---
license: mit
task_categories:
  - other
tags:
  - polymarket
  - prediction-market
  - crypto
  - bitcoin
  - ethereum
  - solana
  - time-series
size_categories:
  - 100K<n<1M
---

# Polymarket Crypto Up/Down Market Metadata

Raw Gamma event metadata for every crypto up/down market on Polymarket
during March 1 - June 3, 2026 (UTC), for BTC, ETH, and SOL on the
5-minute and 15-minute timeframes.

## Structure

```
data/
  raw/
    metadata/
      <asset>/                  # btc, eth, sol
        <timeframe>/            # 5m, 15m
          _manifest.json        # series-level manifest (expected/fetched/missing/failed)
          <DD-MM-YYYY>/         # UTC day folder
            <HH-MM-SS-UTC>/     # UTC time folder (window start)
              metadata.json     # raw Gamma event for this window
```

Total: **108,276 market files** across 6 series (3 assets x 2 timeframes),
plus 6 series-level `_manifest.json` and a cross-series index at
`data/raw/_index/by_window.jsonl`.

## What's in each `metadata.json`

The full raw Gamma `/events` response for one market, including:
- `id`, `slug`, `title`, `description`
- `startDate`, `endDate` (UTC ISO 8601)
- `active`, `closed`, `archived`, `new`
- `markets[]` — array of inner markets (typically 1) with `conditionId`,
  `clobTokenIds` (UP and DOWN token IDs), `volume`, `liquidity`
- `tags[]`, `image`, `icon`, etc.

## Coverage

| Series | Windows | Missing |
|---|---|---|
| btc/5m  | 27,068 | 5 |
| btc/15m |  9,025 | 0 |
| eth/5m  | 27,066 | 7 |
| eth/15m |  9,025 | 0 |
| sol/5m  | 27,067 | 6 |
| sol/15m |  9,025 | 0 |

**18 windows are genuinely absent from Polymarket's history.** They cluster
on 3 specific clock-times across 3 specific days (all on 5m markets, all 3
assets). The collector recorded them in each `_manifest.json`'s
`missing_slugs` list. See the manifest for exact slugs.

## Slug convention

`{asset}-updown-{timeframe}-{window_start_epoch}` where
`window_start = (unix_ts // duration_seconds) * duration_seconds`. The
window is `[T, T+duration)` and the slug encodes the start, not the close.

The 5m and 15m market windows close at `T+300` and `T+900` seconds
respectively.

## Loading this dataset

The dataset preserves the on-disk structure, so you can:

```python
from huggingface_hub import snapshot_download
path = snapshot_download(
    repo_id="krish301/polymarket-crypto-metadata",
    repo_type="dataset",
)
# Walk the tree
import os, json
for root, dirs, files in os.walk(os.path.join(path, "data", "raw", "metadata")):
    for f in files:
        if f == "metadata.json":
            full = os.path.join(root, f)
            with open(full) as fh:
                ev = json.load(fh)
            # ev["slug"], ev["endDate"], ev["markets"][0]["clobTokenIds"], ...
```

Or browse the tree in the HF dataset viewer at:
<https://huggingface.co/datasets/krish301/polymarket-crypto_metadata>

## Provenance

Generated by the `polymarket_bot.data_collection` package:
- `slug.py` — slug builder and window-sweep generator
- `gamma.py` — async Gamma client with retry/backoff
- `collect_metadata.py` — backfill CLI
- `verify.py` — independent post-collection verifier

Collection period: 2026-06-04. See per-series `_manifest.json` for
exact fetched_at timestamps.