Local data cache
A local data cache is the on-disk store in which trading software keeps the bars, ticks and session history it has already downloaded, so that a chart can open, scroll and compute without asking the provider again. It is a copy of what was received, not a source of new data.
Senzoukria · Glossary · Updated September 2026
What the cache holds and why
Downloading history is a real round-trip to the provider: a week of one-minute footprint on an active contract is many requests, and providers cap what they return per request. Keeping the result on disk means the second opening of the same contract costs nothing and works when the market is closed or the provider is down. The cache is keyed by instrument and granularity; a contract charted at one timeframe does not automatically populate another.
- Bars with their bid and ask volume per price level, so footprints and profiles recompute from disk.
- Ticks for replay when the source provides them.
- Coverage metadata: which windows were requested, which arrived, which came back empty.
- No order book history unless a depth-recording source was used; trades cannot rebuild the book.
Cache and coverage
A cache makes gaps easier to see, not easier to hide. Each requested window is either present, partially present or absent, and that status should travel with any study computed on it. A window that a provider returned empty during a weekend is a legitimate absence; the same window returned empty on a weekday is a gap to retry. Some tools stop asking for a window once it came back empty, which is the right default for a weekend and the wrong one after a provider outage, so a manual re-request is needed.
In Senzoukria
The Settings page points to the Account page for data feeds, brokers and the local cache. When a chart opens offline, it reports the age of the last cached bar and the number of bars cached on the timeframe, and offers to replay a cached session; a weekend notice says that every session already in the local cache opens, scrolls and computes as usual, and that only new history needs an open market. The backtest data panel reads the cache, shows how many windows are cached out of those requested, and can download the missing history; an instrument whose tick size is unknown is not cached and is re-downloaded on every load until the tick size is added to the instrument catalogue. A tick export writes the cached session as gzipped CSV with timestamp, price, size and aggressor side.
Common mistakes
- Treating the last cached close as a live price; the welcome screen labels it "last cached close — not live".
- Assuming a cached session at one granularity covers a backtest at another.
- Deleting the cache to fix a display problem and then paying the download again.
- Expecting the cache to contain depth history for a heatmap when only trades were recorded.
Related
- Futures backtesting guide
- Data gap (incomplete history)
- History permission (broker refusal)
- Footprint charts
This page in other languages
Frequently asked questions
- Can I chart with no connection at all?
- Yes, for instruments already in the cache. The chart opens on the cached bars, indicators compute, and a cached session can be replayed. Nothing new arrives until a source is connected and the market is open. An instrument never charted before has no local data and cannot open offline.
- Does the cache fill in missing intervals?
- No. The cache stores what was received and records what was not. A missing window stays marked as missing until a new request succeeds. Filling holes with zeros would corrupt cumulative delta, profiles and backtests, so the software leaves them visible.