Post New

Has anyone here seen numbers for the forward performance of portfolios consisting of yesterday's stock winners? Note that I do not invest this way. Like most investors on this board I look for companies that are profitable, growing and low priced, not companies that have had high returns over the recent past. Nevertheless the scarce data that I have been able to find does indicate that outperformance tends to persist, and that a portfolio of, say, the top performing 5% of stocks in the S&P 500 over the last five years would outperform the average over the next year. If true, this would be the opposite of a strategy of "Buy the dogs of the Dow," or "Don't chase yesterday's winners." It would be more like the saying, "The best forecast for tomorrow's weather is today's weather."

If anyone has data on this question, would you please share? Many thanks.

In short, market tracking.

As an example, S&P 500 companies with the best stock price total return performance in the last 5 years, held for a month or held for up to a year, little difference.
You tend to outperform in the strong "up" markets, like 2003-2009 or 202-2021, then give the advantage back again when those periods end, like 2008-2009 or 2022. A tie at the end.

What works much better is five year rate of sales growth per share. Despite being very painful in bear markets, that has shown a long term advantage of 3-5%/year over the long run. Same test as above: S&P 500 firms, top 25 by the metric, held for 1,2,3,6,12 months then rebalanced to equal weight. Like the total return test, it works well in boom markets and does poorly in bears, but unlike looking at stock performance, it doesn't give back all the relative gains during the bear.

One good version of that: every month, put 1/24 of your portfolio into each of the 8 highest-growth-rate S&P firms you don't already own, and hold for three months. After trading costs, beat the S&P by 3.51%/year 2000-2024. This may or may not be the right juncture to start such a strategy, though...who knows when the current growth bandwagon will end?


Some years back I proposed a system for finding the best equity investment strategy.
* Split all of history into broad swaths of bull markets and bear markets. Don't test a timing signal, look at the market return directly with perfect hindsight.
* Take all the strategies you've been thinking of (assuming all-long equities all the time), and see the average rate of return of each one separately in bullish and bearish markets.
* Throw away the ones that worked best in bull markets. Just use the ones that did best in bear markets all the time. If an all-long strategy does well in bears, the bull stretches will take care of themselves : )
This has the bonus advantage that you never have to try to decide if you're in a bull market or a bear market.

Jim

Has anyone here seen numbers for the forward performance of portfolios consisting of yesterday's stock winners?
...
Nevertheless the scarce data that I have been able to find does indicate that outperformance tends to persist, and that a portfolio of, say, the top performing 5% of stocks in the S&P 500 over the last five years would outperform the average over the next year.

Someone on the MI board posted a backtest of top 5/10 stocks, monthly
thread: https://www.shrewdm.com/MB?pid=963283071

S&P500, depends on starting date.
Nasdaq 100, yes, improved return.

This belongs in the MI board.

Thanks Mungofitch. Thanks Rayvt. Exactly what I was looking for.

If an all-long strategy does well in bears, the bull stretches will take care of themselves : )

Jim, are there any such strategies that perform well in bears that you’d point to?

If an all-long strategy does well in bears, the bull stretches will take care of themselves : )
...
Jim, are there any such strategies that perform well in bears that you’d point to?


I ultimately decided that the most reliable quant strategies had very modest goals. My favourite ended up being one dubbed LargeCapCash, created a few years ago.

There are a few slight variants, but a typical one goes like this:
Start with the ~1700 stocks covered by Value Line, considering only those that have a valid/populated "Timeliness" value. Any value will do (they run from 1 to 5), but it has to have one.
Of those, consider only those currently paying a dividend. (optional step, but it improves performance a hair in backtest)
Of those remaining, take the 30% with the highest reported ROE.
Of those remaining, calculate (cash - long term debt) for each firm. Not per-share values, absolute piles of dollars.
Sort on that [cash-debt] figure to find the firms with the largest cash piles.
Buy equal dollar amounts of the top 40 ranked.
Hold for two months then check: sell any position no longer ranked among the top 45, instead buy 1/40th of the portfolio value worth of the highest ranked pick which is not already held.

2000-2024 inclusive, this tested as beating the S&P 500 by several percent per year, having a better return in 78% of rolling years.
As for the issue of bear market resilience, looking at all rolling 6-month stretches that the S&P total return was negative (average -9.5%), the average 6-month return of this screen was +1.4%.
I ran this screen myself with real money for a while. It picks very large rich companies, and the largest allocation is only 2.5% of the portfolio, so you can trust it with quite a lot of money. The goal was to have something like an S&P 500 alternative which picks only companies that are more solid than average, and has a very good chance of at least a couple of percent per year of outperformance over the long run.

Figures from backtests are always suspect, but it has also beat the market since it was created. And it has lower company-specific risk than SPY which has 7 firms with higher allocations.

Jim

Post New

Berkshire Hathaway FAQ

---

Contact Shrewd'm
Contact the developer of these message boards.