Replicating a Long-Short Stock Pair with Index Futures
Summary
When a trading platform prohibits shorting individual stocks but allows short positions in index futures, a desired long-short stock portfolio may be approximated using an index that contains the relevant names. The example targets a long position in one stock, a short position in another, and no exposure to a third. The proposed construction shorts the index, then buys enough of the third stock to offset its index exposure and enough of the first stock to both offset its index exposure and create the desired net long. Exposure to the second stock comes through the short index position.
This approach depends on having access to a suitable index future and knowing its constituent exposures. It does not perfectly reproduce the individual-stock short: the index future may trade at a basis relative to the replicating portfolio, creating tracking risk. The document offers a conceptual workaround rather than sizing formulas, a backtest, or evidence of performance, so practical use would require further analysis of weights, contract exposure, and basis behavior.
Key ideas
- A short index future can provide indirect short exposure to its constituent stocks.
- Long positions in other constituents can offset unwanted index exposures.
- The resulting portfolio can approximate a long-short pair when single-stock shorting is unavailable.
- Differences between the index future and the replicating stock portfolio create basis risk.
Tags
Full text
# Pairs trading stocks without shorting # Pairs trading stocks without shorting For my high school national trading competition (organised by the national stock exchange, officially starting in a week) I gathered a team of 3 friends and developed a simple pairs trading strategy. One of them fetched historical data of every stock in the exchange and the remaining two friends plugged the data to an open-source portfolio optimization library, to obtain an allocation we wanted to actually use in the game. Later on (a month ago) after reading up the manual/rules of the platform I noticed the creators explicitly described what buttons to press to get into a short position on a stock. My teammate modified the script to fetch live data and made an email sending class, on the basis of which I implemented (and afterwards backtested) said pairs trading strategy. Now the funny part: as the mock trading session started and we exchanged some messages with the organizers, it turned out the platform does not support shorting stocks. With our whole amateur data pipeline working, and me in the process of extending it to include kalman filters/DDQN suggestions (the platform prohibits automatic trade execution, so we wanted to place orders by hand) I am frankly lost in what to do further. My ideas included: - running the script anyways to log the trades and see the returns in 3 months (because this is how long the game is) - find an alternative trading competition and deploy the strategy there. I couldn't really find any as the registration dates passed already and they require me to be a university student. I would greatly appreciate any comments :) thanks Edit: Apart from buying regular stocks/etfs the platform offers access to futures (which will be available only in the second stage) ## Answer by Dimitri Vulis (score 1, accepted) https://quant.stackexchange.com/a/77290 Suppose, without loss of generality, that your universe contains assets $A$, $B$, and $C$. You want to be long $A$, short $B$, and flat $C$. But you're not allowed to short single stocks. What do you do? There may be an index that you can short using exchange-traded futures that consists of $A$, $B$, and $C$. - Short the index, - be long enough $C$ to offset your exposure in the index and flatten your exposure to $C$ - leave $B$, so your exposure to $B$ is via the short position in the index. - be long enough $A$ to not only offset your exposure in the index, but be net long $A$. However there is a risk that the index will be trading at some basis to your replicating portfolio.
Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.