Long/short strategy: what you are actually paid for

A long/short strategy holds two books at once. The long book is ordinary ownership: you buy shares and profit if they rise. The short book is the mirror image: you borrow shares you don't own, sell them at today's price, and profit if you can buy them back cheaper later. Run together, the two books turn a portfolio from a bet on the market going up into a bet on the things you own doing better than the things you are short. That difference — the spread between the two books — is what a long/short equity trader is actually paid for, and almost every other design choice follows from it.

Gross and net exposure, defined

Two numbers describe any long/short book, and confusing them is the most common beginner error. Gross exposure is the long side plus the short side — the total amount of stock you are riding, and therefore the scale of everything that can go right or wrong. Net exposure is the long side minus the short side — how much of a plain directional market bet is still in there. Two books can have identical gross exposure and behave nothing alike.

What the two books are actually paid for

Run the three shapes through three markets. The assumptions, stated once and used everywhere below: each book moves with the market one for one, the names you own beat the market by 2.0 points over the period, and the names you are short lag it by 1.0 point. That is a 3-point selection spread — modest, and the only skill assumed anywhere in the table.

And now the part the sales version leaves out. Reverse the assumption — your longs lag the market by 2.0 points and your shorts beat it by 1.0 — and the dollar-neutral book loses $150 in all three markets, including the bull market. A neutral book cannot be rescued by a rising tide. You have not reduced risk; you have swapped market risk for selection risk, and selection risk is the one you are claiming to be good at.

The short side has arithmetic the long side doesn't

A long position can go to zero and no further: the worst case is −100%, and it is bounded before you enter. A short has no such ceiling, because there is no limit to how high a price can go.

• Borrowing isn't free. You pay a fee to borrow the shares. At 3% a year on a $5,000 short leg held 60 days that is $24.66 — only 0.25% of capital, but 16.4% of the $150 the neutral book earned above. Costs matter in proportion to your edge, not to your account.
• Hard-to-borrow names cost far more. The most crowded shorts carry the highest borrow rates, so the trade you most want is the one whose carrying cost most quietly eats it.
• Shorts need a margin account. A cash account cannot short at all; the usual substitute is holding an inverse ETF, a fund built to move opposite the index, which resets daily and so needs its ratio monitored rather than set.
• Shorts are usually whole-share. Fractional orders are commonly available on the long side but not the short side, so a small book's short leg lands on round share counts and rarely hits the exact dollar figure the arithmetic asked for.

Where the two sides come from

Nothing above says which names go on which side, and that is the whole job. The convention that makes long/short tractable is to pair things that are alike in every way except the one you have a view on: two companies in the same industry, exposed to the same customers, the same input costs, and the same regulator, where you can argue one is better positioned than the other. Pairing alike with alike is what cancels the shared risks and leaves your actual argument standing — pair a chipmaker against a bank and you have not hedged anything, you have taken two unrelated bets and called them a trade.

Two families of reasoning fill the sides. A fundamental pair rests on a story you can state in a sentence — this company's margins are recovering and its closest competitor's are not. A systematic book ranks a whole universe by a measurable score, goes long the top of the ranking and short the bottom, and accepts that any individual name may be wrong as long as the ranking is right on average. Both are legitimate; mixing them without noticing is how a book ends up with no stated reason for half its positions.

Beta-neutral, not just dollar-neutral

The dollar-neutral book above assumed both sides carry the same market sensitivity. Real books don't: $5,000 of a volatile growth name carries far more market risk than $5,000 of a utility, so matching dollars can leave a large directional bet hiding inside a book that looks balanced. The fix is to match risk rather than dollars, which is what a market-neutral strategy does with beta-dollars — and it is why serious long/short books are solved rather than eyeballed.

When the structure is worth it — and when it isn't

Long/short earns its complexity when you have a genuine view on relative performance — two comparable companies where you can argue one is better positioned than the other — and when you would rather not have the market's mood as a permanent partner in your P&L. It is a poor fit when your real view is directional ("chips go up"), because you are paying borrow costs and doubling your research burden to hedge away the exact thing you wanted exposure to. It is also a poor fit at very small size, where whole-share short legs and per-name costs make a clean hedge hard to construct.

The workload is the honest obstacle: two books to research, exposures to re-measure whenever prices move, borrow costs to track, and a hedge ratio that drifts on its own. That is arithmetic and bookkeeping rather than insight, which is why it is worth automating. Quantreno builds long/short baskets on your own Alpaca account — real short legs where the account supports them, an inverse-ETF hedge where it doesn't — with gross exposure, net exposure, estimated beta, and R² computed server-side and shown on the proposal before anything is placed. You approve each basket or you don't; the strategy presets are a way of working, not personal advice. The sizing discipline on this path is the solver's own: each name's weight is chosen to hit the neutrality target and scaled against the sleeve's budget. The two sides then place differently, because the account treats them differently — long legs go out as dollar amounts and can hold a fraction of a share, while short legs are floored to whole shares, since fractional shares can't be sold short. A budget too small for every short leg to reach one whole share doesn't get fudged into a worse book: the basket comes back as not placeable, naming the short legs that fell below a share and the smallest budget that would clear them. (Position sizing by the Kelly criterion is a different path — it applies to the binary event contracts of the event-driven style, where a single probability estimate makes the formula meaningful, not to an equity basket.)