Saturday, December 29, 2012

The Incredible Story (In A Picture) of the Last Day of the Year

Last year in the 12/30/11 blog I showed that while the last day of the year used to be a bullish day for the market, that tendency has reversed this century. Below is an updated equity curve for the NASDAQ Composite on the last day of the year.

Closing up 29 years in a row is fairly astounding. Just as astounding is the abrupt end to the apparent edge.  I have no good explanation for why this may have changed, but it obviously has.

And that is something we always need to keep in mind. The market is constantly changing. It is important to always keep studying it, keep an open mind, and adapt as it evolves. Best to all in 2013! I hope it is a prosperous year for you and I hope Quantifiable Edges proves helpful along the way!


marcusbalbus said...

or it could mean nothing!

Pi said...


Wanted your comments on something. I live in India and was working on creating a sort of portfolio of short term trading strategies on the domestic index. Idea is to run not one or two, but many strategies together, and as such have seen the effect that it lower overall volatility as well as transaction costs for all the strategies put together as longs and short on some cancel each other.

After a point I realized there is no need for all the individual strategies to have equal wt, so decided to allot them varying weights based on avg profit per trade, with a cap on max amount of position open any given point of time.

Further, given the first realization that portfolio vol was much lower than sum of individual vols, I decided maybe using portfolio vol minimization might be a better way to allot weights than avg profit per trade.

Doing that would be a similar exercise to creating an efficient portfolio based on modern portfolio theory.

I created a covariance matrix etc and worked out weights. And results were pretty good. With same profitability and slightly lower avg profit per trade, but much lower volatility and a far smoother profit curve.

But till now I had only considered positive weights as the net expected values of running all strategies was positive. On relaxing the criteria to include negative weight, the system performed even better. It allotted negative weights to a few strategies, thus enabling increasing the positive weights on others while still meeting the constraint of max position at anytime. There was increased profits, better avg profit per trade, lower vol and a really smooth profit curve. This was really an unexpected thing for me, cause it didn't ever make sense to me to actually short a strategy that was making money, but when used in overall portfolio of strategies context, it improves all parameters rather than reduce it.

Now my question at the end of this entire exercise was whether this was genuine portfolio management or this was extreme curve fitting.

Your comments would be appreciated.