Trading Glossary

Every term the 100 lessons use, defined the way the lesson that introduces it defines it. Each entry names the lesson, so a definition you doubt can be checked against the arithmetic it came from.

A

Absorption

Aggressive orders arriving at a price without moving it, because resting size is taking the other side. It describes what the tape shows rather than who is behind it: one large passive seller and a queue of small ones print the same thing. Opposite of exhaustion. See: Lesson 28: Absorption and Exhaustion, Lesson 8: Volume and Delta

Access Fee

What a venue pays or charges for an order, capped by rule at three tenths of a cent a share. You never pay it, which is exactly why it decides where your order is sent rather than you deciding: lesson 54 finds the swing across routing outcomes worth $1.65 on a 300-share order, turning lesson 53’s $3.00 round trip into $4.80 with the extra 60 per cent invisible on your statement. See: Lesson 54: The Fee That Routes Your Order, Lesson 53: What the Spread Is Paying For

Adverse Selection

The market maker’s problem: the orders that hit a resting quote are more likely to come from someone who knows something. It is the reason the spread exists at all, and the reason it widens when the chance of an informed counterparty rises. See: Lesson 53: What the Spread Is Paying For, Lesson 1: What a Market Solves

Ask (Offer)

The lowest price a seller is willing to accept, and what you pay if you will not wait. It is the top line of a queue rather than the queue: on the ordinary book lesson 2 prints, the quote describes 900 shares out of 7,300. See: Lesson 2: The Order Book

ATR (Average True Range)

The average over a lookback of each bar’s true range, which is the greatest of the high less the low, the high less the previous close, and the previous close less the low. Lesson 21 uses it to say how much room ordinary movement needs rather than where a trade is wrong. The lookback is a dial like any other: lesson 44 gets a per-bar standard deviation of 0.32 per cent and of 3.43 per cent out of the same sixty bars, and the difference is the window rather than the market. See: Lesson 21: Where the Stop Goes, Lesson 44: Volatility as a Quantity

Attribution

Splitting a result into the part the market gave and the part the rule added. It is a subtraction, so the whole content sits in what you chose to subtract: lesson 66 finds lesson 63’s winning rule owing 57.7 per cent of its return to a fully invested benchmark and 30.0 per cent to one weighted by the exposure it actually carried, which nearly doubles what the rule is credited with, from 4.16 a share to 7.75. See: Lesson 66: The Benchmark You Chose, Lesson 63: Backtesting as Evidence

Availability

The share of the time a link answers, which multiplies along a chain, so the chain can never be more available than the link you do not own. Lesson 77 finds the largest single improvement is free: a stop resting at the broker takes your own equipment out of the product for an open position and moves it from 96.3 unprotected hours a year to 8.8. See: Lesson 77: The Link You Do Not Own, Lesson 76: The Pace You Actually Trade At

B

Backtest

Running a rule over past prices to see what it would have done. Lesson 63 treats a backtest as evidence with a strength that can be stated rather than as a result, and prices the four subtractions between a gross figure and one worth believing. See: Lesson 63: Backtesting as Evidence, Lesson 62: What Would Have to Happen

Bar

Not something the market produced but a rule for grouping the tape, and the rule has two settings: how wide each group is and where the first one starts. Only the first is ever quoted. Lesson 38 regroups sixty bars four at a time and the sample high and low do not move while the swing count falls from 15 to 2; move only the starting bar and lesson 36’s ratio runs from 0.104 to 0.249 on identical tape. See: Lesson 38: What a Timeframe Is, Lesson 6: The Candle Is a Summary

Base Effect

The part of an annual rate that is decided by the month leaving the window rather than the month arriving. Eleven of the twelve months inside the rate were published before the release, and on the series lesson 47 prints the rate falls 3.37 points across half a year with four fifths of that fall carried by the months dropping out. See: Lesson 47: The Macro Cycle, Lesson 46: Positioning Data

Base Rate

The share of candidates that are genuinely alive before any test is run. It is the input nobody can measure and the one that decides what a passing test is worth: at one candidate in ten, a disciplined test leaves 43.2 per cent of what it accepts real, and two ordinary habits take that to 6.0. See: Lesson 84: The Share That Is Real, Lesson 63: Backtesting as Evidence

Basis Point

One hundredth of one per cent, which is how a cost is quoted so that instruments at different prices can be compared. A penny of spread on a share near 103 is 0.97 basis points and lesson 63’s whole round trip on it is 11.94. Lesson 12 runs the same conversion across instruments and finds a round trip costing 0.2 basis points on an index ETF and 467 on a micro-cap. See: Lesson 83: The Same Thing in Two Places, Lesson 12: What Should You Actually Trade

Bid

The highest price a buyer is willing to pay, and what you receive if you will not wait. Like the ask it is one line of the book: lesson 2’s book rests 7,300 shares behind a quote that describes 900 of them. See: Lesson 2: The Order Book

Bid-Ask Spread

The difference between the highest bid and the lowest ask, which is the price of not waiting. It is not a fee and it does not move with mood: lesson 5 shows it is a fixed cost plus two risks, and an unchanged company goes from a penny wide on 2,000 shares to four cents wide on 300 in a single afternoon. What it costs you is a share of your stop rather than a number of cents, which is lesson 4’s arithmetic. See: Lesson 4: The Spread Is the Price of Immediacy, Lesson 5: Why Anyone Quotes At All

Binding Constraint

The first row of a business card that fails, read in the order the rows depend on each other rather than in the order they were measured. Lesson 100 prints five rows (pace, cost, net earning, withdrawal and verdict) and finds the second binding: the cost takes 79.7 per cent of the edge before any account arithmetic starts, so repairing the withdrawal or the capital repairs rows downstream of a number that has already gone. See: Lesson 100: The Business on One Page, Lesson 75: Which Limit Binds First

Borrow Cost

What it costs to borrow a security in order to sell it short, plus the risk that the lender recalls it. Lesson 82 concedes it is the one price the course never measured, and it is the whole bill for the short leg of a spread. See: Lesson 82: The Second Way to Disagree

BOS (Break of Structure)

Price breaking a prior swing high in an uptrend, or a swing low in a downtrend. It is computed from a swing definition, and that definition is a setting nobody quotes: on the twenty bars lesson 32 prints, the same prices give eight structure events, or two, or none. See: Lesson 32: Market Structure

Breakout

When price moves through a defined level (resistance, range high, consolidation). It either continues or reverses, and nothing visible at the moment of the break tells you which. See: Lesson 35: Sweeps, Beyond the First, Lesson 27: The Liquidity Lie

C

Call Auction

A crossing that collects orders instead of matching them as they arrive, then solves for the single price at which the most shares can trade. Every filled order receives that one price: on the book lesson 42 works by hand it clears at 100.50, two ticks above everything that traded in the session, and charges it on all 2,600 shares where walking the identical orders continuously would have averaged 100.2654. See: Lesson 42: Opening and Closing Auctions, Lesson 2: The Order Book

Candle

Five numbers kept from an interval and everything else discarded: the first price, the highest, the lowest, the last, and the total size. Nothing in it is estimated, and what goes is the order the trades arrived in. Lesson 6 prints two five-minute intervals producing the same five numbers pixel for pixel, where the same bracket makes 0.35 in one and loses 0.15 in the other. See: Lesson 6: The Candle Is a Summary, Lesson 38: What a Timeframe Is

Capacity

The size at which a rule’s own impact eats its edge. Lesson 68 turns that into a number for one rule and finds two thirds of the edge spent to collect the other third at the largest position it can hold. See: Lesson 68: How Much the Trade Holds, Lesson 59: What Hurry Costs

CHoCH (Change of Character)

Price breaking structure against the prevailing direction, which is how a change of trend is declared rather than observed. It shares the swing definition BOS is computed from, and lesson 32 finds two ordinary settings disagreeing about who is in control on four bars in twelve. See: Lesson 32: Market Structure

Configuration

One choice of indicator and lookback, of which any screen holds several. Three indicators at four lookbacks is twelve, and they are not twelve opinions: on the 39 bars where all twelve are defined lesson 52 finds one reading bullish while another reads bearish on 20 of them, and their vote series correlating at 0.5455, which is 1.71 independent opinions. See: Lesson 52: Confirmation Bias, Measured, Lesson 48: What an Indicator Is

Contract

What you are actually holding, as a legal object rather than a line on a chart. A share, a futures contract, an option and a perpetual are four different ones, and which you hold fixes the smallest position you may take, whether it expires, and the most it can take from you: lesson 13 finds the smallest position long the same index running from $500 to $250,000 across the four. See: Lesson 13: What You Are Actually Buying, Lesson 14: Margin and Leverage

Correlation

How alike two series move, between −1 and 1. It is the single input that decides how many independent bets a set of positions amounts to, and lesson 74 shows it moves enough between two halves of one record to change the answer. See: Lesson 45: Correlation, Lesson 74: The Window Decides

Cost Base

A cost that arrives whether you traded or not, which is why it is a share of the edge rather than a bill the profits settle. Because lesson 79 sized the capital so the expected annual gain equals the wage, a cost base a quarter the size of the wage takes exactly a quarter of the edge: lesson 80 finds the verdict moving from 589 trades to 1,047, and from 10.07 years to 17.90. See: Lesson 80: The Cost That Does Not Scale, Lesson 79: Paying Yourself Before You Know

Cross-Correlation

The correlation between two different instruments, as against between two rules on the same one. Lesson 82 runs lesson 71’s divisor on it: four positions across two instruments carry 2.08 bets at a cross-correlation of zero and 1.28 at 0.6, against the 1.04 the two rules on one instrument were already carrying. See: Lesson 82: The Second Way to Disagree

Cumulative Delta

The running total of delta across a session. It inherits delta’s inference bar by bar, so the line is a sum of estimates rather than a sum of facts: lesson 8 shows a bar reported at plus 15 per cent establishes a buy share only between 50.0 and 62.5 per cent. See: Lesson 8: Volume and Delta

D

Daily-Loss Limit

A cap on what one day may cost, fixed before the session. Lesson 75 reads it beside the correlation and weight columns and finds it the one that binds: at three per cent a day and two per cent a trade the four rules module 9 licensed may carry 1.5 positions, which is to say one, at a count no argument about correlation ever reaches. See: Lesson 75: Which Limit Binds First, Lesson 72: The Day Every Stop Hits

Dark Pool

A venue that displays no orders, so its trades reach the tape without a venue you could have acted on. The print carries a price, a size and an off-exchange flag, and never a side. Lesson 56 tests the rule that reconstructs the side from the session average and finds it reconstructs the session instead, tracking the day’s return at a correlation of 0.93. See: Lesson 56: The Side the Tape Leaves Out

Delta

Volume that crossed the spread to buy, less volume that crossed to sell. Every trade has both a buyer and a seller, so delta counts who was unwilling to wait rather than who was more numerous. And the side is inferred rather than reported. Lesson 8 shows a bar reported at plus 15 per cent bounds the buy share only between 50.0 and 62.5 per cent. See: Lesson 8: Volume and Delta, Lesson 29: Volume at Price

Disposition Effect

Closing winners early and holding losers, named by Shefrin and Statman in 1985 and the subject of module 12. It is one habit rather than two, and lesson 86 prices its first half on lesson 63’s seven trades: taking the profit at the first close that shows one leaves the win rate at six of seven, exactly where it was, and removes 2.20 a share of the 4.16 the rule was worth over holding. See: Lesson 86: Taking the Profit, Lesson 89: The Money That Was Showing

Divergence

When price makes a new high/low but an indicator (RSI, momentum, volume) does not. The indicator is a function of the same prices, so a divergence is an arithmetic consequence of the path rather than an independent witness to it. See: Lesson 34: Divergence

Drawdown

The distance from a peak in the equity curve to the lowest point after it, quoted in R or in per cent. Lesson 67 puts the depth you should expect on an edge you believe in, and finds an eight-R line switches off 59.5 per cent of the systems that genuinely work. See: Lesson 24: When the Drawdown Arrives, Lesson 67: The Drawdown You Should Expect

E

Edge

The average result of one trade, in R, after costs. It is a quantity to be measured rather than felt: lesson 18 shows what an edge of a tenth of an R actually looks like from inside, which is mostly like nothing. See: Lesson 18: What an Edge Feels Like, Lesson 17: Expectancy

Equal-Weight Book

A portfolio that puts the same money in every rule it runs, with no optimising of any kind. Lesson 91 holds all 253 of lesson 63’s rules this way over twenty-eight moves and gets 4.16 a share, which is exactly the average of the rules inside it, for a standard deviation 4.21 per cent below theirs. The median return is 4.10 at one rule and 4.16 at 253, so the size of the book moves the range of outcomes and not their centre. See: Lesson 91: What the Whole Grid Earns, Lesson 95: The Book on One Card

Evidence Bar

The count of trades a claim about a rule requires, of which this course names several and the book satisfies none. Lesson 81 sets them against the seven trades lesson 63’s winner actually took: lesson 65’s fixed horizon wants 156, which is 22.3 times the record, lesson 67’s verdict wants 589 at 84.1 times, and lesson 80’s cost base wants 1,047 at 149.6 times. See: Lesson 81: What the Book Clears, Lesson 65: The Horizon You Fix First

Excursion

How far a trade goes against the entry before it finishes, measured here as the worst close inside it and quoted in R. It is the column that decides whether a stop is ever touched: on lesson 63’s seven trades the deepest is 0.453 of an R, so any stop wider than that never fires, and four of the seven never close below their own entry at all. See: Lesson 87: The Second Exit, Lesson 89: The Money That Was Showing

Exhaustion

Aggressive orders moving price because nothing is resting to absorb them. Like absorption it classifies what the tape shows rather than reading intent, and the classification is not forced by the data: lesson 28 prints two bars sharing every quantity a candle records, and the volume at every individual price, whose deltas are plus 2,820 and minus 1,680. See: Lesson 28: Absorption and Exhaustion

Exit Policy

The full set of rules that can close a position, taken together rather than one at a time. Two of them do not stack, they compete, and the earlier one is the only one that ever fires: lesson 90 finds an early exit costing 2.20 a share alone and a quarter-R stop costing 5.90 alone, which would leave 1.74 between them, while running both at once leaves 4.44. See: Lesson 90: The Four Habits Together, Lesson 87: The Second Exit

Expectancy

Win rate times average win, less loss rate times average loss. It is the one number that decides whether a method makes money, and every cost in this course is subtracted from it. See: Lesson 17: Expectancy, Lesson 10: Every Trade Starts Negative

F

Fill

The transaction your order actually got, as against the price you were looking at. It has a counterparty who chose that moment to trade against you, which makes the fills you receive a selected sample rather than a random one: lesson 3 prices the same one-in-ten miss at $1.50 for one trader and $15 for another against the same $2 saving. See: Lesson 3: What a Fill Actually Is, Lesson 58: What an Order Type Gives Away

Footprint Chart

A chart that splits each bar’s volume by price and by inferred side. The split is a classification with a known error rate rather than a record, which is why the same footprint supports two opposite stories in lesson 28. Lesson 29 separates the two halves: the count of volume at a price needs no inference at all, and the number everyone quotes from it does. See: Lesson 29: Volume at Price, Lesson 28: Absorption and Exhaustion

G

Gamma

The rate at which an option hedge changes with the price, which is why a dealer’s trading follows price rather than leading it. Lesson 60 prices 10,000 contracts struck at 103 on this course’s sixty closes: holding a position that never exceeds a million shares takes 3,079,024 shares of trading, and a dollar of movement at the strike moves the hedge by 57,861 shares with twenty bars left and by 258,898 with one. See: Lesson 60: The Flow With No Opinion, Lesson 44: Volatility as a Quantity

Gap

A stretch of price at which nothing traded, which is not a large move but an absence of one. Lesson 40 removes the overlap from the same sixty closes and 19 of 39 transitions become gaps while the high, the low and the efficiency ratio do not move. And a stop resting inside one is filled at a median of 1.5 times the loss it promised, and 4.5 times at worst. See: Lesson 40: Multi-Day Structure, Lesson 21: Where the Stop Goes

Golden Cross

A shorter moving average crossing above a longer one. Its lateness is arithmetic rather than opinion: on lesson 50’s sixty closes the ten crossed the twenty at bar 36, which is 12 bars after the low, with 54 per cent of the whole advance already behind it. See: Lesson 50: Moving Averages as Support, Lesson 48: What an Indicator Is

H

Horizon

The number of trades you fix, before the first one, as the point at which you will judge the record. Lesson 65 shows that judging continuously instead takes a dead system’s pass rate from 5.05 per cent to 24.25. See: Lesson 65: The Horizon You Fix First, Lesson 19: How Long Until You Know

I

Iceberg (Reserve Order)

An order displaying a fraction of its true size, so the screen understates rather than overstates. Detecting one is an observation rather than a read, and the catch is on the other side: across the hundred tests lesson 31 prints, the detector could not have fired at all on eighty, which bounds the rate between 14% and 94% and is why a quiet book proves nothing. See: Lesson 31: Hidden Size, Lesson 26: The Order Book Is Theater

Impact

The price movement your own order causes. It grows with size faster than the spread does, roughly as size to the power of one and a half, which is why a rule that works small can stop working large. See: Lesson 59: What Hurry Costs, Lesson 57: What a Million Shares Takes

Implied Move

The size of move an option chain is charging for before a dated announcement, read out of the straddle. It is a price rather than a forecast, and the hurdle it sets is not the one usually quoted: a stock at 100 with a 6.00 straddle is quoting six per cent, and lesson 43 finds the long call inside it breaking even at 3.10 per cent, roughly half. See: Lesson 43: Scheduled Events, Lesson 44: Volatility as a Quantity

Independent Bets

What a set of correlated positions actually amounts to: the count divided by one plus the count less one times the average correlation. At the 0.9472 lesson 71 measures, two rules are 1.03 bets rather than two, four are 1.04, and however many you add the total tops out at 1.06. See: Lesson 71: How Many Bets You Are Carrying, Lesson 45: Correlation

Indicator

A function of prices you already have, from which four things follow. It cannot add information, because the line was computed from the prices. It discards, and the discarding can be counted: two ten-bar windows of lesson 48’s series have averages 0.09 apart and standard deviations 0.076 apart while one drifts nowhere and the other trends, at efficiency ratios of 0.053 and 0.536. And it is late by an amount you compute rather than guess, which for a ten-period simple average is 4.50 bars. See: Lesson 48: What an Indicator Is, Lesson 52: Confirmation Bias, Measured

Institutional Flow

Orders large enough that working them is itself a decision. Lesson 57 prices that decision as a trade-off between impact and exposure rather than a pattern to spot, and lesson 56 shows the prints such orders leave carry no side at all, so nothing on a chart identifies them after the fact. See: Lesson 57: What a Million Shares Takes, Lesson 56: The Side the Tape Leaves Out

Inventory Risk

The risk a quoter carries in the position nobody wanted to take off them. It is one of the two risks lesson 5 finds inside a spread, and it is why the quote moves without any news: an unchanged company goes from a penny wide on 2,000 shares to four cents wide on 300 in a single afternoon. See: Lesson 5: Why Anyone Quotes At All, Lesson 53: What the Spread Is Paying For

K

Kill Switch (Circuit Breaker)

A rule that halts trading automatically when a stated threshold is crossed. This course prices the threshold rather than the mechanism: lesson 75 finds the daily-loss limit the only one of its four columns that can be enforced before an order is sent rather than argued about afterwards, and lesson 69 is where the three per cent it uses comes from. See: Lesson 75: Which Limit Binds First, Lesson 69: The Delay You Remove

L

Latency

The delay between a decision and its arrival at the market. Lesson 69 prices five minutes between signal and order at 32.10 basis points on one instrument, which is more than a second venue gives back across a whole record. See: Lesson 55: What a Millisecond Is Worth, Lesson 69: The Delay You Remove

Leave-One-Out Test

Removing one member of a book at a time and recomputing every column on what is left. Lesson 95 does it to the four rules and finds three of the four deletions improving the book, the best of them raising net return per unit of depth from 2.000 to 2.155. The one deletion that makes it worse removes the member the return column ranked third, which is why the test needs all four columns rather than one. See: Lesson 95: The Book on One Card, Lesson 91: What the Whole Grid Earns

Liquidity

Orders that have not executed yet. They are not spread evenly across the price axis but pile up at levels you can name in advance, and lesson 25 shows the piling needs no coordination: everyone arrives at the same level by the same reasoning. See: Lesson 25: Where Liquidity Rests

Liquidity Pool

A concentration of resting orders at one level, common at round numbers and just beyond swing highs and lows. A stop is invisible until it fires, so a pool is a quantity of certain future flow at a known price that nobody can observe beforehand. See: Lesson 25: Where Liquidity Rests, Lesson 27: The Liquidity Lie

Liquidity Sweep

A break of a level that triggers the stops resting there, followed by a reversal. A cascade through resting stops needs no author, and lesson 27 sets out both what deliberate conduct actually consists of and why it cannot be identified from price. See: Lesson 35: Sweeps, Beyond the First

Losing Month

A calendar month in which a system with a real edge still ends down, which happens because the month holds too few trades for the edge to show. Lesson 96 puts the rate at 41.3 per cent for a genuine tenth-of-an-R edge at the 4.875 trades a month lesson 76 measured. The most likely year has five of them, more than one year in three has six or more, and twelve clean months happen 0.169 per cent of the time. See: Lesson 96: The Month That Loses, Lesson 76: The Pace You Actually Trade At

Loss Relief

The share of a loss that actually reduces what you pay. It, rather than the rate, is what a tax does to a system: lesson 78 shows a proportional rate with full relief shrinking the edge and the standard deviation by the same factor, so the verdict stays at 589 trades at every rate, while with no relief at all the same tenth of an R stops being an edge above 22.18 per cent. See: Lesson 78: The Deduction That Changes Nothing, Lesson 67: The Drawdown You Should Expect

M

Maintenance Call

The point at which a broker may sell your position without asking, because your equity has fallen through the level the loan requires. It is a price you can compute before you open the position: it is the loan divided by one minus the maintenance requirement, and on a position half borrowed against a 30 per cent requirement lesson 15 finds a fall of 28.6 per cent hands the decision to your broker. See: Lesson 15: When Your Broker Acts Without You, Lesson 14: Margin and Leverage

Margin

The collateral you post to hold a position, set by a clearing house from the instrument’s volatility. It is not a measure of what the position can lose, and sizing to it rather than to the loss is the commonest way an account ends: on the contract lesson 14 works, the deposit is fifty times the money actually at risk. See: Lesson 14: Margin and Leverage, Lesson 15: When Your Broker Acts Without You

Market Maker

A participant who quotes both sides and earns the spread for supplying immediacy. Lesson 53 works out what that spread has to cover: inventory risk, the cost of being adversely selected, and the fee schedule. See: Lesson 53: What the Spread Is Paying For, Lesson 4: The Spread Is the Price of Immediacy

Market Structure

The sequence of swing highs and lows a series is read as. It is not in the prices; it is produced by applying a swing definition to them, and that definition is a setting: on the twenty bars lesson 32 prints, ordinary values give eight structure events, or two, or none. See: Lesson 32: Market Structure, Lesson 33: Order Blocks and Displacement

Minimum Variance

The weights that make a book’s day as quiet as its covariance matrix allows: invert the matrix, multiply by a column of ones, divide by the sum. Lesson 73 runs it on four rules over twenty-eight moves and it asks to short two of them and to carry 3.29 dollars of position per dollar of account; forbid the shorts and the answer is 0.630 and 0.370 with two rules at zero, keeping 59 per cent of the cut at one dollar of position per dollar. See: Lesson 73: The Weights You Can Hold, Lesson 72: The Day Every Stop Hits

N

Net Edge

What is left of an edge after the round trip is subtracted, in the same units. Lesson 98 divides module 13’s 0.1230 a share by lesson 63’s 1.5443 to get 0.0796 of an R, which is 79.7 per cent of a tenth-of-an-R edge, leaving 0.0204. Because a sample size grows with the inverse square of the edge, that subtraction takes the trades a verdict needs from 785 to 18,952. See: Lesson 98: The Number That Says Scale, Lesson 93: The Rule That Trades the Most

O

Open Profit

What a position is worth while it is still open, which no statement records and every trader remembers. Lesson 89 adds up the best close inside each of lesson 63’s seven trades and gets 14.60 a share against 10.70 realised, so 26.7 per cent of everything that ever showed was given back, and four of the seven trades give back none of it. See: Lesson 89: The Money That Was Showing, Lesson 86: Taking the Profit

Order Book

Every resting limit order at every price, of which the quote is only the top line: lesson 2’s book describes 900 shares out of 7,300. Resting size costs nothing to place and nothing to withdraw, so lesson 26 prices what a wall is worth as a read and finds the answer turns on how often walls hold rather than on how large they are. See: Lesson 2: The Order Book, Lesson 26: The Order Book Is Theater

Order Flow

The sequence of executed trades with a side attached to each. Three of the four facts on the tape are reported and the fourth, the side, is inferred: lesson 7 finds the inference right about four times in five, and one wrong print of 900 shares swinging a minute from plus 600 to minus 1,200. See: Lesson 7: Time and Sales, Lesson 8: Volume and Delta

Order Type

A choice about which of two things you leave uncertain, and there is no third option: a market order fixes the fill and leaves the price open, a limit order fixes the price and leaves the fill open. The limit looks free because it collects the spread, and lesson 58 measures what selection does to it: a buy limit posted eight tenths under the close fills 16 times in 56 and still averages a loss of 0.375 a share, against a gain of 0.373 for simply buying at the close. See: Lesson 58: What an Order Type Gives Away, Lesson 3: What a Fill Actually Is

Overfitting

Choosing a rule’s settings on the same data used to judge it, so the judgement measures the search rather than the rule. Lesson 64 turns that into a bar rather than a warning: lesson 63’s winner has to clear a t-statistic of 3.54 rather than 1.65 because 253 settings were searched, which it does at 3.65. See: Lesson 64: The Price of Looking, Lesson 63: Backtesting as Evidence

P

Pace

How often a rule actually asks for an order, counted rather than assumed. Over the stretch each is defined on, 138 of lesson 63’s 253 rules produce exactly one entry and 80.63 per cent produce two or fewer. The four rules module 9 licensed ask for thirteen orders across twenty-eight days, which is 117 orders a year and 58.5 round trips, against the forty trades a month lesson 65 assumed. See: Lesson 76: The Pace You Actually Trade At, Lesson 79: Paying Yourself Before You Know

Participation Rate

The share of a period’s volume your own order represents. It sets the trade-off lesson 57 solves: go faster and pay impact, go slower and carry exposure for longer. See: Lesson 57: What a Million Shares Takes, Lesson 59: What Hurry Costs

POC (Point of Control)

The price with the most traded volume in a period. It is the mode of a binned distribution, so it moves with the bin width: lesson 29 measures it sliding 9.5 ticks across a 15-tick day, and the coarsest reading turns on 21 contracts. See: Lesson 29: Volume at Price, Lesson 30: Volume Profile

Portfolio Heat

The total risk open at once, as a share of the account: positions times risk per trade. Lesson 72 prices the day on which every stop hits together, which is the day heat was defined to survive. See: Lesson 72: The Day Every Stop Hits, Lesson 75: Which Limit Binds First

Positioning Data

A census of who holds what, published on a schedule. It is not a leak, and the standard reading of it counts one fact twice: net positions sum to exactly zero, so commercials at a short extreme and speculators at a long extreme are one sentence written down twice. Lesson 46 finds the two headline series would correlate at -0.71 by accounting alone and run nearer -0.95, which is 1.03 independent readings. See: Lesson 46: Positioning Data, Lesson 45: Correlation

Power (Statistical)

The chance a test accepts a system that genuinely works. Lesson 84’s bar over 156 trades has a power of 0.3461 against an edge of a tenth of an R, which means a real edge fails that test about two times in three. See: Lesson 84: The Share That Is Real, Lesson 19: How Long Until You Know

Profit Factor

Gross profit divided by gross loss, which is lesson 17’s two numbers written as one. Lesson 70 turns it into the bar a filter has to clear: a filter of accuracy a earns its place only on a strategy whose profit factor is below a divided by one minus a, so at 60 per cent accuracy it earns its place only below 1.50. See: Lesson 70: Machine Learning as a Filter, Lesson 39: Trading More Than One

Profit Target

A distance from the entry at which a trade is closed regardless of the signal that opened it. It is a dial and lesson 86 turns it: across seven settings from the first green close to two R and no target at all, the win rate reads six of seven in every row while the net runs from 7.64 a share to 11.74, so the exit setting alone accounts for 98.5 per cent of what the entry rule earned. See: Lesson 86: Taking the Profit, Lesson 90: The Four Habits Together

R

R

One unit of risk: the distance from entry to the stop, times size. Everything after lesson 20 is quoted in it so that trades of different sizes can be added. Lesson 85 shows the unit itself moves, by a factor of 2.711 between two halves of one record. See: Lesson 20: Position Sizing, Lesson 85: The Unit That Moved

Ratchet

The compounding trap that follows a halved account: risking a fixed fraction means the position size falls with the balance, so the earning power falls while a fixed withdrawal does not. Lesson 99 finds 58.67 per cent of ten-year paths halving at the net edge, and a median account ending at 46,526 that earns 46 a month against a withdrawal of 500. See: Lesson 99: How Long the Money Lasts, Lesson 22: Risk of Ruin

Regime

Which of two behaviours a stretch of series is in, measured rather than named. Lesson 36’s ratio is how much of the distance price travelled ended up as progress, and it carries a window: change it from ten closes to thirty and the two readings disagree on twenty-two of thirty bars. See: Lesson 36: Markets Have Modes, Lesson 37: Detecting a Regime Change

Repainting

When an indicator changes its historical signals after the fact, so a backtest reads a chart the trader could never have seen. Lesson 49 shows how to test any indicator for it yourself rather than take an assurance. See: Lesson 49: Repainting

Return Concentration

The share of a record’s result that comes from its largest few periods. Lesson 92 sorts the four-rule book’s twenty-eight moves and finds three of them carrying 3.30 of the 4.10, which is 80.5 per cent; the other twenty-five make 0.80 between them. Buying and holding over the same window is less concentrated, at 57.6 per cent, because a book of one family shares the good days and dilutes the total they are measured against. See: Lesson 92: Where the Result Came From, Lesson 95: The Book on One Card

Risk of Ruin

The chance a sequence of losses takes the account below the point it can recover from. It falls fast with a smaller fraction risked per trade and rises fast with a longer losing run than the one you planned for. See: Lesson 22: Risk of Ruin, Lesson 19: How Long Until You Know

Round Trip

One entry and one exit, and therefore the unit every cost in this course is charged in. Lesson 63 prices one at 0.1230 a share, and lesson 83 splits that into the part a second venue can compete for and the part that is a delay. See: Lesson 63: Backtesting as Evidence, Lesson 83: The Same Thing in Two Places

RSI (Relative Strength Index)

A momentum oscillator on a nought-to-a-hundred scale, and two standard implementations of it disagree: on lesson 51’s sixty closes one prints thirteen readings above 70 and the other prints none. A reading of 70 asks for an average gain seven thirds of the average loss, and what follows those thirteen readings is 58 per cent against a base rate of 58. See: Lesson 51: Oscillators Under Regime

Run and Reclaim

A level traded through and then recovered, which is the event every adversarial argument asks you to supply a frequency for. It is not a property of the market: on lesson 61’s sixty closes the frequency moves from 0.00 to 1.00 across ordinary choices of two definitions, so the page solves for the break-even instead and finds it at half the probability the level simply holds. See: Lesson 61: The Number You Cannot Look Up, Lesson 35: Sweeps, Beyond the First

S

Sequence Risk

The part of a result that is decided by the order events arrived in rather than by what they were. Flat sizing has none of it: lesson 88’s seven trades net the same 9.84 a share in all 5,040 orders. Sizing by the last outcome has a great deal, returning between 0.2559 and 1.5696 of what a flat position of the same average size would have earned. See: Lesson 88: The Size of the Next One

Session

The interval between two closures, whose edges concentrate activity for three structural reasons, none of which is mood. What is not structural is almost every number attached to it: lesson 41 splits sixty bars with no clock in them into three consecutive twenties and the efficiency ratio reads 0.021, then 0.067, then 0.333. See: Lesson 41: The Session Cycle, Lesson 40: Multi-Day Structure

Simulator

A platform that fills you at the price on the screen for nothing. It teaches the mechanics of your software honestly and it cannot teach cost or consequence, because it sets to zero every charge module 2 spent six lessons measuring: lesson 16 puts them back and a strategy reporting +$1,000 over a hundred trades nets −$700, with the breakeven win rate moving from 50% to 58.5%. See: Lesson 16: Sim Against Live, Lesson 10: Every Trade Starts Negative

Slippage

The difference between the price you saw and the price you got, which is your size divided by the size resting in front of you. Lesson 11 prices the same $30,000 at $0.30 in one instrument and $80.00 in another at the same instant, and lesson 59 shows it grows with size faster than the order does. See: Lesson 11: Slippage and Impact at Retail Size, Lesson 59: What Hurry Costs

Smart Money

A label for institutional participants. Lesson 9 names who is actually in the room and what each is optimising for, which is more useful than a two-way split, and lesson 27 finds that the conduct actually proved in court runs at institutional size against other machines. See: Lesson 9: Who Else Is Here, Lesson 27: The Liquidity Lie

Stop

An instruction that becomes a market order when price trades through a level. It is not resting in the book and nobody can see it, and when it fires it asks for whatever price exists rather than the one you named. See: Lesson 21: Where the Stop Goes, Lesson 25: Where Liquidity Rests

Sweep

Price trading past a level and coming back. Whether one happened at all is decided by three settings rather than by the market: on lesson 35’s sixty bars ordinary values of the three find four sweeps, or fifteen, or forty-six. See: Lesson 35: Sweeps, Beyond the First

T

Time & Sales (Tape)

The record of executed trades, carrying a time, a price and a size. The aggressor side your platform shows beside them is inferred rather than reported: lesson 7 finds the inference right about four times in five, and one misclassified print of 900 shares swings a minute from plus 600 to minus 1,200. See: Lesson 7: Time and Sales

Trade Record

The ten fields, written down at the time, that decide whether any question about your own trading can be answered later. Two of them cannot be recovered afterwards at any price and a broker statement contains neither: on the eight trades lesson 23 works, the statement puts the payoff ratio at 2.16 and the record puts it at 2.51. See: Lesson 23: Keeping the Record, Lesson 17: Expectancy

t-statistic

A mean divided by its standard error. It is a ratio, so it does not change when the unit does: lesson 63’s 3.65 is 3.65 in dollars, in R, and in either half of the record’s own R. See: Lesson 63: Backtesting as Evidence, Lesson 85: The Unit That Moved

Turnover

The number of times a position changes, which is what costs are proportional to rather than the number of days held. Lesson 93 charges lesson 63’s 0.1230 round trip to each of the four-rule book’s thirteen changes and takes 4.10 a share to 3.70, inverting the ranking inside the book: the fastest rule keeps 77.4 per cent of what it earned and the slowest keeps 97.0 per cent. See: Lesson 93: The Rule That Trades the Most, Lesson 95: The Book on One Card

V

VAH (Value Area High)

The upper boundary of the value area. The 70 per cent is a convention borrowed from one standard deviation of a normal distribution, and the box labelled 70 per cent holds between 70 and 84 per cent of the session depending on how it was binned. See: Lesson 30: Volume Profile

VAL (Value Area Low)

The lower boundary of the value area. An interval is a steadier summary than the point of control, 3.5 ticks of movement against 9.5, which is the reason to read it instead. See: Lesson 30: Volume Profile

Value at Risk (VaR)

A loss size a period is not expected to exceed at a stated probability. Lesson 75 reads it beside the other limits and asks which one binds first, which is the only question a set of limits actually answers. See: Lesson 75: Which Limit Binds First, Lesson 72: The Day Every Stop Hits

Variance Ratio

The variance of a multi-bar change divided by the number of bars times the variance of a one-bar change. One means a random walk and below one means the series turns; lesson 85 measures 0.3434 over sixty closes. See: Lesson 85: The Unit That Moved, Lesson 44: Volatility as a Quantity

Volume Profile

Volume counted at each price over a period. The count itself needs no inference; the summaries drawn from it do. Lesson 30 finds the point of control sliding 9.5 ticks across a 15-tick day under ordinary bin widths while the value area moves 3.5, which is the reason to read the interval rather than the mode. See: Lesson 30: Volume Profile

VWAP (Volume Weighted Average Price)

Average price weighted by volume. Institutions use it as an execution benchmark, which is what lesson 57 prices when it works a million shares against it. Nothing in this course establishes it as a level price is drawn back to. See: Lesson 57: What a Million Shares Takes

W

Wage Capital

The account a wage requires, which is one divided by the round trips a year times the edge times the risk per trade. At lesson 76’s 58.5 round trips, lesson 67’s tenth of an R and the one and a half per cent lesson 75 licenses, lesson 79 finds it 11.40 years of the wage, with the verdict on whether the edge exists 10.07 years away, and 88.3 per cent of the capital drawn as wages before it lands. See: Lesson 79: Paying Yourself Before You Know, Lesson 75: Which Limit Binds First

Walk-Forward Optimization

Fitting on one stretch and testing on the next, repeatedly, so every test sits on data the fit never saw. It removes one way of fooling yourself and not the others: lesson 65 shows that judging a record continuously, rather than at a horizon fixed in advance, takes a dead system’s pass rate from 5.05 per cent to 24.25. See: Lesson 65: The Horizon You Fix First, Lesson 63: Backtesting as Evidence

Win Rate

The share of trades that finish positive. On its own it says nothing, because the cost comes out of the winners and the losers alike: lesson 4 shows the breakeven win rate is one plus the spread as a share of your stop, all over two, so a spread half the size of your stop needs 75.0% and one the size of it cannot be beaten at all. See: Lesson 17: Expectancy, Lesson 4: The Spread Is the Price of Immediacy

Withdrawal

Money taken out of a trading account to live on, and the only term in the business with no distribution: it is the same size in a good month and a bad one. Lesson 97 puts a hundred-thousand account’s monthly earning at 487.50 and shows that a 500 withdrawal needs 102,564 to break even. It is 0.227 of one month’s standard deviation, so it stays invisible in the balance for 19.5 months. See: Lesson 97: The Money You Take Out, Lesson 80: The Cost That Does Not Scale

Worst Run

The largest distance a cumulative result falls below its own highest point, measured on a record rather than modelled. Lesson 94 puts the four-rule book’s at 1.85 a share, or 1.198 of an R, against 2.10 for a fully long position, so the book kept 88.1 per cent of the pain for 83.7 per cent of the return, and the whole 0.25 of benefit is one rule being out of the market for two days. See: Lesson 94: The Ride You Actually Bought, Lesson 95: The Book on One Card