Evidence before interpretation
Data and research methodology
How we connect an economic release to a recorded market move and, where a public case study exists, to broker-reported execution. This page defines the sources, calculations, publication controls and limits behind ToxicTraders research.
Scope
Three records, kept separate.
An economic number, a market quote and a broker fill are different observations. We do not treat one as proof of another. Public pages connect them by timestamp while retaining their separate provenance.
01 · Economic releases
What the numbers mean and where they come from.
Actual and official source
Actual is the published value stored for a scheduled calendar event. Each permanent report page identifies and links the responsible publisher—for example, the US Bureau of Labor Statistics for CPI and PPI, or the University of Michigan for its Surveys of Consumers.
The official publisher is authoritative for the indicator definition and released value. ToxicTraders is not an official statistics agency and does not replace the source publication.
Forecast is not official data
Forecast is a third-party consensus estimate stored with the calendar event. It is not produced by the official agency. Different calendar providers may show different consensus values, so every comparison should be read against the forecast displayed on that page—not as a universal market forecast.
Previous is the prior value held in the event record and may differ from an earlier publication if the series was revised.
Comparison used on report pages
surprise = actual − forecast“Below”, “above” and “matched” describe this arithmetic comparison only. They are not trading recommendations. Units are normalized for display, but values are not seasonally transformed or economically reinterpreted by the page.
Every mapped component required by that report must have a numeric Actual value before a snapshot can be built.
The automated builder waits 10 minutes after the scheduled release before evaluating a complete snapshot.
Report pages retain the latest three complete releases within the configured three-month history window.
Draft, partial or failed builds are never returned by the public research API.
02 · Recorded market data
Ticks are aligned to one UTC release timestamp.
Economic-report charts use timestamped Dukascopy bid ticks stored in ClickHouse as a neutral market observation. The stored schema also contains ask and available volume fields. This reference series is independent of the broker-history record used for a public execution result.
For each selected instrument, the query window begins 30 seconds before the release and ends five minutes after it. Millisecond timestamps are converted to an offset from the release time:
offset ms = tick timestamp ms − release timestamp ms- Pre
- Last recorded bid at or before the release timestamp.
- +5s / +30s / +1m / +5m
- Bid from the stored tick nearest to each requested horizon.
- Chart line
- Observed bid path; it is not an OHLC reconstruction or a broker fill series.
- Completeness gate
- At least 20 ticks, including observations on both sides of the release.
03 · Public execution case studies
What “verified result” means on this site.
Frozen broker-history snapshot
The published symbol, side, volume, prices, timestamps, commission, swap, fee and P&L originate from a stored closed-order snapshot.
Revalidation before display
The snapshot is matched again to the owner-scoped application order and cached broker history. A missing, open or inconsistent order is not returned publicly.
Execution telemetry
Where recorded, latency, ping, spread and slippage come from PostgreSQL execution-metric or trading-journal rows linked to the same internal order.
Recorded order window
ClickHouse order ticks are grouped into one-second buckets using the last bid and ask in each second. IN and OUT labels retain the exact broker fill price and original fill timestamp.
Explicit editorial publication
A result must be verified, selected for public display and assigned a permanent published URL. Publication is an editorial opt-in, not an automatic export of user history.
Public evidence, private identity
What is deliberately withheld
User, account, broker, server, ticket and raw source identifiers are excluded or anonymized. This limits independent replication at the original account, but protects account security and customer privacy. Public labels such as “Recorded feed 1” must not be read as broker identities.
Audited storage snapshot · July 22, 2026 UTC
The current research corpus.
These are dated database checks, not live counters and not claims about future coverage.
PostgreSQL is the system of record for calendar events, report definitions, publication state, broker-history snapshots and execution metrics. ClickHouse stores the high-volume market and order tick series.
04 · Quality and corrections
A failed update preserves the last known-good page.
Atomic snapshot
Measures, calculated reactions and tick series are stored together as one versioned release snapshot.
No partial replacement
If a required measure or market window is incomplete, the build fails and cannot silently replace a published snapshot.
Publication state
Ready and Published are separate states. Auto-publication can be enabled per report; administrators can also review and publish manually.
Rebuilds
A forced rebuild increments the internal data version. Material public corrections should update the page date and preserve a clear explanation.
Read before using the data
Known limitations.
- 01
Consensus forecasts are third-party estimates and can vary between providers or change before release.
- 02
The reference tick series represents one data source, not every venue, broker or executable price available to a trader.
- 03
Nearest-tick horizon values are observations, not guaranteed prices at precisely +5, +30, +60 or +300 seconds.
- 04
Public execution pages are curated case studies. They are not a complete trading history, audited performance statement or representative sample of all outcomes.
- 05
Broker fills depend on liquidity, account conditions, order size, spread, slippage, routing and rejection rules. A historical fill is not reproducible by default.
- 06
Temporal proximity does not prove that an economic release caused the entire observed move.
Questions or a possible data error? Email support@toxictraders.com with the page URL, release time and disputed field.