Long-Short vs Market Neutral: What the Hedge Leaves

Both books hold long positions and short positions, so the word hedged separates nothing at all. One number separates them, and it is net exposure: the long book less the short book. A long-short book keeps a deliberate net position and still rides the market's direction. A market neutral book drives that number to about nil, and what is left over is the gap between the two sides.

Start with the thing almost every description of these two approaches gets wrong by leaving out. A reader meets the word hedged and hears a promise inside it: something has been covered, something has been taken care of, the sharp edges have been filed off. Then the reader meets two labels, long-short and market neutral, and quietly sorts them into an order, with market neutral filed as the more thoroughly hedged of the two and therefore the quieter one. Both approaches hold a long book and a short book, and both can be very large. Neither the sorting nor the promise survives one minute of looking at what the two books actually hold. The word hedged is true of both and says nothing about either.

One number genuinely separates the two approaches. Worked on an invented fund's own figures, and set beside the second number that is constantly confused with it, that one number also settles what each book still has left after the hedgeA position taken in order to cancel part of the effect of another one. is on. Neither approach is the safer one, and neither arrives with a return attached. The arithmetic below describes what each book holds, and holding is not earning.

If both books are hedged, what is actually different about them?

Picture a vegetable trader in a wholesale market. She has bought a hundred crates of tomatoes and they are sitting in her shed. The crates in her shed are a long positionSomething a fund has bought and gains on if the price rises.: she has paid for them, she holds them, and if the price of tomatoes rises tomorrow she is better off. She has also done a second thing. A restaurant has agreed to buy forty crates from her next week at today's price, and she has not bought those forty crates yet. She has promised to deliver something she does not hold. The forty crates she has promised are a short positionSomething a fund has sold without holding it and gains on if the price falls.: if the price of tomatoes falls before she buys them, she pockets the difference, and if it rises she has to pay up.

Now the question everything turns on. Is this trader hedged? She certainly has both sides on. If the general price of tomatoes moves, one side of her position gains and the other loses, so something is genuinely cancelling. But she is not covered, and the reason is arithmetic. She holds a hundred and she has promised forty. Sixty crates of her position have nothing at all on the other side of them. If the whole tomato market falls twenty per cent she loses on those sixty and no promise anywhere offsets it. Having both sides on is not the same as having them matched, and the difference between the two is the only thing that separates these two approaches.

The sixty unmatched crates have a name, and the name is net exposureLong positions less short positions, expressed as a share of what the fund is worth.: what is held, less what has been promised, expressed as a share of the holder's own worth. The trader's net is sixty crates. A fund's net is the long book less the short book over its net assets. When that number is held away from nil on purpose, the book is being run long-short. When it is driven to about nil on purpose, the book is being run market neutral. Everything else about the two, and there is a great deal of everything else, is shared.

The left panel of that figure repays a moment longer than feels necessary. Five things sit in it, and every one of them is true of both approaches. Both have a long book. Both have a short book. A short position is by definition something the fund does not hold and had to get from somewhere, and a long book is very often part financed, so both are carrying borrowed money and borrowed stock. Both are running many positions at once rather than one big bet. And both have two sides that can drift apart from each other for reasons that have nothing to do with the market going up or down.

The right panel has one line in it. The asymmetry between the two panels is the point. The shared column is long and the separating column has a single row. A label built on the shared column cannot possibly identify which of the two approaches is in front of the reader. Somebody who says a fund is hedged has said it has a short book. The short book is genuinely useful information, and it is the end of what has been said.

What does net exposure actually measure?

Net exposure measures how much of the fund is still riding on market directionThe general movement of prices that most holdings share, whichever way it goes., meaning the general movement that most holdings share. Nothing more than that. The interpretation is where readers slip, so the arithmetic comes first.

Take the long book in rupees. Take the short book in rupees. Subtract the second from the first. Divide by net assetsWhat the fund is worth after everything it owes has been taken off., the fund's own worth once everything it owes has been taken off. The result is net exposure, and it is usually written as a percentage. The entire calculation is a subtraction and then a division.

Nilgiri Absolute Return Fund, invented, carries the worked figures. The fund is open-ended, managed by Nilgiri Alternatives Advisors Private Limited, invented, and at its record date its book stands at long positions of Rs 6,50,00,00,000 and short positions of Rs 2,50,00,00,000, against net assets of Rs 5,00,00,00,000. Every figure here belongs to that fund at that record date, and nothing after that date exists in this worked record. So: Rs 6,50,00,00,000 less Rs 2,50,00,00,000 is Rs 4,00,00,00,000. Dividing that by Rs 5,00,00,00,000 gives 0.80. Net exposure is Rs 4,00,00,00,000, being 80.0 per cent of net assets, and that single number says this is a long-short book rather than a market neutral one.

Read the picture as the trader's shed. The long bar is the hundred crates. The promise cancels against the holding, crate for crate, as far as it goes, so the short bar is laid along the right-hand end of the long one. The strip sticking out on the left is the net: the crates nobody has promised anything about. Four fifths of this invented fund's net assets sit in that leftover strip at its record date. If the whole market moves, that strip moves with it.

Two habits are worth building here, and both of them are about the denominator rather than the numerator. The first: net exposure is a share, and a share always needs the thing it is a share of. Rs 4,00,00,00,000 of net exposure means one thing inside a fund worth Rs 5,00,00,00,000 and something completely different inside a fund worth Rs 50,00,00,00,000. The number people quote is the percentage, and the percentage is meaningless without knowing what sat underneath it on the day.

The second habit: net exposure is measured on a day. Net exposure is a photograph, not a rule. A book that read 80.0 per cent at one measurement date is not promising to read 80.0 per cent at the next one, and a book run to keep its net near nil is doing continuous work to keep it there rather than sitting in a state that holds itself. The continuous work matters more than it sounds, and it returns below.

Why does gross exposure answer a completely different question?

Go back to the shed. Net exposure asked how many crates are riding on the tomato price. There is a second question a landlord, a lender or an insurer would ask instead, and it is not a version of the first one. How many crates is this trader responsible for altogether? A hundred sitting in the shed plus forty she has to find and deliver is a hundred and forty crates of tomato passing through her hands. She needs shed space for a hundred, working capital for a hundred, and somewhere to source forty crates she has not bought yet. Her net is sixty. Her total involvement is a hundred and forty. Both numbers are true, they measure different things, and neither one can be worked out from the other without the pieces.

Gross exposureLong positions plus short positions, expressed as a share of what the fund is worth. is the fund version: the long book plus the short book, over net assets. Same two inputs as before. One sign changed. On this invented fund at its record date, Rs 6,50,00,00,000 plus Rs 2,50,00,00,000 is Rs 9,00,00,00,000, and dividing by Rs 5,00,00,00,000 gives 1.80. Gross exposure is Rs 9,00,00,00,000, being 180.0 per cent of net assets, on the same book, on the same day, as the 80.0 per cent net.

Sit with the size of the difference for a second. Eighty and one hundred and eighty. Eighty and one hundred and eighty are not two estimates of one thing that happen to disagree. The two figures answer two questions related to each other by nothing except their shared inputs. Net exposure asks which way the book is pointing. Gross exposure asks how much of the book there is. A book can point almost nowhere and still be enormous, and a book can point very firmly and be quite small.

Why does gross matter at all, if the direction has been dealt with? Because everything that is not direction scales with gross rather than with net. Every position in the book has to be financed, and the financing is on the gross. Every short position is stock that had to be borrowed from somebody, and that lending arrangement is on the gross. Every position can be marked against the fund in a way that has nothing to do with the market's overall movement, and the room for that scales with gross. A book at 180.0 per cent of net assets has Rs 1.80 of position working for every rupee the fund is actually worth. A book at 260.0 per cent has Rs 2.60. The direction may cancel; the positions themselves do not go anywhere.

There is one more reason to keep the two apart, and it is the practical one. When somebody quotes an exposure figure, they very often do not say which of the two it is. Both are called exposure. Both are quoted as a percentage of net assets. Both come out of the same two rupee figures. On this invented fund's own record-date book the two answers are 80.0 and 180.0. The distance between them makes guessing which one is in front of the reader unreasonable.

Can two books of completely different size read the same net exposure?

Books of very different size can read the identical net exposure, and once that has been seen a bare net figure is never enough again.

Hold net assets at Rs 5,00,00,00,000 and build two books. The first is the invented fund's own: Rs 6,50,00,00,000 long, Rs 2,50,00,00,000 short. Net exposure Rs 4,00,00,00,000. Gross exposure Rs 9,00,00,00,000. The second is a plain arrangement with no short book anywhere in it: Rs 4,00,00,00,000 long and nothing sold short at all. Net exposure Rs 4,00,00,00,000. With no short book the two measures collapse into one figure, so gross exposure is Rs 4,00,00,00,000 as well.

The two books read exactly the same net exposure, in rupees and as a percentage. One of them carries more than twice the positions of the other. The second book is not a fund and is not anybody's approach; it is an arithmetic comparison put here so the point cannot be argued with. Same net. Very different amount of stuff.

Back to the shed, where the same point is obvious. A trader holding a hundred crates and having promised forty has sixty crates of net position. So does a trader who holds sixty crates and has promised nothing to anybody. Their nets are identical. But the first one needs a shed for a hundred crates, the money to have paid for a hundred crates, and a source for forty crates she does not yet hold. The second one needs a shed for sixty. Which of them is running the bigger operation admits of no hesitation, and the net number would not have helped answer it.

Exposure gets reported as a pair rather than as a single number for exactly that reason. One figure describes the tilt and the other describes the scale. Quote the tilt alone and a reader will imagine a small book. Quote the scale alone and a reader will imagine a book pointing hard in one direction. Neither imagined book need have anything to do with the real one.

What happens if net exposure is driven to nil on this book?

Here is the counterfactual that makes the whole argument, and it is worth guessing at before the answer arrives. Nilgiri's book stands exactly as it is, with one thing changed: the short book rises until net exposure comes out at nil. The long book is untouched. Net assets are untouched. More stock is simply sold short until the subtraction lands on zero.

Work it through. The long book is Rs 6,50,00,00,000. For the long book less the short book to come out at nil, the short book has to be Rs 6,50,00,00,000 too. The short book was Rs 2,50,00,00,000, so it has to grow by Rs 4,00,00,00,000. Now take gross exposure on the new book: Rs 6,50,00,00,000 plus Rs 6,50,00,00,000 is Rs 13,00,00,00,000, and against net assets of Rs 5,00,00,00,000 that is 2.60. Net exposure went from 80.0 per cent to nil and gross exposure went from 180.0 per cent to 260.0 per cent, so the book got substantially bigger on the way to becoming neutral.

The counterfactual should not be read as a story about this fund. Nilgiri Absolute Return Fund runs a long-short book, and its record-date figures are the ones already given: 80.0 per cent net and 180.0 per cent gross. The neutralised version is an arithmetic counterfactual built on its own numbers to show what neutralising costs, not a change the fund made. The full ladder of what happens to both measures as a short book is moved across a range of sizes is worked separately.

What is each book still exposed to once the hedge is on?

Most descriptions of these two approaches skip the question, and skipping it is what lets the labels do the misleading. Both books have taken something off the table. Neither book has taken everything off the table. So what is actually left?

Start with the long-short book. Its answer has two items in it. The first is the net position, and this fund's own record date puts a number on it: 80.0 per cent of net assets, being Rs 4,00,00,00,000 out of Rs 5,00,00,00,000. Four fifths of the fund's net assets still sit on whether prices in general go up or down. The net position is not a leftover or an oversight; it is the position. A long-short book keeps a net on purpose, and the net is where a large part of what happens to it comes from.

The second item is the one both books share, and it needs a name because it is the whole subject of what follows. Longs and shorts do not have to move together. The things the fund has bought can fall while the things it has sold short rise, at the same time, on the same day, for reasons that have nothing to do with any general movement of prices. When that happens the fund loses on both sides at once. An adverse move for a hedged book is a dispersion loss, meaning the two sides moving apart, and it is not a market move at all. It is a different event with a different cause, and no amount of matching the two sides in size prevents it, because matching the sizes is not the same as matching the behaviour.

Now the market neutral book. Its answer has one item in it, and the item is the second one. Direction has been removed, and the removal is real and exactly what the label claims. The gap between the two sides has not been removed, and this method cannot remove it at all. The gap is the only thing the book has left, so the whole point of running a book that way is to be left holding it deliberately. A market neutral book has not reduced its exposure to dispersion; it has arranged its book so that dispersion is the only thing it is exposed to. Those are very different sentences and only the second one is true.

Four cells and one of them changes. One cell in four is the honest summary of what the hedge buys. Read the grid across rather than down and the point lands harder: on a flat dayA day on which the general market does not move much either way., meaning a day when the general market goes essentially nowhere, the bottom row is the only row that is live, and the two columns of the bottom row say the same thing. Both books can lose money on that day. Both books can make money on that day. The market has done nothing and the outcome is entirely about whether what was bought did better than what was sold.

The trader has met this too, and she would recognise it immediately. Suppose she holds a hundred crates from one grower and has promised a hundred crates of the same vegetable sourced from a different grower. Her net is nil, so the wholesale tomato price can do whatever it likes and she is untouched by it. Then her grower's crop comes in badly bruised and sells at a discount. The grower she has to buy from has a clean crop and will not come down on price. She loses on both sides. The tomato market did not move an inch. Nothing about matching her two sides in quantity ever protected her from the two sides being different things.

One asymmetry belongs here in a single sentence, and is worked in full under short selling: the gain on a short position is bounded and the loss on one is not. The bound is arithmetic rather than opinion, and the consequence is only that the two sides of a hedged book are not mirror images of each other even before dispersion is considered. How a short is actually executed, where the stock is borrowed from, and what the broker holding all of it can require, are covered separately and are used here without being explained.

So what is the word neutral a claim about?

About net exposure. One number. Net exposure is the entire content of the word. Almost every other reading a person naturally puts on it is unsupported.

Neutral does not say the book is small. The neutralised counterfactual runs at 260.0 per cent of net assets, more position per rupee than the fund carries as a long-short book. A short position is borrowed stock by construction, and a larger short book means more of it, so neutral does not say the book has borrowed nothing. Neutral does not say the two sides will behave alike from here. And neutral does not rule out a loss. The gap between the two sides is all the book is left holding, and that gap is what produces losses on flat days.

There is one more limit on the claim, and it is the quiet one. Neutral describes a measurement taken on a day. A book run at about nil net exposure is doing continuous work to stay there, and the label describes the result of that work at the moment it was measured rather than a property the book holds by itself. Prices move, positions change value, and a book that was matched at the measurement date is not matched by arithmetic alone the next morning. A percentage struck on a date can say that much and no more.

Compare the two panels of that figure and notice the shape of the mistake. The left panel is one line long and it is precise. The right panel is six lines long and every line is a question a reader typically believes has been answered. The gap between one line and six lines is the entire distance between what the label says and what a reader hears. A label that describes one number cannot describe a whole book, and a reader who wants to know about the book has to go and read the two exposure figures.

Neither approach is better than the other, and neither is better for any particular person. The arithmetic contains no ranking and cannot be made to produce one. Neither label states a return and neither implies one. Both approaches are described structurally: by what sits in the book, and by what the book remains exposed to once the hedge is on. Refusing the word neutral as a promise and then quietly implying a ranking would do the same damage by another route.

How does somebody reading a fund report actually use these two numbers?

The two measures are not a classroom exercise. Net and gross are the first two things a person reading a fund's own reporting looks for, and they get used more narrowly and more mechanically than a reader might expect.

An analyst reading a report starts by refusing to accept an unlabelled figure. A line saying exposure at the reporting date is 80 per cent has not said which measure it is, has not said what net assets were on the day, and has not said what date convention the reporting date follows. The three questions come before any interpretation, and on this invented fund's own record-date book the first alone moves the answer between 80.0 and 180.0. An exposure figure quoted without its measure, its denominator and its date is not yet information, and the correct response is to go and get the three missing pieces rather than to guess at them.

Once the pair is in hand, the reading is a comparison rather than a judgement. Net against gross shows whether the fund is running a tilt or running scale, and the two numbers move for different reasons. A net figure that has drifted between two reporting dates says the book's tilt changed. A gross figure that has drifted says the amount of position changed. Both can move at once, in opposite directions. The counterfactual above is exactly that: net down to nil, gross up by eighty points of net assets, on one long book that never changed at all.

A person looking at an open-ended fund's own reporting has a second, duller use for the pair, and it is about consistency rather than insight. If a fund describes itself with one of these labels, the two exposure figures are where the description meets the arithmetic. A book described as market neutral is making a claim about the net figure and only about it, and that claim is checkable against the reported number on the reported date. Nothing about the gross figure follows from the label, so a reader expecting a small gross figure to accompany a small net one is expecting something the label never offered.

The household version is the same discipline, simpler and worth keeping. A person who says their savings are covered has said something about one number, usually the balance against one specific bill. The claim says nothing about how much they owe altogether, how much of it is borrowed, or what happens if two things go wrong in the same week. Somebody who hears a one-number claim as a whole-position claim has made the same mistake at a household scale that a reader makes when hearing neutral as a description of a fund's risk. The fix in both cases is the same: ask which number, ask against what, and ask as at when.