Boolean in Ranking: Graham NET-NET

I have a problem how to backtest [size=3]Graham Net current asset value strategy[/size] .

Net net strategy is buying those and only those stocks that whose Market cap is less than 2/3 of their Current assets minus Total liabilities:

2/3*(AstCurQ-LiabTotQ)/MktCap

Rank doesn’t work here - I need only NCAV stocks, no matter how many of them market offers. So I was thinking to use boolean Stock formula and create a Ranking system. This doesn’t work though - if backtested, portfolio is created always as entire universe.

I ran through all the p123 manuals and didn’t get anwer, but still think this is kind of elementary problem, so you guys should be able to quickly advice.

Thanks for any help,

Ondrej

Why not skip the ranking interface and just set up a simple screen:

MktCap < 2/3*(AstCurQ-LiabTotQ)/MktCap

You can backtest that; just make sure that in Main Settings, you leave the default 0 value for No. passing stocks (that would have the screen return all stocks that pass the test).

But as with anything, you should think about the universe (in screener, it’s handled via the Main Settings tab). The default choice often leaves a lot of minuscule illiquid stocks. I often run initial tests against the PRussell3000 universe. The NOOTC universe is another chice you could make. Many people also add some screening rues addressed to liquidity.

Thanks, this works!

Hi Marc,

I had a look at this as well. Even with no liquidity rules and NoOTC universe, there are very few stocks that are captured with this rule on a 4 week rebalance period (see attached). When adding a 300k liquidity filter, the screener doesn’t pick up any stocks at all. Looks like in Graham’s days there were more stocks around that were fundamentally undervalued than in the last 16 years…


After the discussion about the unexpected operator precedence in the screener, I thought I would try both;

MktCap<(2/3)*(AstCurQ-LiabTotQ)
MktCap<2/3*(AstCurQ-LiabTotQ)

As expected, the results are very different. I think the first example is correct.

Walter

Thanks Walter!