FOOLISH FOUR PORTFOLIO
Can You Top This Portfolio?

The new Workshop Portfolio is still a work in progress. Today we present a potential blend of Workshop strategies that will give us good diversification. It uses six popular Workshop strategies that have performed well in the past individually and appear to work well together. But we are betting that the community can suggest some refinements that will make the mix look even better.

	Format for printing Reuse/Reprint
	Related Links The Backtest Engines Foolanthropy
	Discussion Boards Dow Investing/F4

By Ann Coleman (TMF AnnC)
November 29, 2000

Here it comes, the moment you've all been waiting for: The Official Foolish Workshop Portfolio -- possibility number one. OK, that's not terribly exciting. What we are presenting today is unlikely to be the final version, because we are still looking for input from the community, but it's a starting point we can work with, and hopefully the process will be instructive.

One thing that has become very clear to me in this process of putting together a portfolio of Workshop strategies is just how many good possibilities there are. I am under no illusion that we will be able to pick the optimum mix of strategies. In fact, we might be in coin-flip land once we go beyond the point where we decide that we want:

1. A value strategy
   
2. Some growth strategies
   
3. Some high Relative Strength (price momentum) strategies
   
4. Strategies that go for both large- and small-cap stocks
   
5. Variety in the holding periods
   
6. 15 to 20 stocks
   

There are a huge number of strategy variations and potential combinations. I hope we will design a portfolio with synergy, but we really have no way to know if our blend will give us the synergy we want until we run it for a few years. As with all investing decisions, it comes down to doing your homework then taking your best shot and hoping for the best.

Based on the input we have received so far, Elan, Moe, and I have come up with a potential portfolio that pulls the top three stocks from six strategies. It's not perfect, but I have faith that the community will suggest refinements that will improve it. Consider it something to criticize, something to beat, bearing in mind that improvements can't come at the expense of increased risk.

We aren't just looking for the highest CAGR a backtest can find. The strategies selected need to be time-tested and rational. Also, it's more important to "buy" our high returns with as little risk as possible. So, let's play "Beat the Sharpe Ratio"!

The tentative portfolio is:

Three annual three-stock screens:

• Key-100 -- Large-cap stocks with high Relative Strength
  
• Low PB -- Stocks with low price-to-book ratios and high Relative Strength
  
• Plowback -- Companies that reinvest in themselves and have high Relative Strength

Two quarterly three-stock strategies:

• PEG-26 -- High-growth companies that haven't been wildly overpriced
  
• CAPRS -- Mid- to large-cap companies with strong price momentum (CAPRS is a new Relative Strength variation developed this year)

One monthly three-stock strategy:

• RS-13 -- High Relative Strength

This will give us 18 stocks, but there will likely be overlaps between some strategies that will reduce the number of stocks held at times.

Want to see how that mix did in the past? Use this link to set up the All-Months Summary backtester then click Run.

This blend has an average CAGR (for 12 portfolios per year, 1986-1998) of  44% per year with a GSD(M) of 32%. It has a Sharpe Ratio of 1.15 compared to the S&P 500's Sharpe Ratio of 1.05. That meets our requirement for a good risk vs. return potential, but the blend is still quite volatile with a GSD of 32%.

I've seen strategies with higher Sharpe Ratios, but none that do better with as much diversification. Yet. Can you find one?

You start by building a blend using the backtester's Blender screen. If your blend looks good for January starts, take the URL code from the blender screens to the All-Month's Summary screen. That's the true test.

Before we play with Jamie Gritton's Backtester too much, though, let's get one thing straight. What we are doing isn't scientific by a long shot. I'm not just prattling on about how past performance is no guarantee of future results -- this is not a disclaimer that you can more or less ignore. It's fact. Comparing two similar blends and picking the one with the best numbers doesn't give you the best strategy.

Here's why: If you compare two similar strategies where one has been altered to pick a slightly different mix of stocks, one strategy will very likely beat the other over a limited time period -- even if they were, in reality, exactly the same over the long run. And, even when one is better over the long run, a 16-year backtest might not demonstrate its superiority.

Odds are that if one strategy truly is superior to another over the long run, it will look better on the backtest. But one will always look best even when there is no material difference, and sometimes the worst strategy will appear better. So, we play the odds, but we shouldn't play them with any illusions.

There is one thing, though: Using the backtester to test strategies is a lot more fun than flipping a coin. You can really challenge yourself thinking of ways to improve those numbers. It's a learning process. You have to think about why RS-4 might be better than RS-13, or what kind of strategy might help fill in some bad returns during 1987 and 1989. And, you might even be right.

So, that's your assignment today. Find a better blend. Some excellent suggestions for improvement have already been posted. The thread starts with Elan Caspi's post. Drop by and let us know what you think.

Fool on and prosper!

Read More Foolish Four Reports