Glossary, without the bluff.

Aggressive flow being eaten by passive size at a level without price moving. Reads as a stack of large bid- or ask-side prints with no progress — the ledger of who actually wants the level.

Volatility compression detector. Fires when realised range collapses below historical norms, predicting expansion.

VWAP computed from a chosen anchor (session open, prior swing, news event) instead of the day. Useful for measuring distance from a reference participants actually care about.

The continuous two-way negotiation between buyers and sellers that produces price. Reading the auction means reading where size traded, not just where price went.

Price extends beyond a reference (range high, prior day high, ORB high), then closes back inside within the same or next bar. Failed auctions are mean-reverting — the rotation back is the trade.

Tamper-evident record of every desk decision, gate evaluation, and order — replayable forever.

Reconstructing any past trading decision by replaying the audit log against the strategy version that was live at the time. The thing that makes 'show me why you did this' answerable.

Testing a strategy on historical data. Easy to do; hard to do honestly. Walk-forward and out-of-sample matter more than in-sample fit.

The difference between the best bid and best offer. Tightest in liquid futures (ES often 1 tick); widens with vol and time-of-day.

Mandatory 24-hour platform lock after a -3R day, plus 0.5x size on the next 5 trades. Forces the cooling-off that tilt argues against.

Entry plus pre-attached stop and target as one OCO group. Eliminates the most common discretionary error: forgetting to place the stop after entry.

Annualised return divided by max drawdown over a rolling 36-month window. Penalises depth of pain, not just volatility.

Slicing trades into buckets — by setup, regime, time-of-day, override status — and computing expectancy inside each. Where most edge actually lives or dies.

The Swing Desk's combined score across regime fit, setup quality, location, risk clarity, and edge persistence. Z-normalised so cross-symbol comparison is meaningful.

Cumulative Volume Delta — running sum of (buy-aggressive − sell-aggressive) volume. Divergences with price are leading indicators of exhaustion.

The signed difference between aggressive buy and sell volume on a bar. Positive delta = buyers lifting offer; negative = sellers hitting bid.

The composite-z output, rendered to a 0-100 scale, that ranks setups on the Swing Desk. Combines regime fit, setup quality, location, risk clarity, and edge persistence.

Peak-to-trough decline in equity. Reported as max-DD (worst observed) and current-DD (active).

Deflated Sharpe Ratio. Adjusts naive Sharpe for the number of trials, return non-normality, and sample size — i.e., the bias farm.

A positive expectancy you can defend with statistics — not a feeling, not a backtest curve.

Pattern where state is derived from an append-only log of events rather than mutated in place. The entire desk is event-sourced, which is why every decision is replayable.

Average R-multiple per trade. The headline number that tells you if a setup is worth taking.

How close your actual fill price was to the price you saw when you decided. Measured per-order; aggregated to per-broker scorecards in the journal.

The Nexural manifesto: statistics over stories, receipts over marketing, built in the open, regime over signal, make leaving free. Filters every commit and roadmap item.

GO / NO-GO indicator combining delta, CVD, and tape pressure. The final confirmation layer for STS entries.

Fear of missing out. The single most expensive emotion in trading; the trades you take to avoid it almost always violate your own size and entry rules.

Bar chart variant showing volume traded at every price within each bar — bid vs ask side. A microscope on the auction.

Second derivative of option price with respect to underlying. The rate at which delta changes. Gamma exposure of dealers drives intraday pinning and acceleration around large strikes.

Gamma Exposure. Aggregate dealer gamma at each strike. Positive GEX regimes mean-revert intraday; negative GEX regimes trend and accelerate.

Retrieval-Augmented Generation constrained to cite specific source documents (your trades, journal, indicators). Sage AI refuses to answer when the grounding is too weak.

Hidden Markov Model. Used by the Regime Matrix to estimate the probability the market is in each of six discrete states, given observable features.

High / Low Volume Nodes on a profile. HVNs are absorption zones; LVNs are vacuums where price travels quickly.

A large resting order that displays only a fraction of its true size. Detected by repeated refills of the same level after fills clear it.

The volatility number that, plugged into Black-Scholes, makes the model price match the market price. Forward-looking; reflects what the option market expects, not what was realised.

Initiative activity pushes price out of value (trend); responsive activity defends value (rotation). Reading which side is in control is the first read of every session.

Jurik Moving Average. Adaptive smoother with phase compensation — minimal lag relative to standard MAs at the same noise level.

Position-sizing fraction that maximises long-run log-wealth. In practice trade fractional Kelly (often 0.25×) because edge is uncertain.

Time from event to action. QuantFlow budgets latency across five stages and surfaces violations in the audit log.

Allocation of allowable delay across the decision pipeline: data → indicator → signal → order → fill. QuantFlow surfaces breaches in the audit log.

How much size you can move without moving price. Function of depth at each price level and rate of replenishment.

A fast move through a cluster of resting stops, followed by reversal. The sweep is the move that takes liquidity; the reversal is the trade.

Managed Account Ratio = annual return / max drawdown. Quick proxy for risk-adjusted performance; less rigorous than DSR but useful.

Maximum Favorable / Adverse Excursion — best and worst unrealised P&L during a trade. Powerful for analysing exit quality.

Public Nexural repo. The NinjaTrader 8 ↔ QuantFlow ZeroMQ bridge, plus the C# helpers to keep the strategy loop deterministic.

The price range traced out in the first N minutes of cash session (commonly 5, 15, or 30). Boundary tags become reference levels for the rest of the day.

Nexural's opening-range indicator. Layers participation quality, regime context, and prior-day reference to filter genuine breakouts from third-poke fakes.

The live ladder of resting bids and offers at every price. Useful for seeing where size is parked; less useful as a predictor — most book size is iceberg, spoof, or pulled.

The signed sequence of executed trades. Different from the order book (resting limits): order flow is what cleared, not what was offered. Footprint, delta, and CVD are all cuts of the same data.

Taking a trade against the desk's recommendation. Audit weekly; the override cohort is where most hidden bleed lives.

Weekly review of every trade taken against the desk's recommendation, grouped by setup × regime. Most hidden bleed lives in the override cohort.

Order executes for less than the requested quantity. Common in thin books; the journal tracks the unfilled remainder so risk math stays honest.

Average winner divided by average loser. Combined with win rate, gives expectancy. Most retail traders have a high win rate and a payoff ratio < 1 — the slow bleed.

Probability of Backtest Overfitting. Combinatorial check that the in-sample winner of a parameter grid generalises out-of-sample.

Point of Control. The price at which the most volume traded over a profile period. A magnet, not a line in the sand.

Before sending the order, name the three ways this trade fails. Forces the kill criteria into consciousness while you can still walk away.

A measurable behavioural target (e.g. 'no entries in vol·rising for two weeks') instead of a P&L target. Process goals are controllable; P&L targets are not.

Gross winning P&L divided by gross losing P&L. Above 1.5 is good, above 2 is excellent — but trivially gameable on small samples, so always pair with N.

Volume-normalised momentum. The WHEN indicator in STS — fires when participation rotates with the read.

The strategy-and-automation layer. Runs strategies inside the user's NinjaTrader 8 with a ZeroMQ bridge, six risk gates, and a tamper-evident audit log.

Trade P&L expressed as multiples of initial risk. A +2R trade made twice what you risked. Normalises across symbols and sizes.

Sage AI's posture when grounding evidence is insufficient: it says 'not enough data' and stops. Hallucination is a worse failure than silence.

The market's current behavioural state. Six discrete regimes: trend·long, trend·short, rotation, vol·rising, vol·crush, regime·shift.

Hard filter that blocks orders whose strategy is incompatible with the current regime. One of QuantFlow's six gates.

HMM-based regime classifier with confidence scores. Drives the regime gate in QuantFlow and the regime-fit component of the desk score.

The property that a past decision can be re-derived from the same inputs and code. Non-negotiable for a regulated workflow; enabled by event sourcing + versioned strategies.

Re-entering immediately after a loss, sized larger than rule, to 'get it back'. The single most expensive trade pattern in the book.

Two-way auction inside an established value area. Edges of value are the rotation boundaries.

Relative Strength Index. Default flavour is mediocre. Quant RSI adds trend-adjusted thresholds and volume normalisation.

Grounded RAG-based copilot. Reads your trades, journal, indicators, audit log, and playbooks. Cites every claim, refuses when grounding is weak.

Public Nexural repo. Statistics + risk math primitives — DSR, PBO, expectancy, R-multiple, sizing. 506+ tests across the open stack.

Number of trades behind a statistic. Below ~30 the noise dominates the signal; below ~100 most claims are unverifiable. Always report N alongside expectancy.

A repeatable trade pattern with defined entry, stop, and target. Tagged in the journal so you can compute setup × regime cohort statistics.

(Excess return) / (return volatility). Useful but easily fooled — see DSR for the deflated version that survives p-hacking.

QuantFlow's pre-trade checklist, evaluated in series: regime, sizing, drawdown, correlation, latency, schedule. Failure of any gate blocks the order and writes to audit.

The difference between the price you intended to trade and the price you got. Worse on stops, in vol, and at session turns.

Sharpe variant that penalises only downside deviation. Closer to what risk actually feels like — upside vol isn't risk.

Posting and pulling large size to manufacture the appearance of demand or supply. Illegal in regulated venues; still common enough that you should never trust the book at face value.

Volatility compression. Often (not always) followed by expansion. ACE Squeeze quantifies it.

Whether a process's statistical properties stay constant over time. Markets are non-stationary — backtests assume the past resembles the future, which is why walk-forward matters.

The five-gate process for moving a strategy from research to live: walk-forward DSR, PBO, MAR, paper-traded sample, peer review. No exceptions.

Sage Trading System. The reading method: WHERE (Volume Profile) → WHEN (QPulse) → GO/NO-GO (Flow Pro).

The ranked, regime-gated dashboard of the highest-quality setups across the watchlist, scored by composite-z.

The time-and-sales feed of every individual trade print. Reading the tape is reading the auction at maximum resolution.

The shape of implied vol across expirations. Backwardation (front > back) signals stress; contango (front < back) is the calm-market default.

Emotional state in which discipline collapses. Diagnose with the override audit; treat with the blowup protocol.

Public Nexural repo. Execution primitives — order types, bracket math, fill modeling, latency tracing. The engine room of the desk, open-sourced.

Public Nexural repo. Multi-agent research framework — analyst, risk officer, devil's advocate roles cooperate on a research question with explicit disagreement records.

Multi-timeframe trend classifier. Outputs a discrete state (up, down, neutral) with confidence per timeframe — input to the regime gate and the desk score.

Value Area High / Low. Boundaries of the price range that contained 70% of session volume. Rotation edges.

Cross-Greek: change in delta as implied vol changes. Vanna flows are why a vol crush can drive directional moves that look unrelated to news.

Public Nexural repo. Research notes, backtest harnesses, and the workspace where strategies live before promotion to QuantFlow.

Standard deviation of returns. Realised (historical), implied (from options), or model-based (e.g., GARCH).

Histogram of volume traded at each price over a defined period. The single most important auction indicator.

Volume-Weighted Average Price. Institutional benchmark; price below VWAP often means buyers are losing the day.

Rolling out-of-sample backtest where each window is fit on past data and evaluated on the next chunk. Defends against curve fitting.

Percentage of trades that closed profitable. Almost meaningless without payoff ratio — a 35% win rate at 3R average wins beats a 65% win rate at 1R.

Lightweight messaging library. QuantFlow uses ZeroMQ to decouple the strategy layer from NinjaTrader 8 — the bridge that keeps latency below 60ms p99.