Adjustment convention
Day / week / month / quarter / year bars are always forward-adjusted, so they do not gap on ex-dividend dates and moving averages are not fooled by the hole. Minute bars are unadjusted — they carry the raw price. There is no adjust parameter, and you should not adjust them yourself.
The short version
Day / week / month / quarter / year bars are always forward-adjusted — there is no adjust parameter and none is needed, and prices do not gap on ex-dividend dates. Minute bars are the exception: they are unadjusted and carry the raw price.
What forward adjustment means
Historical prices are back-adjusted by each ex-dividend / ex-rights ratio, using the latest price as the reference point.
The benefit is a continuous price series: on the day a stock pays a dividend or splits, the chart does not drop by a sudden hole, so moving averages, patterns and percentage changes are not distorted by it.
The trade-off is that historical prices change over time — every ex-dividend event recomputes all earlier prices. The same stock’s historical K-line can therefore return different values today than it did last month. For backtesting, re-fetch the data each time rather than caching old price series for long.
Four things to know
1. There is no adjust parameter
/v1/kline accepts code, period, count, start and end — there is no adjust, and passing one will not be recognised. The convention is fixed, there is nothing to choose.
2. No gap on ex-dividend dates
Moving averages, pattern detection and technical indicators are not fooled by the ex-dividend hole. That is the direct benefit of a fixed forward adjustment.
3. Minute bars are unadjusted
When period is one of m1 / m5 / m15 / m30 / m60 / m120, prices are not adjusted — the upstream source returns raw prices for minute-level data. A minute series can therefore gap on an ex-dividend date. That is a different convention from the continuous daily series; do not compare the two directly.
4. Do not double-adjust
If you splice unadjusted data from another source (or data adjusted a different way) together with the prices we return, or apply another adjustment pass on top, the series is wrong — that is double adjustment.
How to check it yourself
Take a stock’s close series (last in each data row) for three days either side of an ex-dividend date and look at whether it is continuous:
- Continuous (no sudden step or hole) ⇒ the convention is working (daily bars should look like this).
- A visible gap ⇒ first confirm the stock really did go ex-dividend that day (check the announcement), then investigate. On minute bars a gap is expected — minute prices are unadjusted.
Common misuses
- Passing
adjust— the parameter does not exist; the convention is fixed. - Aligning minute bars to the adjusted daily series — minute bars are unadjusted; splicing them with forward-adjusted daily prices misaligns the series.
- Splicing in unadjusted data — double adjustment; the series is wrong.
- Caching historical K-lines for a long time — forward adjustment recomputes on each new ex-dividend event, so stale caches go out of date.
Last updated: 2026-10-07
Convention taken verbatim from the /v1/kline endpoint description, checked against the live openapi on 2026-10-07.