediction-market trades ### Source These fills come from Kalshi trade records. Each record contains the market ticker, Yes and No prices, traded count, taker side, trade ID, and execution time. ### Field mapping | Kalshi source | Probalytics field | | ------------------------------------ | ----------------------------------------------------------- | | Market ticker | `market_id`, `market_platform_id` through the linked market | | Trade ID | `id`, `platform_id` | | `taker_side = 'yes'` | Yes `outcome`; `price = yes_price_dollars` | | `taker_side = 'no'` | No `outcome`; `price = no_price_dollars` | | `count_fp` | `size` | | `yes_price_dollars` | `normalized_price` | | Every current Kalshi prediction fill | `taker_side = 'BUY'` | | `size × price` | `taker_cash_flow`, `maker_cash_flow` | | `created_time` | `timestamp` | The source's Yes/No taker side selects which outcome was bought. It does not map to the stored BUY/SELL enum, so all current Kalshi prediction fills are stored as `BUY`. Participant IDs are `null`, and `fee` is `0` because the current trade source does not expose those values. Kalshi `KXMVE` trades are not included in `fills` because their markets are not available to this normalization path. ## Kalshi margin trades Kalshi margin trades come from the margin product feed and link to markets with `market_type = 'PERPETUAL'`. | Kalshi margin source | Probalytics field | | ---------------------- | ----------------------------------------------------------- | | Margin ticker | `market_id`, `market_platform_id` through the linked market | | Trade ID | `id`, `platform_id` | | Synthetic Long outcome | `outcome` | | Count | `size` | | Instrument price | `price`, `normalized_price` | | `taker_side = 'bid'` | `taker_side = 'BUY'` | | `taker_side = 'ask'` | `taker_side = 'SELL'` | | `size × price` | `taker_cash_flow`, `maker_cash_flow` | | `created_time` | `timestamp` | | Product marker | `metadata.instrument_type = 'perpetual_future'` | Participant IDs are `null`, and `fee` is `0` because the current source does not expose those values. ## Reading fills across products * Join `market_id` to `markets.id` before interpreting price or size units. * Use `price` for the stored outcome's execution price. * Use `normalized_price` to create one outcome-0 price series for prediction markets. * Do not treat perpetual `normalized_price` as a probability. * Do not sum the cash-flow fields as signed P\&L; both currently store positive gross/notional values. See [Tables & Schemas](/docs/sql-guide/tables#fills) for access-specific database types and query behavior.