Order Flow and Local Trading Responses

Order-flow analysis expands trade and quote data into finer time and price detail to observe what actually occurs near key structural locations. Common tools include tick trades, time and sales, footprint charts (Footprint or Numbers Bars), classified aggressive buying/selling volume, Delta, and imbalance markers. These are extensions enabled by modern data and software, not Richard D. Wyckoff's original terminology. They best address local execution: once higher-timeframe context, trading-timeframe structure, and candidate zones are established, does the entry timeframe show a response consistent with the hypothesis? Order flow cannot independently identify accumulation or distribution, and colored numbers cannot replace structure, price-volume analysis, or a risk plan.

I. Distinguish Orders, Quotes, and Trades

The limit order book shows currently visible resting buy and sell orders. The best bid and quantity are buying quotes; the best ask and quantity are selling quotes; deeper levels show visible limit volume farther from the market. Resting orders are public instructions to trade under specified conditions and may be changed or canceled before execution. Book quantity is therefore neither executed volume nor supply/demand guaranteed to remain. The same aggressive buy quantity may fill at one level against thick offers but lift successive trade prices after sellers withdraw into a thin book. Replenishing offers may absorb later aggressive buying, producing high volume with little displacement. Cancellation and replenishment change what subsequent aggressive orders can consume, so impact depends on volume, contemporaneous depth, and update sequence. Rapidly added and withdrawn orders may reflect changing intent, market-making updates, algorithmic management, or deception; the display alone cannot establish why.

Tick records describe already matched trades, normally with time, price, and quantity; some sources also provide exchange or vendor direction flags. Every trade has both buyer and seller. Aggressive buying does not mean only a buyer exists; it means the trade is classified as a buyer crossing the spread to execute against a seller's resting order. Aggressive selling means a seller crosses to execute against a buyer's order. Classification describes execution method, not identity or whether the aggressor opened a position. Aggressive buys can open longs, cover shorts, or execute arbitrage; aggressive sells can open shorts, close longs, or hedge.

Public books usually show displayed quantity only. Icebergs show part of their size and may replenish after fills. Visibility of fully hidden orders, off-exchange trades, internal matching, and other venues depends on rules and source. Repeated consumption and replenishment at one price while price struggles to cross can support absorption: persistent replenishment with no result for the aggressive side. It cannot distinguish one iceberg from successive participants or an appearance caused by aggregation/delay, nor identify the counterparty or opening/closing purpose. Footprints show execution results; books show visible quotes at observation time. Combining them narrows explanations but cannot claim discovery of an institution's position.

Coverage determines the visible market portion. Exchange tick data from centralized futures are generally more consistent than one broker's decentralized-market data, but check subscribed depth, historical backfill, spread or implied-trade treatment, corrections, and local packet loss. Stocks trade across venues; one platform may omit others. Spot foreign exchange often uses broker quotes or trades as a local sample, not whole-market order flow. Record source, exchange, sessions, time zone, and contract.

Unfilled limit orders remain in the book only if time in force and venue rules permit waiting; unfilled IOC, FOK, and similar instructions cancel immediately under their rules. Market and marketable limit orders execute against available opposing quotes. A fill establishes matching, not whether either side opened or closed. Visible quotes can cancel, hide, or change during observation delay and cannot equal certain executable supply/demand. Footprints derive from completed trades and books from unfilled quotes, so their time states differ. With local data, first verify exchange timestamps, aggregation, and full trade coverage.

Orders, Quotes, Trades, and Aggressor Classification

The book represents unfilled quotes and the footprint completed trades; their time states differ.

Figure A2-1 Orders, Quotes, Trades, and Aggressor Classification (from Advanced Appendix II)

II. How Aggressor Classification Is Produced

Reliable exchange direction flags can be used directly. Otherwise a common method matches each trade price with contemporaneous best quotes. Trades at or above the ask are usually classified as aggressive buys; trades at or below the bid as aggressive sells. Midspread trades or inaccurate alignment of quote and execution times may leave direction uncertain. Vendors may complete classification using prior direction, quote distance, or proprietary rules.

The tick rule is another approximation: a price above the preceding trade is buy-directed, below it sell-directed, and unchanged price inherits the last nonzero direction or follows platform rules. This is not quote matching. Timestamp precision, network ordering, exchange batch publication, and software delay affect classification in fast markets. Same-price, same-time trades may be aggregated, or one exchange event split into several records. Historical candle totals cannot reliably reconstruct actual tick aggressor direction.

Before using order-flow rules, establish whether direction comes from exchange flags, quote matching, or tick rule; quote timestamps; ordering of simultaneous events; whether consecutive trades are merged; whether history and live data use one algorithm; and handling of omissions and corrections. Old thresholds cannot automatically transfer to a new platform/vendor. Classified volume is an algorithmic output, not traders' true intent.

Aggressor classification commonly uses execution at Ask as buying and at Bid as selling. Midspread trades, asynchronous quote/trade updates, or missing quote snapshots may be unreliable and should remain unclassified U rather than forced to a side. Tick-rule inference from price changes need not agree with Bid/Ask classification. Classification only indicates which quote side was actively accepted under the chosen data rules; it does not prove net inflow of new money.

III. Bid×Ask in Footprint Charts

A footprint shows classified volume at each price bin inside a candle. For bin p, let Ask(p) be volume classified as aggressive buying at the offer side and Bid(p) aggressive selling at the bid side. Common displays put Bid left and Ask right, but layout can reverse and colors are not standardized. Verify field definitions first.

Same-price horizontal comparison examines both categories at one bin, for example Ask(p) versus Bid(p), or bin Delta: Delta(p)=Ask(p)-Bid(p).

This describes the distribution of the two aggressive categories within one trade-price bin. It is useful for composition and bin Delta, but cannot identify who won a literal contest between adjacent quotes: the columns may reflect different times and every trade still has a counterparty.

Diagonal comparison contrasts adjacent-bin trades crossing the spread. With minimum tick size tick, common buy-direction ratio R_buy(p)=Ask(p)/Bid(p-tick), and sell-direction ratio R_sell(p)=Bid(p)/Ask(p+tick).

The first compares aggressive buying at one bin with aggressive selling one bin lower; the second compares aggressive selling with aggressive buying one bin higher. This adjacent-diagonal convention highlights relative distribution of buying higher and selling lower. It is not both sides of the same match and does not require the actual spread always to equal one tick. Trades may occur at different times with intervening quote cancellations and replenishment. Numbers Bars-type construction needs only best Bid/Ask for classification, not historical Market Depth. Diagonal ratios therefore are neither simultaneous book depth nor remaining supply/demand, and imply no win rate. Footprints can use same-price, diagonal, total volume, bin Delta, and intrabar distribution views, each answering different questions.

Ratio algorithms must handle zero denominators. Under this course's mathematical definition, a ratio with zero denominator is undefined, including 0/0. At boundaries, missing counterparts, or zero volume, software may display 0, infinity, substitute one, use an absolute difference, or omit a marker. These are implementation conventions, not changes to the formula or evidence of extreme trading strength. Also require a predefined minimum numerator volume. After bin aggregation, replace tick in the formula with one displayed bin width so comparison matches the chart.

Imbalance usually means a chosen ratio or volume difference exceeds user parameters, not an intrinsic market event. Thresholds, minimum-volume filters, required consecutive bins, zero handling, aggregation, and inclusion of unfinished bars all change markers. Typical trading size and noise differ greatly by market; no percentage, minimum quantity, or consecutive-bin count is universal. Calibrate to the instrument, session, and source's historical distribution and keep live and historical rules identical.

Each Bid×Ask cell normally puts aggressive selling on Bid and aggressive buying on Ask, but verify left/right order. Horizontal ratios compare categories at the same price; diagonal buying commonly compares Ask with Bid in the next lower bin, with the selling direction reversed. Specify numerator and denominator. A zero denominator lacks ordinary division meaning and must use the predefined display rule, not automatically be called infinite imbalance. Consecutive imbalances show same-direction trading across several bins, but are not guaranteed more reliable than one bin: one large order may sweep them all, so they are not independent confirmations.

IV. The Precise Meaning of Delta and Cumulative Delta

Delta for one bar or sample is classified aggressive buying minus classified aggressive selling: Delta=Σ_p Ask(p)-Σ_p Bid(p). If unclassified volume U exists in the same scope, total volume V=Σ_p Ask(p)+Σ_p Bid(p)+U; Delta uses only classified volume. Sources with different unclassified proportions should not have their Delta directly compared even at equal total volume.

Positive Delta means more buy-directed volume under the classification rule; negative means more sell-directed volume. It is not net money inflow: each trade has equal buying and selling amounts, with different aggressor labels. Nor is it total long holdings minus shorts; every open derivative contract has both a long and a short. Delta is not institutional trading volume, since public data usually lack identity fields. Equal Delta can yield different price results: deep replenishing opposition may block heavy aggressive flow, while thin or withdrawn liquidity lets less flow cross many levels. The same total difference concentrated late versus spread evenly throughout also leaves different subsequent conditions. Delta preserves the net classified difference, not full depth, replenishment, cancellation, or sequence.

Cumulative Delta (CD) adds Delta across successive samples. If time unit t has Delta_t, then CD_t=CD_(t-1)+Delta_t.

If two curves include identical increments throughout the same later segment and differ only in accumulation start, they differ by a constant offset. This changes neither each increment, the relative values at any two times, nor the order of local extremes, and cannot manufacture divergence. Session or calendar-day resets interrupt comparisons across boundaries. Changed session coverage, added/omitted trades, contract changes, and classification changes alter continuous structure. Specify reset rules, sessions, overnight inclusion, and historical gaps. Missing ticks leave persistent cumulative error afterward; repairing only the newest bar is insufficient.

Open Interest (OI) counts outstanding contracts, published at exchange-specified intervals. Volume counts contracts traded during a period; Delta classifies those trades by aggressor direction. Their objects differ. Rising OI does not mean more longs than shorts, because each new contract creates both sides. Price, volume, and OI together can suggest changing participation but still cannot identify either side's participants.

Bin Delta equals classified aggressive buying minus classified aggressive selling; U enters neither side. It is not total volume, net money inflow, or position change. CD accumulates Delta from a specified start. If both starts precede the comparison window, the difference normally is constant and does not change within-window slope or relative new-high comparisons. Resets inside the window, changed sessions, contract rolls, or missing trades can change the comparison shape. Compare additional aggressive trading with actual price movement. Large positive Delta without upward progress may mean passive offers absorb aggressive buys; a positive value guarantees no advance.

Figure A2-3 Delta and Cumulative Delta (schematic)

V. Two Main Explanations of Price–Delta Divergence

Specify whether divergence uses single-bar or cumulative Delta and align price and Delta windows and swing endpoints. A price new high without a CD new high may reflect weaker aggressive buying or thin overhead supply letting less volume move price. A CD new high without a price high may reflect aggressive buying absorbed by passive selling. Downside observations can be symmetrical, but actual swings must validate the mechanism. Single-bar divergence with one candle's extremes is more local and cannot be confused with cross-session cumulative divergence. All divergence first describes relative behavior of two series and may also arise from classification errors, omitted venues, or changed sample/session coverage.

Passive absorption describes repeated heavy aggressive trading with limited net progress beyond a location. Negative Delta at key lower support without further decline supports only that aggressive selling met enough buyer response and selling effort temporarily failed to produce proportionate results. It does not exclude a renewed break after selling pauses, classification error, or withdrawal of lower bids. A bullish view still needs upward departure, continuing buy-directed flow, or a retest where sellers again fail. Analyze positive Delta without progress at key upper resistance inversely. The observable fact is more aggressive volume with limited movement, not proof that a particular institution is building a position.

Aggressive exhaustion means reduced activity from the driving side as price probes further without prior participation. Absorption means aggressive trading remains heavy but fails to progress; exhaustion means that trading itself weakens. Reduced driving force does not automatically create opposite aggressive flow or make opposite quotes move price. With less buy-directed trading at an upward thrust's end, price may still rise through thin offers or remain stationary awaiting orders. Subsequent opposing net displacement, a crossed key boundary, or failed retest adds reversal support. Continued holding high and further advance weaken buyer-exhaustion interpretation.

A one-bar divergence, extreme Delta, or large number therefore only initiates observation. For low-level absorption, selling first fails to extend the decline, then aggressive buying at higher prices achieves an upward result. Those sequential facts support directional change. The chart cannot establish whether the same participants first posted absorbing bids and then chased; reverse the logic at tops. The steps depend on one another, so correlated-signal ratios or subjective strengths cannot be multiplied into a win rate. Without departure, absorption may persist or eventually fail. If the old pressure later crosses and holds beyond the absorption area, absorption no longer stopped progress; that does not erase the historical fact that prior aggressive trades were absorbed. Consecutive imbalances likewise guarantee no continuation: concentrated aggressive flow may move price or meet ample passive liquidity.

Price making a new high without CD may mean less aggressive buying crossed thin offers, or reflect classification, resets, or missing data. Rising positive Delta without price progress may mean passive sellers absorbed buyers. Reverse for declines. Check both coordination of classified flow with price results and differences caused by data construction rather than automatically choose one explanation. Then observe local-boundary crossings, persistent imbalances, and shifts in dense trading to identify the interpretation fitting the action.

VI. Using Local Trading Responses in Wyckoff Events

Narrow from structure to local detail. First assess the main trend, range, or possible phase on the higher timeframe; mark trading-timeframe boundaries and Spring, Test, LPS, UTAD, LPSY candidates; finally open the footprint only in predefined zones to assess whether local response supports a trigger. If structure is ambiguous, magnifying to ticks usually only adds selectable noise.

For a possible Spring, crossing below an accumulation range and recovering is the primary price fact. A footprint can show whether sell-directed activity expands on the break without further progress, buy-directed flow follows on recovery, and a retest after return shows less selling. Even then, subsequent demand progress is needed for a long trigger. Continued trading below, Delta and price expanding downward together, and outside acceptance damage the Spring interpretation. Execute protection at the predefined trade boundary rather than wait for the structure's final renaming.

A Test checks the prior hypothesis of reduced supply or demand. In accumulation, compare sell-classified volume, total volume, and downward result with the preceding visit as a pullback enters candidate buying support, and observe whether upward movement resumes quickly. Low Delta does not define a test; significant sell-directed flow may be absorbed. The key remains absent downward result followed by demand. Easy penetration through the candidate zone with continued decline means the test failed or context needs reassessment.

An LPS is a pullback after an SOS. Order flow helps distinguish continued supply control from sellers progressively failing: examine depth, results of sell-directed flow, whether local lows hold, and effective displacement when upward movement resumes. Link the trigger to LPS structure and place protection where this participation rationale invalidates. One bright buy imbalance in the range center lacks LPS location significance and cannot carry equal weight merely because its color stands out.

Assess distribution from actual downside behavior rather than mechanically invert every long condition. UTAD first requires crossing the ceiling, failing to hold, and returning inside. Local trades can show whether rising aggressive buying still yields limited advance, selling follows on the reaction, and a renewed upper test is blocked. LPSY occurs in a rally after SOW; focus on whether buyers can continue raising price and sellers can then depress it again. Abandon the short view if trading persists above the ceiling without the reaction entering the range, or a rally crosses and holds beyond predefined invalidation.

Execution requires separate trigger and protection conditions. A trigger describes new evidence needed for entry, such as progress after reclaiming a key level or renewed departure after a holding retest. Protection specifies where post-fill risk must be cut. A footprint signal may temporarily disappear or reverse; it cannot justify waiting for fuller structural confirmation after protection triggers. Conversely, valid larger structure does not mean every local entry should remain open: slippage, a false trigger, or an overly close local boundary may invalidate the entry first.

If live data disconnect, classification proportions change abruptly, historical and live algorithms differ, or the platform cannot explain its imbalance algorithm, cancel untriggered order-flow entry conditions and use a predefined price-structure trigger or decline the opportunity. For established positions, data failure does not cancel existing stops. Reduce or exit under the contingency plan's price and risk conditions; inability to see order flow does not remove risk. Before reuse, fill missing segments and recalculate. If impossible, start a new sample from a clearly specified new session.