How to download historical pump.fun data with Python

Every backtest starts the same way: you need the raw history. This guide walks through downloading tick-level pump.fun trades and 1-second OHLCV candles with Python and the Raiden pump.fun API — six months of history, every fill on both the bonding curve and PumpSwap (pAMM), with keyset pagination for bulk export.

What you need

• A Raiden API key (private beta — free, invite-only). Every request sends it in the X-API-Key header.
• Python 3.10+ with requests and pandas.

1 · Pick your tokens

The screener endpoint lists tokens with keyset pagination. Grab the freshest launches, or filter by status to only get graduated tokens:

import requests

BASE = "https://terminal.raiden.wtf/api"
H = {"X-API-Key": "YOUR_KEY"}

tokens = requests.get(f"{BASE}/tokens", params={"sort": "newest", "limit": 100}, headers=H).json()
mints = [t["mint"] for t in tokens["data"]]

2 · Download every trade of a token

/tokens/{mint}/swaps returns individual fills — price, SOL and token amounts, every fee component, tip, priority fee, slot and block index. Large numbers come back as strings in lamports (1 SOL = 10⁹ lamports) to avoid float loss. Paginate with order + cursor (a boundary timestamp):

def all_swaps(mint):
    rows, cursor = [], None
    while True:
        p = {"order": "asc", "limit": 200}
        if cursor: p["cursor"] = cursor
        page = requests.get(f"{BASE}/tokens/{mint}/swaps", params=p, headers=H).json()
        rows += page["data"]
        if not page.get("next_cursor"): break
        cursor = page["next_cursor"]
    return rows

swaps = all_swaps(mints[0])
# each row: time, trader, is_buy, sol_amount, token_amount, price,
#           fee, creator_fee, priority_fee, tip, slot, block_index, sig, venue …

Add failed=1 to interleave reverted trade attempts (rows carry "failed": true) — the lost bids most data sources drop, useful to measure real demand during a snipe war.

3 · Or grab candles down to 1-second resolution

For most backtests OHLCV is enough — and far fewer rows. tf accepts 1s, 30s, 1m, 1h, 1d, with from/to bounds (RFC3339):

candles = requests.get(f"{BASE}/tokens/{mint}/candles",
    params={"tf": "1s", "from": "2026-07-01T00:00:00Z", "to": "2026-07-02T00:00:00Z"},
    headers=H).json()["data"]
# [{"time": "...", "open": "...", "high": "...", "low": "...",
#   "close": "...", "volume": "...", "trades": 137}, ...]

4 · Into pandas

import pandas as pd

df = pd.DataFrame(swaps)
df["time"] = pd.to_datetime(df["time"])
for col in ("sol_amount", "price", "fee", "tip"):
    df[col] = pd.to_numeric(df[col])
df["sol"] = df["sol_amount"] / 1e9   # lamports → SOL

buys = df[df.is_buy].resample("1min", on="time").sol.sum()

Tips for bulk export

• Use order=asc + cursor for full-history walks — each page is an indexed seek, so deep pagination stays fast.
• Respect your key's rate limit — it is sized per use case and raised on request during the beta.
• One call instead of nine: /tokens/{mint}/pack returns the full token dossier (launch, holder stats, smart money, trust score, wash traders) in a single response — ideal for enriching a dataset.
• Venue matters: each swap carries its venue (bonding curve or PumpSwap), so a token's history is continuous through graduation.

Full request/response shapes for every endpoint are in the API reference. Next up: track pump.fun wallets and their PnL or stream trades in real time over WebSocket — or jump straight to backtesting on 1-second candles, the other half of this pipeline.