Data providers & their limits

Where the candles come from, why history depth varies, and what the Verified badge means.

STLab tests against real market data from five sources. They differ in market coverage, how the data reaches the backtest engine, and how much history you can load — which is why two charts of the same pair can carry different depths.

The five sources

  1. DeepCoin — crypto futures, fetched by our servers by default.
  2. Binance, Bybit, OKX — crypto, fetched by your browser directly from the exchange (these exchanges block server-side requests from our region, so your browser does the fetching).
  3. Yahoo Finance — US and Korean stocks and ETFs, fetched by our servers by default.

Verified vs Browser-fetched

Every result shows where its candles came from. If our servers fetched them, the run is marked Verified. If your browser sent us the candles it had already loaded, the run is marked Browser-fetched — the server validates their shape and OHLC integrity but does not re-fetch them. The label answers who fetched the data, not which provider you picked: the same DeepCoin pair can come back either way depending on how you started the run.

  1. Started from the chart — your browser hands over the bars already on screen, so the run is Browser-fetched even on DeepCoin or Yahoo. That is deliberate: the run then covers exactly the window you were looking at.
  2. Binance, Bybit or OKX — always Browser-fetched. These exchanges block server-side requests from our region, so your browser is the only thing that can reach them.
  3. Everything else on DeepCoin or Yahoo — our servers fetch the window themselves, so the run is Verified.

History depth

How many candles the chart can load depends on your plan: Trader 5,000 · Pro 20,000 · Studio 50,000 · Quant 200,000 bars. One notable exception: DeepCoin serves at most ~5,000 candles regardless of plan.

What this means for backtests

  1. The result page's Source line is the ground truth: provider, candle count, and the exact tested window.
  2. If a backtest covers less time than you expected, the timeframe × candle-cap math above is almost always the reason.
  3. Short-selling cost caveat: shorts run on every provider, but borrow/funding costs aren't modeled anywhere — so short results on any provider (especially Yahoo stocks/ETFs, where borrow fees are real) exclude carry costs.