Interval & Data Cadence Reference
Use this guide to match your strategy analysis to the right candle interval, understand data latency, and decide when to switch between dataset versions.
Key facts at a glance
- Candles are sourced from partner exchanges and normalised to UTC to avoid daylight-saving drift.
- Analytics defaults to Dataset V1 (fast lane) for all users; V2 remains available for deep history but can be 3-5x slower.
- New candles land in V1 within ~45 seconds on average; bulk backlog jobs run every 15 minutes to reconcile gaps.
- CSV exports include the interval, timezone (UTC), and ingestion timestamp so you can audit latency.
Supported intervals
Each interval aggregates trades into the candle size shown below. The evaluation window indicates how much history feeds the current score, not the total data stored.
| Interval | Candle length | Evaluation window (rolling) | Best for | Typical V1 delay |
|---|---|---|---|---|
| 1m | 60 seconds | Last 21 days (~30K candles) | High-frequency scanning, scalping backtests, monitoring execution drift. | 20-45 seconds |
| 3m | 180 seconds | Last 30 days (~14.4K candles) | Intraday trend following and VWAP-style reversion checks. | 45-60 seconds |
| 5m | 300 seconds | Last 45 days (~12.9K candles) | Balanced intraday setups, automation pilots, session bias checks. | 60-75 seconds |
| 15m | 900 seconds | Last 90 days (~8.6K candles) | Session swing strategies, breakout confirmation, lower-noise comparisons. | 90-120 seconds |
| 30m | 1,800 seconds | Last 120 days (~5.7K candles) | Cross-exchange confirmation, managed swing portfolios. | 2-3 minutes |
| 1h | 3,600 seconds | Last 180 days (~4.3K candles) | Swing trading, cross-asset studies, checklist validation. | 3-4 minutes |
| 4h | 14,400 seconds | Last 270 days (~1.6K candles) | Position trading, macro thesis validation, funding overlays. | 4-6 minutes |
| 1d | 86,400 seconds | Last 540 days (~540 candles) | Macro dashboards, long-term score monitoring, investor updates. | 6-8 minutes |
Tip: Use the limit dropdown in Analytics to keep comparisons within the same interval so scores remain apples-to-apples.
Dataset versions
- Lowest latency, 1m ingestion loop with auto backfill.
- Focus on the most recent 6 months of strategies with continuous scoring.
- Recommended for day-to-day analysis, dashboards, and CSV exports.
- Extends back 2+ years with broader factor combinations and experimental scoring.
- Runs on a longer processing queue; expect 3-5x slower responses.
- Use selectively for research sessions; crawlers and automation should stay on V1 until further notice.
Switch datasets in Analytics via the dataset toggle above the filters. Your choice is remembered per session.
How updates flow through the system
- Raw ticks stream in from exchange APIs and are batched into 1-minute candles per pair.
- Higher intervals (3m and above) are derived from the 1m feed to keep timestamps aligned.
- Evaluation jobs recalculate strategy scores as fresh candles arrive and roll forward the window.
- Daily at 00:10 UTC we snapshot metrics for audit and purge stale series.
If the Analytics timestamp is >10 minutes behind the wall clock, raise a support ticket with the pair, interval, and screenshot of the status banner.
Choosing the right interval
- Scalping / automation pilots — Stay on 1m or 3m. Pair with responsive indicators (EMA/HMA) and monitor latency via the ingestion timestamp.
- Intraday swing teams — Blend 5m and 15m for signal confirmation. Require >=250 closed trades and PF >=1.4 before scaling.
- Swing / position desks — Evaluate 30m, 1h, and 4h side by side. Align with funding windows and macro catalysts logged in your playbook.
- Macro / investor relations — Use 1d for narrative reporting. Combine with V2 snapshots when you need multi-year context.
FAQ
Which timezone do charts use?
All timestamps are stored and displayed in UTC. When exporting to CSV, the `ingested_at` column shows the UTC time the candle was processed.
Do public holidays or exchange outages change intervals?
No. Crypto markets run 24/7. If an exchange is unreachable, we mark candles as partial and backfill once connectivity returns.
How far back can I request history?
The interface surfaces the active evaluation window. For deeper history, export from V2 or contact support with the pair, interval, and required span.