For 1990 to 2018 around the world using monthly returns w/div reinvested:the top 306 stocks (0.5%) account for 73% of global wealth creation;the top 811 firms (1.3%) account for 100% of global wealth creation;56.3% of U.S. stocks and 60.7% of non-U.S. stocks underperformed T-bills;the single top-performing firm typically explains a substantial portion of gross wealth creation in each country (New Zealand was least concentrated);Switzerland and Columbia had the highest percentage of stocks beating US T-bills;Japan and China had the worst wealth destruction (and Worldcom)."Do Global Stocks Outperform US Treasury Bills?" 9 Jul 2019 Hendrik Bessembinder et al https://ssrn.com/abstract=3415739 or http://dx.doi.org/10.2139/ssrn.3415739
What does 'top' mean? Biggest market cap?rrjjgg
What does 'top' mean? Biggest market cap?I believe it's based on "biggest total return if you held forever up to the end of the test".This is an interesting, no doubt correct, yet in many ways misleading study.Not that it's wrong, but it does assume that you hold every position forever, including all the way down till it goes bust.Thus, it seems to suggest that most stocks are losers. Yes, but only in that hold-forever situation: most companies do eventually go bust.Thus, the conclusion to draw is not that stock investing is a losing game unless you pick a rare winner,or that buying an index fund is the only sensible way to go.Those are perhaps the sensible conclusions from their work, but *only* if you do what they did: never sell anything.The companion observation, not mentioned, is that the average stock in the average month or year makes you money.The life of most firms could be seen as lifetime of varying length, most of which is profitable, followed by a relatively short decline and fall.A randomly selected interval from that trajectory will make money on average, far in excess of the short term cash rate.A broad portfolio of randomly selected stocks will tend to do extremely well over time--provided you refresh a bit, say every year or two.It will do well even if it does not contain any of the few very long term big winners.There is no need to be lucky enough to pick the very long term winners...provided you trade once in a while.Jim
provided you refresh a bit, say every year or twoprovided you trade once in a while.Jim, would you expand a bit? - trade/refresh how?Hank
Jim, would you expand a bit? - trade/refresh how?The goal is just to avoid following a dying company down into the grave.Any system you devise that would have got you out of Kodak would probably be fine.The death phase tends to be a small number of very bad (and obviously bad) years for the company's business economics and, usually, price and market cap.On average, the death watch tends to be a relatively short stretch of a firm's otherwise long and fruitful life.It seem to me that the way to avoid losses from the "hold till death" phase is merely to check once in a while that the things you currently own aren't are their way there, and replace those that seem that they might be.There are probably lots of different ways to accomplish this. Any half way plausible check for investibility would probably do.Continued index membership.Continued membership in the set of "most profitable N firms".Membership of the Value Line set.ROE still over a given modest threshold.Or crowd source it: ratio of price to 52 week high still among the top 90% of stocks.Dying firms seem to percolate to the worst 10%.Many good old MI screens start with a few "crap filters".Just don't hold stocks that fail those tests.If you merely eliminate a few big losers, it's amazing how much a portfolio's long run returns improve.Jim
Great fundamental set of observations Jim.I just did a fast and crude backtest selecting S&P 500 stocks that pass justa few of your criteria which I could implement quickly in GTR1.From 9/2/1997 till present selecting random stocks passing just some ofyour criteria with a one year hold; CAGR was 10.25% while the index was 5.4%.I say random because the screen passes almost 200 of the 500 S&P stocks.http://gtr1.net/2013/?~SP500_Filter:h253::styp.a:ne14!15!44!...RAM
trade/refresh how?A few years ago, Blackstar had a study on a trend-following strategy that only bought stocks when they hit all-time highs, because that would be the ultimate trend. They also had an exit criteria.As I recall, it did pretty well in the backtesting. I think the post-discovery results were also good, but not as spectacular as the backtest.Found a copy of the study:http://www.cis.upenn.edu/~mkearns/finread/trend.pdfA trend-following system backtested from 1991 to 2004 is kind of a self-fulfilling prophecy (19% CAGR vs 12%), but it did lose far less than the S&P 500 during the Internet collapse from 2000 to 2002.
A few years ago, Blackstar had a study on a trend-following strategy that only bought stocks when they hit all-time highs, because that would be the ultimate trend. They also had an exit criteria.Holy cow!I was setting up a screen on GuruFocus that would notify me of stocks hitting 10-year highs. I was shocked when Chipotle (CMG) showed up on the screen and went to see what I had done wrong. The last time I looked at that stock, it had dropped almost 66% from it's high just over 4 years ago. But in the last 22 months, it's up over 200%. Hitting a new high today.$758.61 -> $247.52 -> $828.80The skeptic in me wonders about all of the news and stock trading that occurred in between. :(
A few years ago, Blackstar had a study on a trend-following strategy that only bought stocks when they hit all-time highs, because that would be the ultimate trend. They also had an exit criteria.Another follow-up...I tried a very simple model on Portfolio123.com:https://www.portfolio123.com/app/screen/summary/231671Basically, hold anything that has stayed within 85% (arbitrarily chosen) of its highest price in the last 10 years (P123 limitation) since that highest price was made.Universe(SP500)ShowVar(@NHV, HighVal(2500, 0))ShowVar(@NHB, HighValBar(2500, 0))ShowVar(@LV, LowVal(@NHB, 0))@LV > 0.85 * @NHVIf I do a daily backtest that goes back to 1999 with no trading costs and no friction, it about doubles the S&P 500 return with about half the drawdown. But a very high turnover rate.Group TR CAGR DD Sharpe Sortino StdDev Correl R^2 Beta AlphaScreen 950.79% 12.03% -27.35% 0.83 1.12 12.43% 0.72 0.52 0.61 7.46%SPY 256.44% 6.33% -55.42% 0.35 0.46 14.80% - - - -Extending it to trading weekly, 4-week, and 12-week instead of 1-day progressively deteriorates the CAGR (to 9.6% on 12-week). But drawdown holds steady at about 33-35%.Unfortunately, far too much trading going on to be practical.Oddly enough, the outerformance seems to come from down days, not up days. I think that implies the "stop loss" sell criteria is more important?
Group TR CAGR DD Sharpe Sortino StdDev Correl R^2 Beta AlphaScreen 950.79% 12.03% -27.35% 0.83 1.12 12.43% 0.72 0.52 0.61 7.46%SPY 256.44% 6.33% -55.42% 0.35 0.46 14.80% - - - -
the outerformance seems to come from down days, not up days. I think that implies the "stop loss" sell criteria is more important? Yes of course.That's something I read long time ago in some SSRN paper on timing. When the market is going up, all you can do is go up with the market. But when the market is going down, you step aside and let it go down without you. You beat the market on the downside by losing less than the market.
Best Of |
Favorites & Replies |
Start a New Board |
My Fool |