Designing a Short Ranking System

Research
1

I tried reversing the directions of the leaf nodes of a well-performing Long ranking system (Simple Value Momentum), but the short performance isn’t very good. See http://www.portfolio123.com/app/ranking-system/94186

Note i cannot simply use Rank<1 long recommendations, because they contain N/As.

Questions

  1. Is there a good public Short ranking system i can copy? Pl. recommend

  2. What are good factors to use for a Short ranking system? I could try reversing the top-performing factors (i.e. yielding best return in top 5 percentile) in Dan Parquette’s spreadsheet (http://www.portfolio123.com/mvnforum/viewthread?thread=886#2797) , but not sure it’d help - result might be similar to the one above. If this doesn’t help, any other ideas for designing a short ranking system?

  3. In Short simulations, I’m considering using -ve trading costs and evaluating them based on worst performance - will that give me a good Short trading system?

thanks
Kiran

2

Hi,
Short strategies have been discussed in the forums before. Not sure all the sims posted are still available, but the first one in this thread is:
http://www.portfolio123.com/mvnforum/viewthread?thread=2660

I have had good success with rankings by identifying top factors (updating Dan’s spreadsheet) and then inverting most factors for a short ranking. But some of the factors I use are not in the public version of the spreadsheet and there are some issues with the fundamental factors due to NA’s.

To accurately track trading costs, as you suggest, you need to add negative slippage and then evaluate based on the worst performance. In lieu of drawdown I review the stats page to ensure there aren’t too many significantly positive months.

Don

3

The only systems still around from that thread are based on a ranking system with one big flaw, in my opinion. One of the factors is short interest as a percent of float, with stocks having a high short interest getting a higher ranking, and therefor being shorted in the Sims or Ports. This is unrealistic… you’re gonna have less chance of being able to find shortable shares on stocks with high short interest, since a large amount have already been lent out.

A possible way to address this is to compare the percentage of a stocks’ float held by institutions to its short interest. If stock XYZ has 30% of its float held by institutions and 15% of it shorted, then you can assume another 15% is available for shorting, not the remaining 85% of float.

So for stocks that have low institutional ownership and high short interest, although the short interest might on average predict a price decline, your chance of taking part is low.