agentleFS
voyager67voyager67/JaneStreet-Agentic-Skills/skills/tick-data-engineering/SKILL.md

tick-data-engineering

Skill13 starsChanged 5 days agoApache-2.0View on GitHub

Build or audit reliable tick, quote, trade, and order-book data pipelines for electronic trading research. Use for feed ingestion, normalization, timestamp alignment, sequence recovery, schema design, and historical data quality.

The file

We split this skill into the rules it gives an agent

7 rules. About 0 are common, 7 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈7

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Treat market data as event streams with venue-specific semantics, not as an ordinary table of prices.

    line 7
  2. Distinctive:

    Detect feed defects: Check sequence gaps, duplicates, out-of-order records, invalid types, stale updates, zero/negative sizes, crossed …

    line 15
  3. Distinctive:

    Do not forward-fill missing book updates across an unknown sequence gap.

    line 23
  4. Distinctive:

    Do not use local receive time as exchange event time or vice versa.

    line 24
  5. Distinctive:

    Do not combine data from multiple vendors without mapping differences in trade conditions, corrections, and timestamp semantics.

    line 25
  6. Distinctive:

    Do not drop outliers without preserving a quarantine log and reason.

    line 26
  7. Distinctive:

    Avoid floating-point prices as primary keys.

    line 27

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Tick Data Engineering

Treat market data as event streams with venue-specific semantics, not as an ordinary table of prices. Preserve raw evidence so normalized datasets can be reproduced and challenged.

Workflow

  1. Inventory the sources. Record venue/provider, protocol and version, instrument coverage, session coverage, message granularity (trades, L1, L2, MBO), entitlements, timestamp fields, delivery guarantees, and known corrections.
  2. Keep raw immutable. Archive raw messages or vendor files with checksums, acquisition time, source metadata, and version. Normalization must be a derived layer, never the only surviving representation.
  3. Define a canonical event schema. Include venue, instrument identifier, event type, exchange timestamp, local receive timestamp, sequence/channel, price ticks, size lots, order/trade ID if available, side, and source/version metadata. Use explicit units and nullable fields.
  4. Normalize safely. Parse fixed-point price/quantity correctly, handle symbol mapping and instrument lifecycle, convert timezones once with provenance, and model sessions/auctions/halts explicitly. Never infer missing events from OHLCV bars.
  5. Detect feed defects. Check sequence gaps, duplicates, out-of-order records, invalid types, stale updates, zero/negative sizes, crossed books, implausible prices, and session discontinuities. Quarantine suspect intervals; do not patch them silently.
  6. Synchronize clocks. Preserve source time and local time. Quantify precision, timezone, clock source, synchronization error, and timestamping point. Never compare unlike clocks without documenting the uncertainty.
  7. Build reproducible partitions. Partition by venue/date/instrument or a workload-appropriate key. Keep deterministic sort order and schema versions. Ensure reruns over the same raw input generate identical outputs.
  8. Test at event boundaries. Include session open/close, daylight-saving transitions, halts, auctions, instrument rolls, sequence resets, corrections, and partial files.
  9. Publish data quality metrics. Report coverage, gap rate, duplicate rate, stale rate, parse failures, recovery intervals, and last trusted sequence per channel.

Guardrails

  • Do not forward-fill missing book updates across an unknown sequence gap.
  • Do not use local receive time as exchange event time or vice versa.
  • Do not combine data from multiple vendors without mapping differences in trade conditions, corrections, and timestamp semantics.
  • Do not drop outliers without preserving a quarantine log and reason.
  • Avoid floating-point prices as primary keys; store integer ticks or fixed-point values.

Deliverable

Provide source inventory, schema contract, transformation steps, defect policy, recovery semantics, test cases, quality report, and replayable provenance from normalized row to raw source.

Further reading

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.