Long/Short Backtesting Set up

Question

I am going to be backtesting a Long/Short equity strategy and need some guidance on how best to deal with the short book. I was thinking that for each portfolio I would go long 50 equities and go short 50 equities. I plan on doing this test in excel by the way.

I am trying to wrap my brain around how to appropriately handle this.

My thought was:

Portfolio 1

Longs MSFT CSCO ORCL AMZN Portfolio Returns Benchmark Returns Out/Under-performance

Shorts HII HP BBBY YHOO Portfolio Returns Benchmark Returns Out/Under-performance

My first question is whether I would simply flip the sign on the short book in order to get performance for that piece of the portfolio. And secondly whether I would simply add short performance to long performance to get net performance. I would like to include some assumptions about margin but not sure how to appropriately do this -- If you have any suggestions I would appreciate it.

Answer 1

You should first determine whether you want to look at relative or absolute returns. You may want to use position weights relative to the benchmark rather than market value if interested in relative value.

For absolute returns consider your three components (long, short and cash - where cash includes borrowing, other costs and in/out flows) P&L and compute simple returns for each period. A period would be created when a change occurs to any position within any of the three components. An alternative would be to compute simple returns after all costs for only long and short positions rather then managing costs through the cash components. This second approach would consider your margin in the divisor (compare this demonstration).

Once all period returns are known you would typically chain-link the period returns to compute the time-weighted return.

Answer 2

To add an important point to existing answers, the overall portfolio beta should be 0.

The usual motivation of a long short strategy is to invest in a portfolio that captures your view of individual equities, but is immune to overall market movements. For example you have reason to believe that MSFT is going to perform better than HP, but you do not want your portfolio to be affected by (for example) the release of macro-economic data would cause all stocks to drop (or go up) by some amount.

The main risk factor for every stock is the market, usually approximated by the S&P 500 index. To create a long short strategy that is immune to the S&P 500, you have to estimate the beta's of each of the stocks. You can either calculate it or get it Google finance.

Long equities will have positive beta and short equities negative beta. The portfolio value weighted beta should be close to zero. This strategy, can theoretically offer a great Treynor ratio, if there is value added by the stock picking!