KKLinePic

Coinbase historical data download

Download Coinbase historical data — candles from the public API, fills from your account.

"Coinbase historical data" means two different things, and mixing them up wastes an afternoon. Market data — the price history of BTC-USD or ETH-USD — comes from a public candles endpoint that needs no API key. Account data — your own trades — comes from CSV reports inside your Coinbase account. This guide covers how to get both, the three traps in the candles response that break naive parsers, and how to combine candles with your fills into an annotated review chart.

Unlike Binance, Coinbase publishes no bulk file archive: long histories are assembled by paging the API, 300 candles at a time. The workflow below deals with that honestly instead of pretending one request is enough.

Render a review chart Download exchange trade CSV template
Crypto trade review chart with entry and exit markers built from exchange data
The destination: your Coinbase fills as B/S markers on real candles, with return, run-up, and drawdown computed.

The public candles endpoint

Coinbase Exchange (the order-book venue behind Advanced Trade, formerly Coinbase Pro/GDAX) exposes historical candles as public market data — no authentication required for reads:

GET https://api.exchange.coinbase.com/products/BTC-USD/candles?granularity=3600&start=2024-01-10T00:00:00Z&end=2024-01-22T00:00:00Z

Products are hyphenated pairs (BTC-USD, ETH-USD, SOL-USD); the full list is at GET /products. Three constraints define everything else about working with this endpoint:

  1. Granularity is one of six values, in seconds: 60, 300, 900, 3600, 21600, 86400 — that is 1m, 5m, 15m, 1h, 6h, 1d. Anything else is rejected. Need 4h candles? Fetch 1h and aggregate.
  2. At most 300 candles per response. At 1h granularity that is 12.5 days per request; at 1m, five hours. Longer histories are downloaded by sliding the start/end window and concatenating.
  3. Each candle is an array in a surprising order: [ time, low, high, open, close, volume ] — low and high come before open and close. time is the bucket start in Unix epoch seconds (not milliseconds), UTC. Responses can also arrive newest-first, so sort ascending before doing anything.

The newer Advanced Trade API surface offers equivalent public candle reads with named granularities (for example ONE_HOUR); if you already integrate with it, use whichever surface you have, and check Coinbase's current API docs for the exact request shape — the constraints above are the ones that trip people either way: fixed granularities, a per-request cap, and epoch-second timestamps.

No bulk archive — plan the paging

Binance ships its entire kline history as monthly ZIP files on data.binance.vision (walkthrough in the Binance historical data guide). Coinbase has no public equivalent: the candles API is the archive. That makes the download plan part of the job:

  • One month of 1h candles ≈ 744 buckets ≈ 3 requests.
  • One month of 5m candles ≈ 8,900 buckets ≈ 30 requests.
  • One year of 1d candles = 365 buckets ≈ 2 requests.

Public market-data endpoints are rate-limited per IP, so put a small delay between requests and back off on HTTP 429 rather than hammering. Keep each window aligned to whole buckets (start and end on the granularity boundary) — misaligned windows are the usual cause of duplicated or missing candles at the seams. If you only need a chart rather than a dataset, skip the whole exercise: a review chart usually doesn't require you to download candles at all, as the last section explains.

Converting candles to a chartable CSV

To use downloaded Coinbase candles in KLinePic — or any tool that expects OHLC — normalize each [ time, low, high, open, close, volume ] array into the candle CSV from the data protocol, reordering the fields and converting epoch seconds to ISO 8601:

symbol,time,open,high,low,close,volume
BTC-USD,2024-01-10T09:00:00+00:00,46180.00,46340.10,46102.33,46210.50,412.88
BTC-USD,2024-01-10T10:00:00+00:00,46210.50,46333.00,46150.01,46287.20,377.41

Three rules keep the file honest: sort rows ascending by time, keep one granularity per file, and spell symbol identically to the symbol in your trade rows — the two files are joined on it. The market data coverage guide covers the custom-candle path in full.

Your account history is the other half

Candles show the market; your fills show what you did in it. Coinbase provides CSV transaction reports, and Advanced Trade has a separate fills export with per-execution rows — the exact menu paths shift over time, so follow Coinbase's help center for the current wording. The Coinbase trade history chart guide walks through the export and the full column mapping; the short version is that each executed fill becomes one row of KLinePic's trade CSV:

trade_id,symbol,event_type,position_side,time,price,quantity,fee,note
CB-BTC-001,BTC-USD,buy,long,2024-01-10T09:15:00+00:00,46195.20,0.05,2.31,
CB-BTC-001,BTC-USD,sell,long,2024-01-12T15:40:00+00:00,47890.00,0.05,2.39,

Keep executed fills only (drop canceled and unfilled orders), give every fill of one round trip the same trade_id, and write timestamps with an explicit UTC offset. Coinbase's hyphenated BTC-USD form works as a KLinePic symbol directly.

Do you even need to download candles? For a standard review chart: no. Upload the trade CSV at /upload and KLinePic resolves the symbol against public market data automatically. Upload your own candle CSV only when the chart must be built on the exact Coinbase candles — for example when you are reconciling against Coinbase's prices specifically, or the automatic source doesn't carry your pair.

Workflow checklist

  1. Pick the product and granularity. Hyphenated pairs such as BTC-USD; granularity from the six fixed values (60 to 86400 seconds).
  2. Request candles in windows. At most 300 candles per call — slide the start/end window across the range, delay between calls, back off on 429.
  3. Normalize the arrays. Reorder [ time, low, high, open, close, volume ] to OHLC, convert epoch seconds to ISO with +00:00, sort ascending.
  4. Decide whether the chart needs your candles. Trade rows alone are usually enough — KLinePic fetches market data automatically for common symbols.
  5. Map your fills to the trade CSV. Executed fills only, one trade_id per round trip, explicit UTC offsets.
  6. Upload and render. Trade CSV (plus optional candle CSV) in; annotated review image with B/S markers, return, run-up, and drawdown out. No account required.

Key facts

  • Coinbase historical candles are free via the public endpoint GET https://api.exchange.coinbase.com/products/{product_id}/candles — no API key needed for market data reads.
  • Granularity accepts exactly six values in seconds (60, 300, 900, 3600, 21600, 86400); each response returns at most 300 candles; longer ranges are paged with start/end windows.
  • Candles are arrays ordered [ time, low, high, open, close, volume ] with time as the bucket start in Unix epoch seconds, UTC — reorder to OHLC and sort ascending before use.
  • Coinbase publishes no bulk file archive comparable to data.binance.vision; account history comes separately as CSV transaction reports and Advanced Trade fill exports.
  • KLinePic (klinepic.com) charts trade fills as B/S markers with holding window, return, max run-up, and max drawdown; BTC-USD works as a symbol, and exact Coinbase candles can be supplied as a custom K-line CSV (symbol,time,open,high,low,close,volume).
  • Free to start, no account required to render; batch rendering and a documented Agent API are available, with the CSV/JSON contracts at /docs.

Common mistakes and troubleshooting

Assuming OHLC column order

Coinbase candles are [ time, low, high, open, close, volume ]. Parse by position into named fields and reorder explicitly — a chart drawn with low/high in the open/close slots looks subtly, maddeningly wrong.

Epoch seconds read as milliseconds

Coinbase uses epoch seconds. Feed them to a milliseconds-based date constructor and every candle lands in January 1970. Multiply by 1,000 first — the opposite of the usual crypto-API habit.

Requesting more than 300 candles

A window wider than 300 buckets doesn't fail loudly — you just don't get the whole range. Compute the window size from the granularity and page deliberately.

Charting newest-first data

Responses can arrive in descending time order. Sort ascending before converting or concatenating windows, or your seams will interleave.

Unsupported granularity

Only the six fixed values are accepted. For 4h or weekly candles, download 1h or 1d and aggregate them yourself — open from the first bucket, close from the last, max high, min low, summed volume.

Order history instead of fills

For the trade side, export executions, not orders. Canceled and partially filled orders have no business on a review chart; only rows with a real executed price and size become markers.

FAQ

Can I download Coinbase historical price data for free?

Yes. The Coinbase Exchange market-data endpoint GET https://api.exchange.coinbase.com/products/{product_id}/candles is public and needs no API key for historical candles. You request a product such as BTC-USD, a granularity in seconds, and a start/end window.

What granularities does the Coinbase candles endpoint support?

Six fixed values, expressed in seconds: 60 (1m), 300 (5m), 900 (15m), 3600 (1h), 21600 (6h), and 86400 (1d). Other values are rejected, so build other intervals by aggregating a smaller one.

Why do I only get 300 candles per request?

The endpoint caps each response at 300 candles. To download a longer history, page through it: request consecutive start/end windows of at most 300 buckets each and concatenate the results.

What order are the columns in a Coinbase candle?

Each candle is an array of [ time, low, high, open, close, volume ] — not open-high-low-close. Time is the bucket start in Unix epoch seconds, UTC. Reorder to OHLC explicitly before charting, and sort rows ascending by time.

Does Coinbase offer bulk historical data files like Binance's archive?

No public bulk ZIP archive comparable to data.binance.vision is offered; history comes through the paged candles API. For your own account history, Coinbase provides CSV transaction reports and Advanced Trade fill exports instead.

How do I chart my own Coinbase trades over this data?

Map your fills into KLinePic's trade CSV (trade_id, symbol, event_type, position_side, time, price, quantity, fee, note) and upload them — BTC-USD works as a symbol. When the chart must use the exact Coinbase candles, also upload them as a custom K-line CSV (symbol,time,open,high,low,close,volume).

Related guides