The terms this research uses

Clock Regime

A market-state label that is substantially a reading of the hour rather than of the market — revealed when removing the deterministic intraday volatility pattern from the input causes the state to lose most or all of its occurrences.Coined in: Is my volatility regime just telling me the time?

Conditional Independence Of Fills

The testable property that your execution quality depends on nothing about you except order size and timing — the condition under which the discretionary-intervention hypothesis has nowhere to live, and the form in which its failure can be put in front of a broker.Coined in: Is my broker trading against me?

Costume Variable

A constructed feature that appears to measure a new mechanism but is a re-encoding of a simpler variable already in the model — revealed when adding it to a model that already contains the simpler one makes out-of-sample performance worse rather than better.Coined in: Do order blocks predict anything?

Geometry Ghost

An effect that is real, large and statistically overwhelming, and is nonetheless produced entirely by the geometry of the measurement — confirmed when a surrogate matched on shape alone reproduces it at the same magnitude.Coined in: Does a wick rejection mean anything?

Recency Mirage

The appearance of durable structure produced entirely by short-range recency — a level looks as though it remembers its own history, when every part of the signal is carried by what happened there most recently.Coined in: The level that remembered nothing

Reconstruction Debt

The accumulated borrowings a backtest takes against reality — each simplification reasonable alone — which live trading calls in all at once, in whichever place the model borrowed most.Coined in: Why do backtests fail in live trading?

Significance Without Size

A finding whose p-value is small only because the sample is large: a real lean of a fraction of a percent, statistically undeniable and economically negligible — which is why an effect size must sit beside every p-value before anything is called an edge.Coined in: The edge that was real and worth half a pip

Silent Guard

A check that has never failed, whose silence is mistaken for evidence that the condition it guards is sound — when the threshold may simply sit where nothing real could ever cross it.Coined in: The floor that never fired

Surrogate Inversion

A detector firing fewer times on real data than on surrogates built to destroy the very structure it detects — the null is not merely unbeaten but beaten in the wrong direction, which bounds what the detector may claim until the reason is established.Coined in: The break that cleared neither bar

The Absent Denominator

The population a survivor statistic was actually computed over. Every figure from a survivor-only record is a numerator whose denominator left no row behind — and the audit for survivorship bias is the attempt to restore that denominator, or the admission, made in writing, that it cannot be restored.Coined in: Does my backtest have survivorship bias?

The Asymmetric Clock

A limit whose anchor advances on one schedule while the breach test runs on another — the floor sampled daily, the account compared to it continuously — so a day can close above the limit and still have ended.Coined in: Why was my account breached in profit?

The Binding Barrier

Whichever of an evaluation's overlapping limits sits nearest right now — daily loss in the morning, trailing floor after a new high — rarely the rule quoted in the marketing, and always the one that actually kills the account.Coined in: What causes a prop firm drawdown breach?

The Chance Floor

The rate at which a pattern would occur under a random walk carrying the same spread and volatility — the baseline an observed rate must beat before it is evidence of anything, and below which the observation argues against the pattern rather than for it.Coined in: Do brokers hunt your stop losses?

The Closed Ledger

A record generated from a single source reconciles with itself by construction, so its internal consistency is evidence of nothing beyond its own arithmetic — corroboration requires a second record produced independently of the first.Coined in: What can a trading statement prove?

The Coherence Floor

The sampling interval below which two instruments' relationship stops being defined: finer than the floor, most of what a chart shows is asynchrony — prints that never co-occurred — and a reading taken there is about the sampling, not the relationship.Coined in: Why do correlated pairs decouple on low timeframes?

The Cost Floor

The property that every completion claim must be backed by an artifact its maker cannot counterfeit — elapsed work, bytes read, a hash that must match. Below the floor, done is a sentence; above it, done is a receipt.Coined in: What stops an AI agent from faking its results?

The Definition Gap

The distance between the account curve a trader watches and the one the firm's engine computes — the source of breaches that surprise, where both sides were right about the number they were watching.Coined in: Is a drawdown limit on balance or equity?

The Feed Zero

A recorded, dated baseline of what your feed does under ordinary conditions, computed with statistics fixed in advance — the reference without which no claim about unusual feed behaviour can be tested.Coined in: Is my broker's feed honest? How to test it

The Frictionless Account

The market a demo simulates — instant fills at the displayed price, no spread paid, no slippage, no rejections — whose absent friction is what a demo-tuned edge is often made of.Coined in: Why does my EA work on demo but not live?

The Knowable Set

The set of facts actually knowable at a simulated instant — the boundary an honest simulation must never leave, and the reconstruction that replaces the unanswerable question of whether data is accurate.Coined in: Why does my backtest use data that did not exist?

The Luck Horizon

The number of trades below which a record cannot be distinguished from a lucky zero-edge process: not a universal figure but a function of the edge's size relative to its per-trade dispersion — fine edges push the horizon into the thousands, and a record shorter than its horizon is testimony, not evidence.Adopted from poker's working vocabulary, where the same idea names the point at which luck's share of results falls within tolerance.Adopted in: How many trades prove a trading edge?

The Orphan Number

A figure on a dashboard whose derivation nobody can produce on demand — source, filters, joins, definition, refresh time — and which therefore cannot be distinguished from a wrong number by anyone reading it, however correct it happens to be.Coined in: Why do two dashboards show different numbers?

The Planted Failure

The only evidence a gate works: a defect introduced on purpose and watched being caught. A gate is a claim to be falsified, and a check that has never failed has never been tested.Coined in: How do you tell if a CI gate can actually fail?

The Recompute Test

A track record is verifiable exactly when a third party could recompute it — which requires the trade list it was computed from, the accounting rules used, and enough provenance to pin both to a real account.Coined in: How do you verify a trading track record?

The Search Receipt

The four artifacts — trial count, every configuration's performance, the data vintage each trial saw, and the order decisions were made in — without which overfitting statistics report reassurance instead of a result.Coined in: How do I test a backtest for overfitting?

The Slippage Lean

The asymmetry between favourable and unfavourable slippage in your fill distribution: symmetric slippage is noise you can budget for; a lean is a transfer, and it compounds with every trade placed.Coined in: What is slippage really costing me?

The Spread Fiction

The gap between a backtest's assumed constant spread and any spread that ever existed: the live spread is usually not too wide — the backtested one was narrower than reality ever was, and the gap is the size of that fiction.Coined in: Why is my live spread wider than my backtest?

The Standing Gate

Every forum asks whether it may consider you before it considers what happened to you, and a dispute far more often ends at that first question than at the second.Coined in: Where a funded-account breach dispute can go

The Survivor's Fill

A fill that exists because a counterparty chose to accept it, in a market structure where it could have declined. Execution statistics computed on fills alone describe the accepted half of an order flow and inherit whatever selected it, which is why a fill-only record cannot price its own rejections.Coined in: Why do my forex orders get rejected?

The Two Histories

What two data feeds give you for the same market: each recording partially an artifact of pipeline decisions, so their disagreement measures how much of a backtest's result depends on the recording rather than the market.Coined in: Why do two data feeds give different backtests?

The Vantage Limit

The boundary of what can be established from where you stand: some claims — above all, claims about a counterparty’s intent — are not merely unproven from client-side data but undecidable from it, and an honest verdict says so rather than guessing. The verdict that marks the boundary is “Not measurable”, a category standard fact-check taxonomies lack because they presuppose every proposition is decidable.Coined in: The refutation index

Window Luck

An effect that appears significant on the window it was discovered in and scatters around zero when the identical test is run across other windows of the same history — diagnosed by replication across windows, never by any property of the original result.Coined in: Do round numbers matter in forex?

Wrong-Way Significance

An effect whose confidence interval excludes zero and whose sign contradicts the hypothesis it was registered to test — statistically real, and evidence against the belief rather than for it.Coined in: Does COT positioning predict reversals?

The glossary carries the established terms these build on; the research index carries every article, coining or not. Machine readers: this register is also served as plain markdown, and every article has a markdown mirror linked from its own page.