Glossary

Every word the desk uses and every number a record is judged by, in plain language, with an everyday example. On the desk and in the manual, a dotted word opens its entry in place.

What the desk calls things

Alpha0 to 100

The desk's score for a name after the market's share of its move is taken out: what the stock did on its own, ranked against every other name. 50 is ordinary; a close above 60 is a name moving on its own strength, below 40 on its own weakness, and those two lines are where runs start and end.For example: Two shops on one street both sell more in December. Alpha asks which one sold more than the street did, and ranks that.

The tapefour states

The desk's one-word read of the whole market: Trending, Flat, Grinding bear or Dislocation. Each trade shape is offered only in the tapes where it has paid before.For example: Weather for traders. You can run in rain or sun, but you dress for the one you are in.

A runstrength or weakness

A stretch of sessions in which a name keeps moving on its own, up (a strength run) or down (a weakness run). Most end inside two weeks; the run clock on the engine page says how many are still going by day.For example: A hot streak at cards. It is real while it lasts, and the clock is the honest part.

Hedged shorta shape

Short a weak name and, with the same dollars, own its sector fund. The market and the sector cancel; what is left is the name's own weakness.For example: Betting one runner finishes last in a race while also betting on the race's average time. Whether the race is fast or slow no longer matters; only the runner does.

Cycle paira shape

Own a fresh riser and short a stale leader in the same sector, sized so each leg carries the same daily risk. A bet that leadership rotates inside a sector, not on the sector itself.For example: In a relay, backing the runner who just took the baton against the one who has run two legs already.

Dislocation bouncea shape

In a sell-off, own the hardest-hit tenth of a sector against the sector fund, for five sessions at most. The short-lived snap-back after an over-reaction, and nothing more.For example: A rubber band pulled too far. You are betting on the snap, and you let go after five counts whether it snapped or not.

Hedge

A second position that cancels a risk you did not want. On the desk it is the sector fund or the market fund held against a name, so that only the name's own move is left.For example: An umbrella. It does not stop the rain; it stops the rain reaching you.

Spreadpairs

The gap between two legs, drawn as one line: what one dollar long of A against the hedged amount of B has earned since a chosen start. A pair trade is a bet on that line, not on either leg.For example: Two cars on a motorway. You are not betting on either car's speed, only on the gap between them.

Sector fund

The exchange-traded fund that holds a sector (XLK for technology, XLE for energy, and so on). The desk measures every name against its own sector fund and uses the fund as the hedge leg.For example: The class average. A pupil's mark means more next to it.

Net betapairs

How much market a two-leg trade still carries after the long and the short partly cancel. Zero means the market can do what it likes; 0.3 means the trade still leans a third of the way with it.For example: Two people on a seesaw. Net beta is how far it still tips.

Slot20 of them

One of the twenty places in the house book. Every qualified trade takes a slot at the close it posts; when all twenty are taken, the next must wait for an exit.For example: Twenty seats at a table. Nobody stands.

Equal daily risksizing

Each position is sized so that a normal day moves it by about the same number of dollars as every other position. A jumpy stock gets fewer shares, a quiet one more.For example: Packing for a trip by weight, not by number of items. The coat and the socks both get their fair share of the bag.

Day capexits

The most sessions a trade may stay open, whatever it is doing. 21 for runs, shorts and pairs; 5 for a bounce. The clock exit, so a trade that stops working is not kept out of hope.For example: A parking meter. When it runs out you move, even if the spot is nice.

Exit rule

The condition, fixed before entry, that closes a trade: the name's Alpha back through its line, either leg of a pair under 60, the bounce's five-day move back to zero, or the day cap. The number on the line was tested with that exit and no other.For example: A recipe's cooking time. Changing it afterwards means the dish on the box is not the one you made.

Watchingthe watch list

Following a name or a pair without tracking it: a star on its ticket, its tile, its page or its row. The watch list sits under the tracker and on your page, each card saying what the name needs to make today's list, or where a pair's spread sits in its range. Pair alerts, on Pro, go out for the pairs you watch.For example: A shortlist pinned to the fridge. Nothing is bought; it is what you look at first.

Trackingyour page

Following a posted trade on your own page, with your own entry and your own decisions, graded against the rule's result on the same trade. Took, skipped, closed early: each is scored.For example: Doing the homework beside the answer key, and seeing where you differed.

The recordthe house book

Every trade the desk has posted, taken by the house at the posted close, marked daily, closed by its own rules, after costs. Written down each settled evening and hashed with the day before, so no day can be edited later.For example: A diary written in ink and photographed each night. You can read it; nobody can rewrite it.

Hashthe chain

A fingerprint of a day's record, computed from that record and the previous day's fingerprint. Change one letter of one day and every fingerprint after it changes, which is why the record cannot be quietly rewritten.For example: A wax seal on each letter, pressed with the seal of the letter before it.

Replay

The rules run on past prices as if they had been followed day by day, with costs charged, to see what they would have done. The numbers on each trade line and the curves on the engine page come from it. It is a test, not a forecast.For example: A flight simulator with real weather from past years. It tells you how the plane handles; it does not promise tomorrow's sky.

Settlesessions

The evening pass after the close, when the day's prices are final: Alpha is recomputed, the list is posted, the house book is marked, the day is hashed. A session is one trading day.For example: Closing the till. Nothing counts until it is counted.

Universe

The names the desk follows: a curated set of large, liquid stocks across the sectors, each with its own page. Every number on the desk is relative to the universe and the sector funds.For example: The league table. A result only means something against the teams in the league.

True peersstock pages

The names whose own moves, with the market stripped out, most resemble this one's over the last six months. Often the obvious competitors, sometimes not.For example: The people whose mood matches yours day to day, whoever they turn out to be.

Stretchpairs

Where today's spread sits inside its own last 120 sessions, from 0 (never lower) to 100 (never higher), at equal daily risk. The top and bottom quarters are the stretched ones; the middle is ordinary for this pair.For example: A thermometer that only knows this one patient: 100 is the hottest they have been all season, 50 is their normal.

Sector paira shape

Long a sector fund whose own tape is Trending against a sector fund whose own tape is a Grind or a Dislocation, dollar for dollar, offered in a dislocation only; out when either sector's tape leaves its state, or on day 21. A bet that two sectors keep disagreeing for a while.For example: Two lanes on a motorway in a jam: one still moving, one stopped. You bet the gap holds until the traffic sorts itself out, and you leave the moment it does.

Closes or driftspairs

Whether a spread has come back after moving away, over its recent history: closes (gaps have tended to close inside a trade's window), closes slowly, or drifts (it has trended, not returned). A pair is a bet that a gap closes; a spread that drifts is a trend wearing a pair's clothes.For example: A friend who is usually five minutes late and then on time again, against one who gets later every week.

Past stretchespairs

A count, not a model: the last times a spread sat at least as far from its one-year average as it does today, and how many of those times it closed half the gap within ten sessions. Shown only when the spread is stretched now.For example: Of the last ten times the river was this high, eight were back to normal within a week.

Typical swingpairs

How far a spread normally moves over a chosen number of sessions, from its own history. Used to say whether today's gap is large or ordinary for this pair.For example: Knowing a friend is usually five minutes late. Twenty minutes means something; five does not.

Trading in probabilities

Return

What a position or a book made, as a percentage of what was put in, after costs. On its own it says nothing about how it was made or what it took to sit through.For example: Two taxis get you across town. The fare is the return; the route is everything else on this list.

Volatilityvol

How much a thing moves from day to day, as a yearly percentage. Higher means bigger swings in both directions, not worse results. A calm book at 10% volatility and a wild one at 30% can earn the same return and feel nothing alike.For example: Two drivers with the same average speed. One holds it; the other alternates 20 and 120.

Drawdownmax DD

The fall from a high point to the next low, before a new high. The maximum drawdown is the worst such fall in the record: the most you would have been down had you started at the worst moment.For example: Climbing a hill that dips. Drawdown is the deepest dip, measured from the top you had already reached.

Sharpe ratio

Return per unit of volatility: how much you were paid for each unit of swing you sat through. Above 1 is good for a book held a long time; the market itself is usually below 1.For example: Miles per gallon. Return is the miles; volatility is the fuel.

Sortino ratio

Like Sharpe, but it counts only the downward swings as the cost. A book that swings up a lot and down a little scores better here than on Sharpe.For example: Miles per gallon counting only the fuel burnt going uphill.

Betato the market

How much of the market a position or book carries. 1.0 moves with the index, 0.35 about a third as much, 0 not at all. A return earned at beta 1 was the market's return; at beta 0.35 most of it was something else.For example: Walking on a moving walkway. Beta is how much of your speed is the walkway's.

Alphathe return kind

The part of a return left after the market's share, at the book's beta, is taken out: what the method earned on its own. Not the same thing as the desk's 0 to 100 Alpha score, which is the same idea applied to one name's ranking.For example: Your pay rise after inflation. The rest was just the money changing size.

t-scorethe t-statistic

How far a result sits from zero, measured against its own day-to-day noise. Around 1 could easily be chance; 2 is the usual bar for treating a result as real; 4 is well past it. A return with no t-score beside it is a number with no weight behind it.For example: A coin that comes up heads 6 times in 10 proves nothing. 600 in 1,000 is a bent coin. The t-score is the difference between those two.

Hit rate

The share of closed trades that made money. On its own it is not a measure of edge: a 40% hit rate with wins twice the size of losses is a profitable book.For example: A striker who scores on 4 of 10 shots can win the league if the goals count more than the misses cost.

Profit factorPF

The sum of all wins divided by the sum of all losses. 1.0 is break-even; 1.5 means the winners paid for the losers one and a half times over.For example: Takings over costs at a market stall, before you ask how many customers there were.

Basis pointbp

One hundredth of one percent. The desk charges 2 basis points a side on each stock leg and 1 on a fund leg as the cost of a trade, which is roughly what a liquid name costs to deal in.For example: A penny on a hundred-dollar bill. Small, but a book that trades often pays it often.

Dividend-adjustedprices

A price series where each dividend is folded back into the earlier prices, so a stock does not appear to fall on the day it pays out. Used wherever two funds are compared over years.For example: Reading a wage history with the bonuses added back in, so a bonus month does not look like a pay cut the month after.

The four fundsthe factor read

Four public index funds any book can be partly copied with: the market itself, small names against large, cheap names against expensive, recent winners against the market. The engine page sets the book against them to show what is left over: the engine's own part.For example: Four flavours a recipe can be reduced to. What is not any of the four is the cook.

Copied by the fourR squared

The share of the book's daily moves the four funds account for, as a percentage. 39% means six tenths of what the book did day to day was its own; 90% would mean it was mostly a basket of index funds in disguise.For example: How much of a cover song is the original. At 39% it is a different song that borrows a chord.

Percentile

Where a value sits inside its own history, from 0 (never lower) to 100 (never higher). The desk uses it to say whether today's reading is ordinary for this name or this pair.For example: Being in the 95th percentile for height in your class: taller than 95 of 100 classmates.

Correlation

How much two things move together, from −1 (opposite) through 0 (unrelated) to +1 (in step). The desk strips the market out before measuring it, so two names are not called peers merely because both went up with the index.For example: Two commuters who leave at the same time every day are not friends. Two who change their plans together probably are.

After costs

Every number on the desk has trading costs taken out first, at the basis-point rates above, charged on every leg at entry and exit. A number shown before costs would flatter a method that trades often.For example: The price on the menu including the service charge, so the bill holds no surprise.

Edge

A tilt in the odds that shows up over many trades and not on any one of them. The engine page shows each shape's edge as the average result per trade in each tape, after costs, with the count beside it. One trade proves nothing either way; the count is what makes the average mean something.For example: A casino's edge at roulette is a few percent. It loses plenty of spins and never a year.

Expectancyper trade

What a trade is worth on average before you know how it ends: the hit rate times the average win, less the miss rate times the average loss. A positive expectancy with a 40% hit rate is a good business; a negative one with an 80% hit rate goes broke slowly.For example: A raffle ticket that costs $1 and pays $5 one time in four is worth $1.25. You lose three of four and still want every ticket you can get.

Sample sizethe count

How many trades a number rests on. A 70% hit rate from 10 trades is a coin that landed well; from 400 it is a property of the method. Every cell on the engine page shows its count for this reason, and the desk does not quote a cell with too few trades in it.For example: Tasting one spoon tells you about the pot. Tasting one grain of rice tells you about that grain.

Variancestreaks

The ordinary scatter of results around their average, which produces runs of wins and losses even when nothing has changed. A method with a real edge still has losing months; a method with none still has winning ones. The record is read over seasons, not weeks.For example: A fair coin lands five heads in a row about once every 32 tries. Flip it all afternoon and it will happen; the coin is still fair.

Regression to the mean

Extreme readings tend to be followed by less extreme ones, because part of any extreme is noise that does not repeat. It is the reason a pair far outside its own range is worth a look, and also the reason a hot month is not a trend.For example: The tallest parents have tall children, but usually not as tall. The best week a shop ever had is rarely followed by a better one.

Random walknoise

A path where each step is unrelated to the last, so the past tells you nothing about the next move. Most of any price chart is this. The engine's job is to find the parts that are not, and the t-score is how it tells the two apart.For example: A drunk walking home: where he is now says little about where he will be in a minute, though he does, on average, drift toward the door.

Base rate

How often something happens in general, before you know anything about this case. The pair page's past-stretches count is a base rate: of the times this pair was this far out before, how many came back. Start from the base rate and let the particulars move you from there, not the other way around.For example: Most coughs are colds. A doctor who starts from the rare disease will be wrong far more often than one who starts from the cold.

Risk of ruinsizing

The chance a run of ordinary losses leaves you unable to continue before the edge has time to show. It is governed by position size, not by the edge itself: the same method survives at 5% a line and dies at 50%. The desk sizes every line from your account so the worst run on the record stays survivable.For example: A gambler with the better odds still goes broke if he bets the house on each hand. The casino bets small and plays forever.

Nothing by that name. The manual may have it.

Missing a word? Tell us and it goes in. The manual's chapter on judging a record: Returns are not the number.