Skip to main content

Cascade Analysis Terminology

Document Purpose​

This glossary defines every specialized term used in the CL Cascade Analyzer discussion and blueprint. Terms are grouped by domain and ordered from foundational concepts to advanced analytical constructs. Each entry includes the term, a plain-language definition, the mechanical reality behind it, and where applicable, the quantitative criteria used in the Cascade Analyzer.

1. Market Structure Fundamentals​

Order Book​

The collection of all resting limit orders (bids and asks) at every price level for an instrument. On CL futures, the book is relatively thin (20–80 contracts per level) compared to ES (500–2000+ per level). The Cascade Analyzer tracks the top 20 price levels on each side.

Book Depth​

The total quantity of resting orders across multiple price levels on one side of the order book. Measured as the sum of all resting bid sizes (bid depth) or ask sizes (ask depth) across a specified number of levels. The Cascade Analyzer sums the top 20 levels and tracks this value in a rolling window to compute rate-of-change.

Best Bid / Best Ask (BBO — Best Bid and Offer)​

The highest-priced resting buy order (best bid) and the lowest-priced resting sell order (best ask) at any given moment. The difference between them is the spread. The BBO is the most immediately actionable liquidity in the market.

Bid-Ask Spread​

The price difference between the best ask and the best bid. On CL, normal spread is 1–2 ticks ($0.01–$0.02). On ES, normal spread is 1 tick (0.25 points). Spread behavior is the single most reliable indicator of market maker participation and cascade severity.

Price Level​

A single discrete price at which orders can rest. On CL, each price level is separated by $0.01 (one tick). On ES, each level is separated by 0.25 points (one tick). When we say a sweep consumed 10 price levels on CL, that means it moved through $0.10 of price.

Tick​

The minimum price increment for an instrument. CL ticks at $0.01 with a value of $10 per tick per contract. ES ticks at 0.25 points with a value of $12.50 per tick per contract. MES ticks at 0.25 points with a value of $1.25 per tick per contract.

Resting Order / Passive Order / Resting Liquidity​

A limit order sitting in the order book waiting to be filled. These orders provide liquidity — they are available for aggressive orders to trade against. The term "passive" reflects that the order does not initiate the trade; it waits. In the Cascade Analyzer context, resting bid liquidity is what gets consumed during a sell cascade.

Aggressive Order / Market Order​

An order that crosses the spread to execute immediately against resting liquidity. A market buy lifts the best ask; a market sell hits the best bid. Aggressive orders consume liquidity and move price. Every sweep is composed of aggressive orders eating through successive levels of resting liquidity.

Aggressor Side​

Which party initiated the trade. If a buyer sent a market order that matched against a resting sell limit, the buyer is the aggressor (bid aggressor). If a seller sent a market order that matched against a resting buy limit, the seller is the aggressor (ask aggressor). The Bookmap API exposes this via TradeInfo.isBidAggressor. Aggressor-side identification is essential for determining cascade direction and computing delta.

2. Market Participants and Behavior​

Market Maker​

A participant who continuously posts both bid and ask orders, profiting from the spread while providing liquidity. Market makers are critical to cascade dynamics because their withdrawal (pulling quotes) creates the liquidity vacuum that turns a normal sell-off into a cascade. When market makers detect a cascade pattern (aggressive-to-passive volume ratio exceeding their risk threshold), they pull bids, widening the spread and removing the cushion that normally absorbs aggressive orders.

Institutional Flow / Institutional Order​

Large orders from institutional participants (hedge funds, asset managers, pension funds, sovereign wealth funds). Characterized by patient execution (TWAP/VWAP algorithms), large aggregate size distributed across time, and minimal market impact per individual clip. Institutional flow is distinct from cascade mechanics — institutions accumulate over hours, cascades unfold in seconds.

Retail Flow​

Smaller orders from individual traders. Characterized by smaller clip sizes (1–5 contracts on CL/ES), less execution sophistication, and higher susceptibility to emotional decision-making. In cascade contexts, retail stop-loss orders are frequently the fuel that sustains the cascade once it begins.

Mean-Reversion Algorithm​

An automated trading system that detects price dislocations (like cascade extremes) and trades in the opposing direction, betting that price will revert toward its recent average. These algorithms operate in the 1–50ms latency range and are typically the first participants to buy at a sell cascade's low. They capture the first $0.10–$0.30 of the recovery before human traders can react.

Stacker Algorithm​

An algorithmic pattern where a participant places consistent-sized orders at regular price intervals, creating an artificial appearance of depth. Identified by equal or near-equal sizes at evenly spaced price levels (e.g., 50 contracts every 4 ticks). These orders frequently get pulled before they can be filled, making them unreliable support/resistance. First widely observed around 2010 in electronic futures markets.

3. Sweep and Cascade Terminology​

Sweep​

A broad term for any event where aggressive volume consumes resting orders across multiple price levels within a short time window. The Bookmap built-in Sweeps indicator defines this as: within X seconds, Y contracts traded across Z or more price levels. A sweep can be a single aggressive clip, a stop cascade, or anything in between. Not all sweeps are cascades.

Sweep Event​

A specific instance where the Sweeps indicator criteria are met. Marked on the chart with a colored circle and volume annotation (e.g., @141 for 141 contracts). Each marker represents one qualifying burst. A chart can have many sweep events, most of which are noise (normal aggressive bursts), with only rare genuine cascade events.

Normal Aggressive Burst​

A sweep event that meets the indicator's threshold criteria but is NOT a cascade. It's a single instance of aggressive volume crossing several price levels — a large clip, a stop trigger, or an institutional execution — that does not trigger a chain reaction. The book absorbs it, spread doesn't blow out, and price stabilizes quickly. These constitute the majority of sweep markers on any chart and are the primary source of noise.

Aggressive Clip​

A single execution or tightly grouped set of executions from one aggressive order. When an institutional trader sends a 50-contract market sell, it may fill across 3–5 price levels as a single aggressive clip. This is normal market activity, not a cascade. The term emphasizes that the volume originates from one decision/order, not a chain reaction.

Cascade Event / Liquidity Sweep Cascade​

The specific phenomenon the Cascade Analyzer is designed to detect. A cascade is a self-reinforcing chain reaction where an initial aggressive order triggers stop-loss orders, which become new aggressive orders, which trigger more stops, while market makers simultaneously withdraw liquidity, creating a vacuum that accelerates the price move. Distinguished from normal sweeps by multi-wave structure, spread blowout, book thinning, and temporal density. The March 12 CLJ6 event at 10:46 — price dropping $2.00 in 3 seconds — is a textbook cascade.

Liquidity Vacuum​

The condition that develops during a cascade when market makers pull their resting orders faster than the aggressive side can consume them. Price must fall (or rise) further to find willing counterparties. Visible in the order book as the bid (or ask) side thinning dramatically ahead of the aggressive flow. In the millisecond BBO screenshots from the CLJ6 analysis, the bid stepped down in discrete jumps with almost no fills between levels — that's a liquidity vacuum.

Stop-Loss Cascade / Stop Cascade​

The chain reaction mechanism that drives most cascade events. Traders holding long positions place stop-loss sell orders below their entry prices. When price drops to those levels, the stops trigger as market sell orders. Those sells push price lower, triggering more stops at the next level, creating a self-reinforcing downward spiral. The cascade continues until either the stop clusters are exhausted or genuine absorption occurs.

Sweep Gap​

The price displacement created by a cascade — the distance between the pre-cascade price and the cascade extreme. On the March 12 CLJ6 event, the sweep gap was approximately $2.00 ($95.80 to $93.80). The hypothesis under investigation is that sweep gaps tend to revert (price returns to pre-cascade levels) with statistically significant frequency.

Sweep Gap Reversion​

The tendency for price to return toward or beyond the pre-cascade level after a cascade event completes. This is the core trading hypothesis: that cascades create temporary dislocations (driven by mechanical stop triggering and liquidity withdrawal, not fundamental repricing) and price subsequently reverts. Requires statistical validation across a large sample of cascade events before it can be treated as a tradeable edge.

4. Cascade Anatomy and Wave Structure​

Cascade Wave​

A single burst of aggressive volume within a larger cascade event. Each wave represents one discrete episode where aggressive orders consumed resting liquidity across multiple price levels within the burst time window (150ms in the Cascade Analyzer). A full cascade is composed of multiple successive waves. On the March 12 CLJ6 event, we identified 10–15 distinct waves with volumes ranging from 7 to 297 contracts per wave.

Wave Decomposition​

The process of breaking a cascade into its constituent waves and analyzing each one separately. Metrics per wave include: volume, price levels consumed, duration, and inter-wave pause time. Wave decomposition reveals whether a cascade is accelerating (increasing volume per wave, shorter pauses) or decelerating (decreasing volume, longer pauses).

Temporal Density of Waves​

How closely spaced the cascade waves are in time. High temporal density (waves firing every 100–500ms with minimal gaps) indicates a severe cascade with continuous chain-reaction dynamics. Low temporal density (waves separated by 1–2+ seconds) may indicate the cascade is losing momentum. This metric is the primary differentiator between a genuine cascade (dense cluster of waves) and noise (isolated single bursts scattered across time).

Inter-Wave Pause​

The time gap between the end of one cascade wave and the start of the next. Short pauses (< 500ms) indicate the cascade is actively self-reinforcing — each wave's stop triggers are immediately feeding the next wave. Longer pauses (500ms–2s) suggest the cascade may be transitioning to absorption. Pauses exceeding 2 seconds indicate the cascade has likely exhausted.

Cumulative Price Displacement​

The total price distance from the first wave's starting price to the most extreme price reached during the cascade. For the March 12 event: $95.80 start to $93.80 extreme = $2.00 cumulative displacement. The Cascade Analyzer requires a minimum displacement threshold ($0.30 on CL, 4 points on ES) to distinguish cascades from normal aggressive activity.

Staircase Pattern​

The visual pattern created when the BBO steps down (or up) in discrete jumps during a cascade, visible at millisecond resolution. Each step represents a price level being consumed, followed by a brief pause, then the next level. The staircase pattern confirms that the move is not a single large order but a sequential chain of triggers eating through successive levels with gaps between them.

5. Pre-Cascade Detection​

Book Thinning​

The reduction of resting order depth on one or both sides of the book, typically occurring 500ms–2 seconds before a cascade initiates. Market makers and algorithmic liquidity providers begin withdrawing their quotes as they detect emerging directional pressure or increased toxicity in the order flow. The Cascade Analyzer detects book thinning when the top-20 depth drops by more than 30% within a 500ms window relative to the rolling average.

Depth Rate-of-Change​

The percentage change in total book depth (top 20 levels) measured over a short time window. Computed as: (currentDepth - depthNmsAgo) / depthNmsAgo * 100. Negative values indicate the book is thinning. The Cascade Analyzer computes this every 100ms comparing current depth to depth 500ms ago. Values below -30% trigger the BOOK_THINNING flag.

Spread Blowout / Spread Expansion​

The rapid widening of the bid-ask spread beyond its normal baseline range. On CL, normal spread is 1–2 ticks; a blowout is 5+ ticks. The Cascade Analyzer flags a blowout when the current spread exceeds 3x the session EMA baseline AND is at least 5 ticks wide. Spread blowout is the most reliable real-time confirmation that market makers have withdrawn and a cascade is in progress.

Spread Expansion Rate​

The velocity at which the spread is widening, measured in ticks per second. Computed by comparing the current spread to the spread 100ms ago. A high expansion rate (spread widening by 10+ ticks per second) indicates market makers are pulling quotes rapidly — a strong cascade confirmation signal.

Spread Baseline​

The exponential moving average (EMA) of the bid-ask spread over the session, used as the reference point for blowout detection. Updated with a very slow alpha (0.001) so it represents the session's normal spread conditions rather than being skewed by cascade events. On CL RTH, this typically settles to 1.0–1.5 ticks.

Pre-Cascade State (PRE_CASCADE)​

A state in the Cascade Analyzer's state machine indicating that early warning conditions are present (book thinning detected, spread beginning to expand) but no cascade burst has yet occurred. This state has a 2-second timeout — if no cascade initiates within 2 seconds, the system returns to IDLE. This is the early warning phase that gives a human or automated system the maximum possible lead time.

6. Cascade Detection and Filtering​

Burst Window​

The time interval (150ms in the Cascade Analyzer) over which aggressive trades are accumulated to evaluate whether a cascade wave has occurred. All trades within one burst window are aggregated by volume, price levels consumed, and directionality. At the window's expiration, the accumulated metrics are evaluated against the burst criteria.

Burst Volume Threshold​

The minimum aggregate aggressive volume (25 contracts on CL) that must trade within a single burst window to qualify as a cascade wave. Set based on the instrument's typical book depth — the threshold should be roughly 30–50% of the average resting size at a single level multiplied by the minimum levels threshold.

Burst Levels Threshold​

The minimum number of price levels that must be consumed within a single burst window (4 levels / $0.04 on CL) for the burst to qualify as a cascade wave. This filters out large aggressive clips that fill at a single level or across only 2–3 levels, which is normal market activity.

Directionality​

The percentage of trade volume within a burst window that is on the same aggressor side. A directionality of 80%+ means the burst is predominantly one-directional (almost all sells or almost all buys). Mixed directionality (50/50) indicates two-sided trading, not a cascade. The Cascade Analyzer requires 80%+ directionality for a burst to qualify.

Pending Bursts Buffer​

A temporary storage mechanism that holds qualifying bursts before they're promoted to cascade waves. When the first qualifying burst fires, it's stored in the buffer but no cascade is declared. The second burst within 5 seconds transitions to PRE_CASCADE. The third burst with spread and depth confirmation transitions to ACTIVE_CASCADE. If fewer than 3 bursts accumulate within 5 seconds, the buffer is flushed. This is the compound filtering gate that eliminates noise (isolated single bursts) from genuine cascades (multi-wave events).

Multi-Wave Requirement​

The rule that at least 3 qualifying bursts must fire within a 5-second window, all in the same direction, for an event to be classified as a cascade. This is the single most important noise filter. Normal aggressive bursts are isolated events (1 burst, then nothing). Cascades are chains of successive bursts. Requiring 3+ waves eliminates the vast majority of false positives.

Compound Filtering Criteria​

The full set of conditions that must ALL be met simultaneously for an event to be classified as a tradeable cascade. The four criteria are: (1) multi-wave requirement (3+ bursts in 5 seconds), (2) cumulative displacement threshold ($0.30+ on CL), (3) spread expansion confirmation (2x+ baseline), and (4) book depth deficit (depth below 70% of rolling average). All four must be satisfied. Meeting only 1–3 of the four criteria results in the event being logged for statistics but not flagged on the chart.

Signal-to-Noise Ratio​

In this context, the ratio of genuine cascade events to total sweep markers on a chart. The built-in Sweeps indicator at calibrated settings (0.15s, 30 contracts, 5 levels) produced 20+ markers on the March 12 CLJ6 session, of which only 1 was a genuine cascade. Signal-to-noise ratio = 1:20. The compound filtering criteria aim to bring this to approximately 1:1 — only genuine cascades produce indicators and alerts.

7. Post-Cascade Analysis​

Absorption​

The process by which resting orders and aggressive counter-directional volume halt a cascade's price movement. At a sell cascade's low, absorption manifests as: aggressive buyers appearing, passive bid orders being placed at or near the extreme price, spread beginning to normalize, and price stabilizing (no new lows). Absorption is the mechanical evidence that genuine counterparties have arrived and are willing to hold the level.

Absorption Score​

A composite metric (0.0 to 1.0) quantifying the quality of absorption at a cascade's terminus. Composed of five weighted components: counter-aggressive volume (0–0.30), passive liquidity rebuilding (0–0.25), spread normalization (0–0.20), price stabilization (0–0.15), and delta divergence signal (0–0.10). Higher scores indicate stronger absorption and higher probability of genuine reversal. Scores below 0.3 suggest a dead cat bounce; above 0.7 suggests institutional-quality absorption.

Counter-Aggressive Volume​

Aggressive volume in the direction opposing the cascade. During a sell cascade, counter-aggressive volume is aggressive buying (bid-aggressor trades). This is the most direct evidence of willing buyers at the cascade's low. Distinguished from passive bid placement — counter-aggressive volume represents participants actively and urgently buying, not just posting limit orders.

Passive Liquidity Rebuilding​

The re-establishment of resting limit orders at or near the cascade's extreme price after the cascade pauses. Visible in the order book as new bid sizes appearing at the low (for a sell cascade). Indicates market makers are returning and/or new participants are placing limit buy orders, rebuilding the structural support that was destroyed during the cascade.

Spread Normalization​

The process of the bid-ask spread returning toward its pre-cascade baseline after the cascade ends. Spread normalization indicates market makers are re-establishing two-sided quotes, which restores normal market function. The absorption score awards points when the current spread drops below 1.5x the session baseline.

Price Stabilization​

The cessation of new extreme prices at the cascade terminus. Measured as time since the last new low (sell cascade) or new high (buy cascade). If 500ms+ passes without a new extreme, price is stabilizing. If 2+ seconds pass, stabilization is confirmed. Price stabilization distinguishes genuine absorption (price holds) from temporary pauses within a still-active cascade.

Dead Cat Bounce​

A brief, weak price recovery after a cascade that fails to sustain — price bounces slightly from the extreme but then resumes moving in the cascade direction or simply goes flat. Characterized by low absorption score (< 0.3), minimal counter-aggressive volume, and failure of the spread to normalize. The term comes from the grim observation that even a dead cat bounces if dropped from high enough. Distinguishing dead cat bounces from genuine reversals is the core purpose of the absorption scoring system.

Recovery​

The sustained price movement back toward the pre-cascade level following confirmed absorption. In the Cascade Analyzer state machine, RECOVERY state is entered when aggressive counter-volume exceeds 30% of cascade volume AND book depth is rebuilding. The recovery phase is tracked for up to 30 seconds to measure recovery quality.

Recovery Quality / Recovery Ratio​

The ratio of aggressive counter-directional sweep volume (during recovery) to the cascade's total aggressive volume. On the March 12 CLJ6 event: ~446 contracts of buy-side sweep volume divided by ~1,477 contracts of sell-side cascade volume = 0.30 recovery ratio. A ratio below 0.20 suggests the recovery is almost entirely short-covering (weak, likely to fail). A ratio above 0.40 suggests fresh institutional buying (strong, likely to sustain). This metric directly informs the hold-vs-scalp decision.

Short Covering​

The buying activity from participants who sold short during or before the cascade and are now closing their positions by buying back. Short covering produces aggressive buy volume and upward price movement, but it's mechanically different from fresh buying because it's finite — once the shorts have covered, the buying pressure stops. Recoveries driven primarily by short covering tend to be weaker and more prone to reversal than recoveries driven by fresh institutional buying.

CVD Divergence (Cumulative Volume Delta Divergence)​

A condition where price recovers from a cascade extreme but CVD (the running total of aggressive buy volume minus aggressive sell volume) does not recover proportionally. For example: price recovers 80% of the cascade range but CVD only recovers 40%. This indicates the price recovery is driven more by short covering and passive bid placement than by aggressive new buying. CVD divergence is a warning signal that the recovery may be fragile.

8. Cascade Analyzer State Machine States​

IDLE​

No cascade activity detected. Normal market conditions. The system monitors book depth and spread for early warning signals but no cascade indicators are active on the chart. The majority of trading time is spent in this state.

PRE_CASCADE​

Early warning phase. Book thinning and/or spread expansion detected, but no cascade bursts have fired yet. The system has taken a snapshot of pre-cascade book depth for later comparison. This state has a 2-second timeout — if no cascade initiates, the system returns to IDLE. This is the maximum lead-time signal a human trader can receive.

ACTIVE_CASCADE​

A cascade is in progress. Three or more qualifying bursts have been detected within the time window, compound filtering criteria are satisfied, and the system is actively tracking waves, computing cumulative metrics, and updating indicators. Each new qualifying burst adds a wave to the cascade record. The cascade remains active until a 500ms gap between waves triggers transition to ABSORPTION (if counter-volume present) or a 2-second gap triggers transition to IDLE (cascade exhausted).

ABSORPTION​

The cascade has paused and counter-directional activity is appearing. The system is computing the absorption score across its five components. This state has a 5-second timeout. Possible outcomes: transition to RECOVERY (absorption score rises, counter-volume threshold met), transition back to ACTIVE_CASCADE (new cascade wave fires — absorption failed), or transition to IDLE (timeout with no resolution).

RECOVERY​

Absorption is confirmed and price is recovering. The system tracks recovery ratio and logs the final cascade summary. This state has a 30-second timeout or terminates when price recovers 50%+ of the cascade range. Upon exit, the complete cascade event record is finalized and stored for session statistics.

9. Execution Terminology​

Latency​

The time delay between sending an order and it reaching the exchange's matching engine. At 400–600ms (Nasser's measured latency for MES), the order takes roughly half a second to travel from the trading platform to CME. This latency means competing with sub-millisecond algorithms for the first fraction of a recovery is not viable, but trading the sustained recovery (which unfolds over minutes) is entirely feasible.

Slippage​

The difference between the expected fill price and the actual fill price. During and immediately after a cascade, slippage is elevated because the spread is blown out and the book is thin. A market buy during a 10-tick spread on CL gives up $0.10 on entry. Waiting for spread normalization (an absorption score component) reduces slippage to near-normal levels.

Execution Quality​

How closely a trade's actual fill matches the theoretical ideal. For cascade reversion trading, execution quality is measured by: how much of the recovery range was captured (entry price relative to cascade extreme and pre-cascade level), slippage on entry, and slippage on exit. The recognition advantage from the Cascade Analyzer compensates for latency-related execution quality loss.

Recognition Advantage​

The edge derived from correctly identifying and classifying market events faster than other participants — not through lower latency, but through superior analytical processing. A human trader with 400ms latency but correct cascade identification, absorption confirmation, and recovery quality assessment will outperform a trader with 10ms latency who can't distinguish a cascade from a normal aggressive burst. This is the strategic foundation for the Cascade Analyzer: converting recognition speed into execution edge.

10. Statistical Validation Terminology​

Reversion Rate​

The percentage of cascade events where price returns to the pre-cascade level within a specified time window. Measured across multiple time horizons (30 seconds, 1 minute, 2 minutes, 5 minutes, 10 minutes). A reversion rate significantly above 50% suggests a tradeable edge; at or below 50% suggests no edge. This is the primary metric for validating the sweep gap reversion hypothesis.

Expectancy​

The average profit or loss per trade, accounting for win rate, average win size, average loss size, and costs (commissions + slippage). Calculated as: (winRate * avgWin) - (lossRate * avgLoss) - costs. Positive expectancy means the strategy makes money over a large sample. The sweep reversion hypothesis must demonstrate positive expectancy after realistic execution costs (2-tick slippage assumption + round-trip commissions) to be viable.

Confirmation Bias​

The cognitive tendency to remember instances that confirm a pre-existing belief and forget instances that contradict it. In the context of sweep reversion: after two years of observation, a trader may remember the dramatic reversions vividly and forget the cascades that never reverted. Statistical testing against tick data is the antidote — the data includes every event regardless of how memorable it was.

Sample Size Sufficiency​

The minimum number of observed events needed to draw statistically reliable conclusions. For cascade events (which are relatively rare — perhaps 2–5 per week on CL or ES), achieving sufficient sample size requires months of data. The 3-month tick data extraction in progress is the minimum viable dataset; 6–12 months would provide higher confidence.

See Also​