Portfolio Design

Advanced concepts in stock investment portfolio design.  Fundamentals, technical analysis and many other related topics are discussed.

The Market is Always Right

What should you trade based on, your current theory of what you believe should work, or what the market tells you works? Mainstream opinion would have you believe that the "proper" way to design a stock system is to first come up with a plausible theory of what you believe should work, then implement a design based on the theory, and then backtest the design to confirm the design, and thus the theory.

Stock market analysis tools provided by Portfolio123.


There is one slight problem with this design methodology.  For every design/theory that is "confirmed", the designer may toss several (five, ten, twenty?) attempts prior to achieving a design that can be profitably traded.  Each time a design (and hence theory) is discarded, it should be counted as one of a series of optimization steps.  The designer is in fact optimizing, and in all likelihood, not keeping track of the number of optimization steps during development.  When the designer finally comes up with a design that "works", his/her heart fills with warmth, believing that the design is "proper", matching theory and the world is wonderful.

But in reality, the designer is completely unaware that the design is actually quite optimized.  In addition to selecting/discarding theories, there is the minor point of parameter optimization; a momentum trade for example, where a 200 day moving average is chosen versus a 50 day moving average, simply because the results look better.  Just a little discretion, one "works", the other doesn't, but is justifiable because it is still a momentum trade :-)  Then there is the issue of what fundamental parameters should be used.  What should be chosen?  Something based on the Dividend Discount Model... well the designer is working on a smallcap model so there are no dividends to consider. How about the Dupont business model. Lots and lots of fundamental factors that could be tried.  One of them will work just great, all in the name of proper fundamental design.  And so forth.  Do you get the point?

OK - Lets take a simple example of theory versus markets.  I have a theory that when the Federal Reserve lowers the (effective) Fed Funds Rate that the face value of LT TBonds will rise. This is a result of lowered yield making the pre-existing TBonds with higher yield more valuable. Likewise, when the Fed raises the Fed Funds Rate, the face value of TL TBonds will decrease. Does this make sense?  Its a pretty good theory, don't you think?

Well, lets see what the market thinks about this theory. I set up a one rule strategy that buys LT TBonds (Symbol: TLT) when the Fed Funds Rate is lowered (three month change in rate).  The result is shown below:

Backtest results of a system that holds LT TBonds (ETF Symbol: TLT) when Fed Funds Rate is decreasing

Backtest results of a system that holds LT TBonds (ETF Symbol: TLT) when Fed Funds Rate is decreasing

Well, that doesn't look too good.  What happens when the Fed Funds Rate is increasing?  Perhaps shorting TBonds will be a good strategy. Again, market action proves this theory wrong, as shown below.

Backtest results of a system that shorts LT TBonds (ETF Symbol: TBF) when Fed Funds Rate is increasing

Backtest results of a system that shorts LT TBonds (ETF Symbol: TBF) when Fed Funds Rate is increasing

 
 

Well that strategy wasn't too cool!  By the way, LT TBonds do not actually have to be shorted.  There is an Exchange Traded Fund (ETF) TBF that you can hold (long) that will accomplish this quite nicely.

So we have established that this theory does not "work".  It can be discarded, we can spend the next week dreaming up another theory, test it, discard it, etc.  Maybe, just maybe, a few months from now a theory will be discovered that is actually "confirmed" by backtest. We can always hope.

Or, we can have a closer look at what the market (backtest) is really telling us.  That is, when the Fed Funds Rate adjusts, it is already too late to enter a position aligned the direction of the rate adjustment. But the market overreacts and it may actually be a good time to go opposite the assumed direction.  If you were to do so, your system backtest would look like that shown below.

Performance graph of Long-Short TBonds based on direction of the Fed Funds Rate.

Performance graph of Long-Short TBonds based on direction of the Fed Funds Rate.

Now don't get too excited.  No slippage was applied and the profit per trade is slim, approximately 1% per trade.  But, for the creative designer, this could be the initial stages of the design of a tradeable system.  Or a hedge to be used instead of TLT alone.

Will the market characteristics persist?  That is the sixty four thousand dollar question. But the market action behind this phenomenon is as valid as any contrived theory.

Take care and have fun trading.

Categories:
by Steve Auger