Monthly feature—input wanted
Shamelessly stealing from The Wall Street Journal’s Dartboard Portfolio (which was borrowed from Burton Malkiel’s A Random Walk Down Wall Street, who owed William Sharpe a debt for the original dartboard idea), I thought I would start a monthly Monkey Portfolio. Makliel’s premise was “a blindfolded monkey throwing darts at a newspapers financial pages could select a portfolio that would do just as well as one carefully selected by experts.”
So here’s what I plan to do: near the end of each month I will use a random number generator to choose five S&P 500 stocks. For that month I will measure the return or loss of those stocks during the month versus the S&P 500 as a whole. Like an economist who assumes away the world in order to model it, I will ignore transaction costs, dividends, etc. This isn’t meant to be anything other than entertainment, but I am interested in seeing the results over a long period of time.
If you’re expecting a scientific test versus the Efficient Market Hypothesis (all stocks are fairly valued since everyone has the same information), this ain’t it. If you want a fun competition and some cool “See No Evil” monkey graphics, this is your place.
There have been many papers presented on why the dartboard…….errrrr, that is the monkey portfolio suffers from the same problems as an index fund while at the same time not taking advantage of diversification and other features that a broad index fund offers. Well, yeah, in theory. But I’m looking to see what real life yields. After all, I don’t care what theory says about a stock, other than what it puts in my pocket.
I’m planning on keeping this fairly simple, but if you think of anything I’ve overlooked or additional features you would like to see, let me know. Since I’m the one keeping up with this and compiling the stats, I’ll end up refining it as works best for me (and automating this as much as possible is my goal). But I post this now a week before I start to see what others think. For example, does it make more sense to use the Russell 2000? Thanks for any input.
So here’s what I plan to do: near the end of each month I will use a random number generator to choose five S&P 500 stocks. For that month I will measure the return or loss of those stocks during the month versus the S&P 500 as a whole. Like an economist who assumes away the world in order to model it, I will ignore transaction costs, dividends, etc. This isn’t meant to be anything other than entertainment, but I am interested in seeing the results over a long period of time.
If you’re expecting a scientific test versus the Efficient Market Hypothesis (all stocks are fairly valued since everyone has the same information), this ain’t it. If you want a fun competition and some cool “See No Evil” monkey graphics, this is your place.
There have been many papers presented on why the dartboard…….errrrr, that is the monkey portfolio suffers from the same problems as an index fund while at the same time not taking advantage of diversification and other features that a broad index fund offers. Well, yeah, in theory. But I’m looking to see what real life yields. After all, I don’t care what theory says about a stock, other than what it puts in my pocket.
I’m planning on keeping this fairly simple, but if you think of anything I’ve overlooked or additional features you would like to see, let me know. Since I’m the one keeping up with this and compiling the stats, I’ll end up refining it as works best for me (and automating this as much as possible is my goal). But I post this now a week before I start to see what others think. For example, does it make more sense to use the Russell 2000? Thanks for any input.
posted by Dwight at
9:51 PM
4 Comments:
Why conduct the experiment from a point-forward basis when you can just use historical quote data to replicate the experiment? Then repeat the experiment until your statistics stabilize.
There's really no need though. As you increase the number of iterations on your monkey experiment, your model portfolio will simply revert to the market's mean. If it doesn't, then either your random number generator is biased, or you haven't conducted enough iterations to be statistically significant.
You're effectively modeling an idealized equally weighted total-stock-market index fund. (as opposed to a classical cap-weighted index fund)
I hate to rain on your monkey parade, but this is one of those cases where a pound of theory is worth a ton of experiments.
Matthew @ Crazy Money
By
Matthew, at 4:51 PM
Possibly. But I don't buy it after seeing a decade's worth of the dartboard portfolio. The question I want to see is the cumulative impact of each iteration.
There are a lot of other factors, like losing companies being replaced by better performers. The companies are weighted when calculating the index. Etc.
Even with a true random generator I think it is possible to significantly deviate from the index mean... but I'm stubborn that way.
By
Dwight, at 5:49 PM
Wouldn't it be more fun throwing darts than picking random numbers?
By
Anonymous, at 11:56 PM
Yep, it would. Alas I don't have a dart board.
By
Dwight, at 6:58 AM
Post a Comment
<< Home