File size: 2,817 Bytes
260389b
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
9f2a22c
260389b
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
---

license: mit
pretty_name: Polymarket on-chain fills (predict-ngin)
tags: [polymarket, prediction-markets, finance, polygon, trades]
size_categories: [1B<n<10B]
configs:
- config_name: fills
  data_files: "lake/polymarket/fills/*/*/*.parquet"
  default: true
- config_name: bars_1h
  data_files: "lake/polymarket/bars_1h/*/*/*.parquet"
- config_name: bars_1m
  data_files: "lake/polymarket/bars_1m/*/*/*.parquet"
- config_name: markets
  data_files: "lake/polymarket/markets.parquet"
---


# Polymarket on-chain fills

Every Polymarket fill on Polygon from 2022-11 to 2026-10-18 (exclusive): 1,419,910,711 rows,
$62.0B notional, YES-normalised (each fill priced as the YES outcome), sorted by
category, event, market and time inside monthly folders. Also hourly and 1-minute OHLC / VWAP
bars per market, and a market table with categories.

Built by [predict-ngin](https://github.com/BonelessWater/predict-ngin) from the
[SII-WANGZJ/Polymarket_data](https://huggingface.co/datasets/SII-WANGZJ/Polymarket_data) releases
(MIT) and, after them, Polygon `OrderFilled` logs read from public RPC nodes.

## Get it

With the predict-ngin repository (downloads into your data folder, then fetches the days since this
snapshot and updates the lake):

```bash

uv run predict-ngin data pull polymarket

```

Without it:

```bash

hf download domdd/predict_ngin --type dataset --local-dir predict-data

```

or query it in place with DuckDB:

```sql

SELECT count(*) FROM 'hf://datasets/domdd/predict_ngin/lake/polymarket/fills/year=2026/month=*/*.parquet';

```

## Layout

| Path | Contents |
|---|---|
| `lake/polymarket/fills/year=YYYY/month=MM/` | fills; `fills_state.json` has per-month row counts |
| `lake/polymarket/bars_1h/`, `bars_1m/` | per-market bars from maker fills |
| `lake/polymarket/markets.parquet` | one row per market: question, event, tags, category |
| `polymarket_clean/trades_ext/` | source files predict-ngin uses to keep recent months current |
| `polymarket/hf_20261005/` | market lists used to map token ids to markets |
| `polymarket_clean/event_tags.parquet` | Gamma event tags behind the market categories |

## Notes

- **Count volume with `NOT is_taker_order_row`.** Rows from before 2025-12-30 include the taker

  order's aggregate rows; counting them double-counts volume.

- **`ts_chain`** is the block time. Before 2025-12-30 it comes from Polygon block anchors, because
  the source release's own timestamps there are off by minutes to hours; `ts_local` keeps the
  release value. `timestamp_unreliable` flags rows with no anchored time.
- `yes_price` and `yes_side` are YES-normalised; `price_raw` is the traded token's price.
- Wallet addresses (`maker`, `taker`) are public on-chain data.
- No Kalshi data is included.