Point-in-Time Integrity
Verify that a backtest cannot use information before it was actually available.
Evidence before conclusions
SCYLLA
Point-in-Time Market Data & Research Integrity
We audit quantitative research pipelines for timestamp leakage, invalid data handling, and reproducibility.
SCYLLA helps quantitative teams verify that their research uses information that was actually available at the decision time — with explicit treatment of timestamps, gaps, stale observations, duplicates and data lineage.
Audit surface
What We Audit
Focused checks for the temporal and data-handling defects that can invalidate otherwise careful quantitative research.
Timestamp & Timezone Alignment
Check event timestamps, bar boundaries and timezone conversions for temporal shifts.
Missing & Stale Data
Identify hidden gaps, unsafe forward fills and observations that remain in use after they should be considered stale.
Duplicates & Replays
Detect duplicated records, reconnect replays and conflicting observations.
Data Provenance
Trace where observations came from and how they were transformed.
Reproducibility
Check whether the same inputs and rules rebuild the same research dataset.
Backtest Fill Timing
Check whether simulated executions use prices that would actually have been available after the trading decision.
Split Contamination
Check whether validation or holdout information leaks into research or training.
A narrow, evidence-led process
How It Works
- 01
Scope
We agree on a narrow dataset, public repository, or pipeline component.
- 02
Audit
SCYLLA tests temporal integrity, data handling and reproducibility.
- 03
Evidence
Every finding must be reproducible and tied to a specific rule or artifact.
- 04
Report
You receive a concise PASS / FAIL report with findings, impact and recommended fixes.
Current offer
Free Design-Partner Audit
We are currently offering a limited number of free temporal-integrity audits to quantitative teams and open-source trading platforms.
The initial audit can begin with a public repository and a small synthetic test fixture.
No private strategy code, exchange credentials, wallet keys or customer data are required for the initial public-repository audit.
There is no sales obligation.
Start with a narrow, reviewable scope.
Request a Free AuditWhy this matters
A convincing backtest can still be wrong.
A backtest can look profitable because of a one-row timestamp shift, future-bar access, stale data, an unsafe fill assumption or an unreproducible vendor transformation.
Finding one such defect early can save weeks or months of invalid research.
Tangible evidence
Audit Deliverable
A clean result is also a valid audit outcome.
- Technical scope
- PASS / FAIL verdict
- Critical / major / minor findings where applicable
- Reproduction steps
- Impact explanation
- Recommended remediation
- Retest status when applicable
Clear boundaries
What SCYLLA Does Not Sell
We audit whether the evidence behind quantitative research can be trusted.
- Trading signals
- Guaranteed returns
- Guaranteed alpha
- Individual whale identities
- Claims of hidden stop-loss orders
- Retrofitted backtest results
Data minimization
Security Starts With Scope
- Start with public code whenever possible.
- Do not request trading credentials.
- Do not request withdrawal credentials.
- Do not request wallet seed phrases.
- Minimize private dataset access.
- Define audit scope before any private material is shared.
Independent research infrastructure
About SCYLLA
SCYLLA is an independent market-data and research-integrity project focused on point-in-time correctness, reproducible datasets and evidence-based quantitative research.