Learn to read order flow, then prove you can trade it.

Every quote has two prices: the bid, the highest price a buyer will pay, and the ask, the lowest price a seller will accept. The gap is the spread.

Buying at the ask fills you now and costs you the spread. Resting your own limit inside the spread gets a better price but may never fill. That trade-off sits underneath almost every scalping decision you will make, and Lesson 11 explains why the spread exists at all.

Time and sales prints every trade as it happens. A green print means the buyer was impatient enough to lift the ask. A red print means the seller was impatient enough to hit the bid. Colour tells you who paid up — which is a different question from which way price moved.

One print means nothing. A run of prints on one side, especially with size, tells you who is in control right now. Beginners watch the last price; tape readers watch which side keeps paying the spread.

Level 2 shows resting size at each price. Comparing the two sides shows imbalance — which side has more firepower parked and waiting.

01 — Be precise about what you actually know

A genuine breakout is the thin side of the book being eaten. Ask size depletes print by print until there is nothing left, and price has nowhere to go but up. Downside is the mirror.

The trigger is the depletion, not the candle that closes above the level afterwards. By the time the breakout is obvious on a chart, the cheap entry and the tight stop have both gone.

Sometimes a wall is hit again and again but keeps refilling instead of depleting. That is absorption: a large participant taking the other side of aggressive flow without letting price move. It is the single most useful tell on the ladder.

Absorption at resistance often precedes a reversal once the big seller finishes, or a sharp breakout once they are done and step away. The tradeable moment is usually when the refill finally fails to appear — not while it is still refilling.

01 — The most important distinction in this lesson

Spoofing is placing large orders with no intent to execute, to shape what everyone else believes. On a still frame a real wall and a fake one look identical. The difference is only visible over time.

Never trade displayed size alone. Wait to see whether it gets tested and trades, or vanishes the moment price arrives. The same discipline applies to every pattern in this course: the shape proposes, the tape confirms.

The same tape pattern means different things in different conditions. Filter every setup:

• Time of day — roughly 9:35 to 11:00 ET carries the most directional follow-through; midday is where good patterns go to die.
• VWAP — a crude but effective bias filter. Longs above, shorts below, all else equal.
• Spread and liquidity — wide, jumpy spreads mean your stop is less reliable than it looks on the chart.
• News — fresh headlines produce flow that has nothing to do with the structure you were reading. Let it digest.
• Relative volume — a stock trading at its normal volume is not in play, and patterns on it resolve at random.

Rules used throughout this site, for a $25,000 account with roughly 4x intraday buying power and no PDT restriction:

• Risk $25 to $50 per trade. No exceptions, regardless of how good it looks.
• Stop for the day at three losing R. The trading desk in Tape Lab enforces this.
• Around four trades a day. Scalping rewards selection, not activity.
• Progression: replay historical data bar by bar, then simulator, then 100 shares live, then size up only once your process scores are consistent.

Everything so far describes who is trading. Auction theory describes where they are willing to trade, which is what tells you whether a break should be chased or faded.

The market as a two-sided auction

Price rises until buyers stop showing up, then falls until sellers dry up. The price where the most volume changed hands is the market's own statement about what is fair, for that session, by the only measure that counts: where business actually got done.

Point of control and value area

• Point of control — the single price with the most volume. The centre of gravity.
• Value area — the band holding roughly 70% of the session's volume, grown outward from the point of control.

Initiative versus responsive activity

• Initiative — price pushes away from value and keeps going. Someone needs to be filled badly enough to pay ever-worse prices.
• Responsive — price pokes outside value and gets dragged back in. The market disagreed that the new price was fair.

Your first half hour has one job: decide which of those two days you are in. Everything downstream — entry style, stop width, target — follows from that answer.

Classical tape reading reduces to one habit: compare the effort spent with the result achieved. Volume is effort. Price change is result. When they disagree, someone is doing something.

The two readings

• High effort, low result — heavy volume, price stuck. Absorption. Lesson 5's ladder pattern, viewed through volume instead.
• Low effort, high result — thin volume, price travels. Nothing is in the way, and the move can continue cheaply.

The spring

A break below support that fails quickly and reverses. Shorts are filled at the worst price, then squeezed. Modern vocabularies call it a stop hunt or a liquidity sweep.

The upthrust

Deciding whether to believe it

01 — Be precise about what a sweep actually confirms

This lesson explains why markets behave the way the previous two lessons describe. None of it requires a conspiracy; it falls out of incentives.

Who is on the other side

• Dealers and market makers — quote both sides, earn the spread, manage inventory risk rather than predict direction.
• Informed traders — have a genuine short-term edge and want to be filled before it decays.
• Liquidity traders — trade for reasons unrelated to short-term value: hedging, rebalancing, ordinary retail flow.

Adverse selection, and why spreads exist

A dealer cannot tell in the moment whether the next order is uninformed, which is profit, or informed, which is a loss about to happen. The spread is the fee charged for that uncertainty. When spreads widen, dealers are telling you they think informed flow is around.

Why obvious levels get swept

Everyone is looking at the same chart, so stops and limits pile up just beyond the same swing highs and lows. That concentration is the cheapest available liquidity for anyone who needs to move real size. A large buyer does not sweep the lows to hurt you; they sweep the lows because that is where the shares are.

Your order type is a trade-off, not a preference

That last cell is adverse selection aimed at you. Your resting bid fills reliably in exactly one situation: when someone informed is selling into it.

Everything above is worthless without the discipline to execute it the same way on the twentieth trade as on the first. Professional desks solve this by grading the process, never the profit.

What counts as a good trade

A good trade is one you would take again given the same information. A loss taken at the planned stop, in the planned size, on a planned setup, is a good trade. A win taken in triple size with no plan is a bad trade that happened to pay — and it teaches you the wrong lesson at exactly the moment you are most receptive.

Grade yourself on five questions, every trade

1. Did you know your levels before the trade existed?
2. Did you know the stop and the target before you clicked?
3. Did size come from the stop distance, or from how sure you felt?
4. Did you exit where you said you would?
5. Did you wait for your setup, or manufacture one?

Traders fail in a small number of repeated ways: trading an opinion instead of the tape, preferring to be right over being paid, expecting a win rate no strategy delivers, and loosening risk control at the exact moment being wrong should tighten it.

The advanced modules taught frameworks that sound unrelated and describe the same handful of behaviours from different decades and different seats.

Reading order

1. Harris, Trading and Exchanges — chapters 1 to 4 give you the vocabulary for why spreads and impact exist.
2. Wyckoff, Studies in Tape Reading — public domain, linked in the library.
3. Weis, Trades About to Happen — the most directly applicable of the set.
4. Dalton, Mind Over Markets — lands far better after some screen time.
5. Bellafiore, One Good Trade — read whenever discipline slips.

Everything in Lessons 1 through 11 was qualitative: size, aggression, absorption, described in words. Market microstructure researchers measure the same phenomena as numbers, and the relationship turns out to be almost embarrassingly simple.

Order flow imbalance (OFI)

Define OFI over a short window as the net change in displayed depth, weighted by which side it happened on:

In plain terms: count new size arriving at or above the previous best bid as positive, new size arriving at or below the previous best ask as negative, and net the two. A print that lifts the offer and depletes it contributes differently from one that just adds to the bid — OFI is built to capture that from the book itself, before you even look at the tape.

The empirical result that matters

Cont, Kukanov and Stoikov (2014) regressed short-horizon price change against OFI across liquid futures and equities and found something close to a straight line:

λ (lambda) is a single coefficient — price impact per unit of order flow imbalance — and it holds up well within a stock or contract over a session, though it shifts across names and across volatility regimes. This is Kyle's lambda (1985) rediscovered empirically: a microstructure model built to describe how a market maker prices the risk of trading against someone informed collapses, in practice, to a linear relationship you can estimate from an afternoon of data.

01 — Why this explains the whole first half of this course

You cannot tell whether you have an edge by looking at your win rate. You need to know how much that win rate could plausibly be wrong by — and with the sample sizes most traders actually have, the honest answer is usually "a great deal."

Expectancy, restated properly

Per-trade expectancy in R is the mean of your realised R-multiples:

That number alone tells you almost nothing about whether it is real. You also need its variance:

where σ(R) is the sample standard deviation of your R-multiples and n is the trade count. A 95% confidence interval on your true expectancy is approximately E[R] ± 1.96 · SE. If that interval contains zero, your data cannot yet distinguish your process from a coin flip — regardless of how good the story behind each trade felt.

Why n = 20 tells you almost nothing

R-multiples on a discretionary strategy are noisy — a typical σ(R) sits around 1.0 to 1.5. At n = 20 with σ(R) = 1.2, your standard error is about 0.27R, so a 95% interval is roughly ±0.53R around whatever mean you measured. An observed edge of +0.4R per trade — which would be an excellent result — is statistically indistinguishable from zero at that sample size. You need somewhere in the range of n = 80 to 150 trades, depending on your actual variance, before a modest real edge reliably clears its own confidence interval.

01 — What this means for a discretionary reader, specifically

Lesson 8 fixed your risk at a flat $25 to $50 per trade regardless of conviction — a deliberately simple rule for a reason this lesson makes precise. Once you have an actual edge, there is a formula for the size that grows an account fastest. It is almost never the size a sane person should use.

The Kelly criterion

For a strategy with win probability W and a win/loss size ratio R (average win size divided by average loss size), the growth-optimal fraction of capital to risk per bet is:

Example: a 45% win rate with an average win of 2R against an average loss of 1R gives f* = 0.45 − 0.55/2 = 0.175 — risk 17.5% of account equity per trade at full Kelly. Reading that number should make you uncomfortable, which is the point of the next section.

Why full Kelly is not investable

• W and R are estimates, not truths. Lesson 15 just showed you how wide your confidence interval is at realistic sample sizes. Kelly's formula assumes you know W and R exactly; feed it an optimistic estimate and it happily recommends a fraction that ruins you.
• Full Kelly has brutal variance. Even with the true parameters, full Kelly portfolios commonly see 50%+ drawdowns on the way to their long-run growth rate. Nobody survives that psychologically, and most accounts have a real-world stop-out point Kelly's math does not know about.
• Returns are not identically distributed across regimes. Kelly assumes a stable repeated game. Lesson 7's time-of-day and Lesson 15's advanced-context material both say your actual W and R shift by session and by news proximity — the inputs to the formula are moving targets.

Fractional Kelly, the version people actually use

Most quantitative shops that size by Kelly at all use a quarter- to half-Kelly fraction:

Quarter-Kelly cuts long-run growth rate by roughly half relative to full Kelly, but cuts variance by close to three-quarters — a trade almost everyone should take. It is also a built-in hedge against the estimation error in the previous section: if your true edge is actually zero and your measured W and R were noise, quarter-Kelly loses you a quarter as fast as full Kelly would.

01 — Reconciling this with Lesson 8's flat $25

Everything so far reads volume as a single bar-height number. A footprint chart splits that number by side and by exact price tick, turning one candle into a small grid: buy volume and sell volume, printed against every price the bar traded through.

Delta and cumulative delta

Delta is the simplest possible order-flow number:

Ask-side volume is trades that lifted the offer (buyers paying up); bid-side volume is trades that hit the bid (sellers giving it away). A single bar's delta tells you which side was more willing to pay the spread during that bar. Cumulative Volume Delta (CVD) is the running total of every bar's delta across the session — a scoreboard of aggression that never resets, plotted directly beneath the candles.

The four named divergences, and where you've already seen every one

• Classic bearish divergence — price makes a higher high, CVD makes a lower high or stalls. This is your Upthrust (Lesson 10) expressed in delta terms.
• Classic bullish divergence — price makes a lower low, CVD holds or makes a higher low. This is your Spring (Lesson 10) expressed in delta terms.
• Hidden divergence (continuation) — in an established trend, price makes a shallow higher low but CVD makes a lower low: sellers threw volume at the market and still couldn't break it. This is the pullback-and-hold read behind Initiative trading (Lesson 9), one step more specific.
• Absorption-based divergence — CVD surges hard in one direction while price does not move at all. This is Absorption (Lesson 5), named exactly the same way in both vocabularies.

01 — CVD and the OFI from Lesson 14 are not the same measurement

Two remaining gaps: the specific order types that produce the patterns you already read, and an honest answer to which markets this entire course actually applies to.

Iceberg orders — the mechanism behind absorption

An iceberg (reserve) order shows only a small slice of a much larger order on the book. When that slice trades, it silently refreshes to the same displayed size, again and again, while the tape prints thousands of shares against one price. This is the literal mechanism behind absorption (Lesson 5) — absorption is what an iceberg looks like from outside; the iceberg is the order type that produces it.

Intermarket sweep orders — urgency made visible

An Intermarket Sweep Order (ISO) routes simultaneously across multiple exchanges to take resting liquidity at several price levels at once, bypassing the usual order-protection checks. On the tape it looks like a millisecond burst tearing through two or three price levels in one motion — a strong tell that a large participant needs to be filled now, not eventually.

Stacked imbalances — the footprint-level version of Lesson 3

Lesson 3 compared resting size on the book. A stacked (or diagonal) imbalance is the executed version at the tick level: three or more consecutive price ticks where one side's traded volume overwhelms the other by roughly 3:1 or more. A run of these is often read as short-term structural support or resistance, since it marks exactly where one side was recently overwhelming the other.

Who is actually trading against you

Where does any of this actually work?

01 — What years of screen time actually buys you