= 27d567bc2d48d9f40e9ccc20ea370b15)
GuySrinivasan comments on Why you should consider buying Bitcoin right now (Jan 2015) if you have high risk tolerance - Less Wrong
You are viewing a comment permalink. View the original post to see all comments and the full post content.
= 783df68a0f980790206b9ea87794c5b6)
Loading…= b715d02e85f7b3b4ae65ca3d3e450868)
Subscribe to RSS Feed
Powered by Reddit
= f037147d6e6c911a85753b9abdedda8d)
Comments (136)
Fleshing out my intuition.
For that argument for DCA to go through, we must justify that it's the correct argument to choose from these three:
For example, lets say I take $3000 and buy a stock all at once at $50 for 1 share. I get 60 shares. Now instead, what if I buy $1000 each at 3 different times, once at $40 for 1 share, once at $50 for 1 share, and once at $60 for 1 share. I end up with 25+20+16.66 shares = 61.66 shares, even though the average price per share I bought at was identical. (original argument)
For example, lets say I take $3000 and buy a stock all at once at 0.02 shares for $1. I get 60 shares. Now instead, what if I buy $1000 each at 3 different times, once at 0.025 shares for $1, once at 0.02 shares for $1, and once at 0.0167 shares for $1. I end up with 25+20+16.66 shares = 61.66 shares, because the average shares per dollar I bought at was 0.02057 which is better than 0.02. (original argument with prices preserved, "average metric was the same" changed, and conclusion changed)
For example, lets say I take $3000 and buy a stock all at once at 0.02 shares for $1. I get 60 shares. Now instead, what if I buy $1000 each at 3 different times, once at 0.025 shares for $1, once at 0.02 shares for $1, and once at 0.015 shares for $1. I end up with 25+20+15 shares = 60 shares, because the average shares per dollar I bought at was 0.02 which is identical to 0.02. (original argument with prices changed, "average metric was the same" preserved, and conclusion changed)
Off the top of my head whether the DCA (dollar cost averaging) works depends on the persistence of the underlying time series (returns on the asset you're buying).
If the asset returns are following a random walk, DCA is useless. It won't hurt you and it won't help you.
If the asset returns are persistent (momentum dominates), DCA will hurt you and decrease your performance.
If the asset returns are anti-persistent (mean-reversion dominates), DCA will help you.
Thing about the stock market is, if either momentum or mean-reversion dominated, people would use that in trading algorithms and destroy the phenomenon. Over the long run, it can be safely assumed to be neither, so DCA doesn't hurt you(other than perhaps by delaying your investments, if you're thinking of streaming a lump-sum in over time instead of investing it all right away, but most people invest from paycheques instead of from lump sums), but it does lower variance.
Some people disagree.
Compared to what?
That may be the single most colloquial academic paper I've ever seen. They do lay out a decent case, but remember that they're discussing a different kind of momentum than we are - we're discussing momentum of market returns, they're discussing momentum in relative ranking of different securities. Also, I tend to take it as given that if a simple stock-trading strategy produced returns that were that superior, the hedge fund crowd would have jumped on it with both feet by now. Some of the hedge fund strategies I've seen have exploited far smaller inefficiencies to make significant returns.
Compared to saving up a lump sum and investing that. And it sure returns better than just not saving at all, which is the usual default of most would-be investors.
In practice, DCA is usually used by retail advisors(of which I am one) as a psychological argument that investors should ignore short-term market turmoil in their long-term investments. Frankly, any argument that makes retail investors stop buying high and selling low is doing the lord's work, and DCA is even mildly accurate.
Do note that the main author of that paper runs a hedge fund.
That's apples and oranges, isn't it? All you're saying is that holding cash is less volatile in nominal terms :-)
That's an entirely different claim from saying that DCA improves your returns (or your Sharpe ratio).
Noted.
It's the same argument that the above paper is making, just in the opposite circumstances.
DCA says investing into a down market improves your returns, and the best way to systematically do so is to to just always be investing, which is accurate and a useful psychological argument besides.
This claim is not compatible with EMH.
I'm not speaking of trying to time investments. I'm speaking retrospectively - if you can invest at a time that is a local minimum in retrospect, that money will grow faster.
Also, I'm not a big believer in the EMH - it's valid over a period of seconds and over a period of decades, but in between, psychological effects and the joys of market timing(i.e., knowing an investment is crap, but riding the bubble anyways) can swamp EMH easily.
It's certainly true that a time machine can produce outstanding investment returns :-/ but I don't see what does it have to do with DCA.
That's perfectly fine, but if you don't believe in EMH and believe that DCA improves your returns, then this necessarily means that you believe that markets are mean-reverting and in that case there are better than DCA ways to take advantage of it.