Markets & timeframes
What data the lab can test on, at what granularity, and — because this matters more than the brochure version — exactly where the limits are.
US stocks and ETFs
Any US-listed stock or ETF works by ticker: AAPL, QQQ, BRK.B. The data cache serves more than 5,000 listed symbols. Type tickers into the symbol controls or just tell the chat — “add NVDA and drop TSLA” is a valid edit. How many symbols a single run may include depends on your plan; see Plans & limits.
The ALL_US universe
Setting the symbol list to ALL_US scans a broad universe of currently listed US common stocks instead of a hand-picked list. The universe is built from public exchange listings and filters out warrants, preferred shares, and fund products; each run takes a capped slice (200 symbols) so results come back in seconds. This is a Max plan feature — a universe scan asks far more of the engine than a five-symbol test.
One honest reason to care about universe scans: testing on a hand-picked list of famous winners is the classic way backtests flatter themselves. A broad universe is the antidote.
Crypto
Crypto uses Polygon-style tickers with an X: prefix — X:BTCUSD, X:ETHUSD. Crypto strategies test on daily bars. These markets trade 24/7, and the engine treats them that way: no trading-day filter, candles on weekends, and days bucketed at UTC boundaries — so calendar rules like month and day_of_week refer to UTC days.
Timeframes
Six bar sizes are supported. Set them from the timeframe control or by asking the chat.
| Timeframe | Bars | History per run |
|---|---|---|
1d (default) | Daily | Whatever range you pick — the default starts January 2016 |
60m / 30m / 15m | Intraday | Up to 60 days |
5m / 1m | Intraday | Up to 60 days |
Intraday data is the genuinely expensive thing in this product, so it’s a plan capability rather than a per-run charge: Pro and Max unlock every timeframe down to 1m, with no per-run cost once you’re on them. See Plans & limits.
The 60-day intraday cap
Intraday runs are rejected server-side when the requested date range spans more than 60 days — you get a clear validation error telling you to tighten the range, and when you switch timeframes in chat, the assistant shrinks the range for you and says why.
The benchmark
Every run automatically includes SPY as a benchmark, even when it is not in your symbol list. Its buy-and-hold curve is plotted next to your equity curve so you always see what doing nothing clever would have returned over the same window. The benchmark rides along free — it does not count as a billable symbol.
Settled candles only
Everything in this product runs on completed OHLCV candles — bars that have closed and settled. There is no live tick feed, no streaming quotes, nothing that changes after you look at it. Fills are simulated at bar prices with slippage applied against you; stops are checked against each bar’s high and low.
This is a philosophy, not just a data plumbing detail. Numbers computed from closed bars are reproducible: run the same spec on the same range twice and you get the same answer. The forward ledger works the same way — deployed strategies are scored on closed bars after the day ends, which is what makes its track records auditable rather than vibes.