Being a trained applied statistician, I can tell you that trying to use statistics in this context is somewhat dubious. Why? Because these returns are not samples coming from a larger population; they are the actual numbers. When we're looking at population parameters, there is no need for statistics because the differences are 100%, guaranteed, bona fide, real.rbaldini wrote: ↑Wed Jan 24, 2018 11:20 pmCan you help me unpack that table? I'm guessing that they're comparing the annual returns following days that are either above or below the 200-day moving average. I'm guessing the "p" row is the p-value for each of those being above or below the average annual return. If I'm right about that... then that is not very impressive. A p-value of 35% means that the observed result (or greater) would have been observed 35% of the time even if the null hypothesis (i.e. no difference) were true (calculated via normal or t distribution, usually). Usually people don't start perking up until you get down into the 10% or 5% level, though that's totally arbitrary. (Not sure why they wouldn't show a p-value of the difference between the two effects - looks like they only compared each to the baseline...)willthrill81 wrote: ↑Wed Jan 24, 2018 10:21 pmThe notion of the 'random walk' of stock returns has been debunked many times over and is no longer widely accepted in the academic community.rbaldini wrote: ↑Wed Jan 24, 2018 10:14 pmIMO trying to time the market isn't *terrible*. Why? Because the market seems to be pretty close to a random walk. That means that any time you exit the market, you're just replacing a random sample of (unknown) returns with 0% return over that time period that you exit. The effects of this are ultimately to (1) reduce long-term expected return and (2) reduce down-side risk / volatility, since you can't lose money while you're out. So it's not terribly different from putting some large chunk of your money into a 0-interest checking account: not optimal, but it does technically reduce risk. Of course, once you factor in the higher costs of buying and selling more frequently (e.g. short term cap gains), then it's worse.
Here's a simple table illustrating this.
When stocks are in a downward trend (i.e. below 200 day moving average), their return is far lower than when it is above the 200 DMA.
The bigger question is will this replicate itself in the future. The answer is obviously unknowable, but the 'trend' (no pun intended) would have to turn itself on its head for simple trend following to fall significantly behind buy-and-hold, and I don't think the likelihood of that event is high.
In tax-advantaged accounts these days, trading costs are low to zero. I could trade in my 401k and IRAs every day if I wanted to for virtually no cost. In taxable accounts, you're right that trend following would make less sense.rbaldini wrote: ↑Wed Jan 24, 2018 11:20 pmI've read that the important thing is that non-randomness has to be actionable enough to be profitable. One could buy and sell more often to try to take advantage of trends... but then you're paying more in taxes and perhaps trading costs, so maybe not. Seems to me that the efficient market hypothesis posits that the only predictable trends remaining would be hard to profit on - and that would include relatively short-term (< 1 year?) momentum. Similarly, house prices show a lot of momentum, I think - but, then, flipping houses is slow and costly.
As far as it being actionable, take a look at this graph from Portfolio Visualizer. A 10 month (~200 day) moving average is used with only one trade 'allowed' per month. When VTSMX is above the average, that's where all the money is. Otherwise, all of the money is moved into VBMFX. It's very simple and actionable. And this is just the result of data-mining as this type of strategy has been used for decades.
Over this 24 year period, trend following smashed buy-and-hold with annual returns of 12.11% vs. 9.58% and a lower max drawdown of -17.57% vs. -50.89%. I do not expect identical performance going forward, but this kind of performance is in line with the table above. When stocks are in a downward trend, volatility goes up and returns go down. During this time, bonds are the 'safe haven'. When stocks are in an upward trend, volatility drops and returns go up.
Why then do more investors not do this? I believe it's primarily due to a lack of patience. During a bull market, buy-and-hold is very likely to beat trend following. This occurred during the bull market of the late 1990s and again in the current bull market. It's difficult for investors to see that they're lagging behind the total market, potentially for a decade or more. The fear of missing out is very real. But for those willing to pay that price, they have been spared from the worst ravages of bear markets and saved their portfolios from the big drawdowns.
I do not expect the absolute returns of trend following to exceed or even match those of buy-and-hold of equities going forward. But I believe it will continue to be a more efficient strategy than buying-and-holding balanced portfolios.