Hacker News Reader

Comment by milesvp | Hacker News Reader

You are thinking about it backwards. Humans have a tendency to buy high and sell low. It seems to be a psychological benefit of some sort that holds us back in abstract market scenarios.

By having a fixed percentage portfolio you are forcing yourself to sell high and buy low.

This was also the only basic strategy that mathematically beats the market based on papers I read during undergraduate (there may be others now). Basically, by splitting investments among higher and lower investments that are out of phase you can make sure that you are moving money out of an investment before it falls and into it before it rises.

What I find interesting is that the advantage only works with discrete periods of rebalancing. Instantaneous rebalancing doesn’t provide any advantage. I do not understand why but I saw a paper that showed that being able to take advantage of phase shifts in nearly correlated signals goes to zero as delta t goes to zero.

View on HN

> By having a fixed percentage portfolio you are forcing yourself to sell high and buy low.

Yes, and the things you sell high are the ones that performed well in the past, so you'll have less of those in the future, which is what I said. I'm not thinking about anything backwards.

The question is whether "thing that did well in the past" is more or less likely to do well in the future than "thing that did less well in the past." This seems to vary somewhat by "thing."

If you can time the market, then by all means, do that. The reason periodic rebalancing works, is because stocks and bonds don’t exclusively go up (or down). By rebalancing you can take advantage of a racheting effects as a result of signal variance. By rebalancing at set times, you can overcome the psychological effects of waiting just one more day to get gains that then evaporate while you watch.

I’m having a hard time finding the paper around instantaneous rebalancing eroding the effects (or any good papers atm). But you can model this very easily. You can take 2 signals that randomly walk up or down. One at a “high apr” and one with a “low apr”. I’m not sure if it matters, but typically I’d expect the lower apr to have lower variance of the 2. Most of the literature around rebalancing assumes lower volatility of at least one asset class, but I’m not convinced it’s necessary from some of the math I’ve seen. You may need to add an assumption of correlation between the 2. Be sure to include code that if a signal reaches 0 it stays there. Be sure to backtest as well. Few strategies work in a bear market, but rebalancing is expected to still outperform when markets go down.

Kelly criterion is another thing to look up. It’s a mathematical look at betting stategies and what’s the biggest bet you can afford to make in the long term given that no bet is 100% gauranteed.

You're making this sound more complicated that it is. Correlations, random walks, backtesting, strategies, rebalancing, kelly criterion, have nothing to do with this.

Bonds give you cash later. Cash loses value over time.

Stocks give you a participation in the best companies in the world.

Bonds versus S&P I know which one I'm holding. Good luck with your thing.

Your falling for the same trap as most novice investors - past performance has no predictive value of future performance.

In fact, high performing equities if anything tend to fall and regress to the mean.

> In fact, high performing equities if anything tend to fall and regress to the mean.

Ok, honest question. Have you ever looked at the S&P, say over 50 years? Just simple yes or no.

Yes, have you?

It’s pretty clear high performing company don’t maintain it.

I'm talking about the index.

No, you haven't.

Because if you did, you woulnd't be looking at an exponential and saying "but but but it reverts!!!!"

Please give me an example of a company that has consistently produced annual return higher than the SP500 average over 50+ years (hell, 20+ years).

The answer is there are none. Company tend to have stretches of very high returns, followed by flat or decreasing periods.

So what most people do - buy a stock that has already gone up a lot, are basically buying high and selling low.

Reply on news.ycombinator.com