QQQE

A quick look at QQQ’s info, I find Invesco says the PE of the index is 37. I couldn’t find the same figure for QQQE.

The P/E of QQQE is 24.52 as of Friday close.

Calculated as:
1 / (Average trailing four quarter earnings yield)
Never try to do math with P/E ratios : )

Median P/E among the 100 stocks is currently 26.32
The average across all days of that median-within-the-100 is a P/E of 25.6 since 2005.
The median among all days of that median-within-the-100 is a P/E of 26.9 since 2005.
So, to the extent that that’s a reasonable era to consider, it’s basically at an average valuation level.
(you get a very slightly different result by adding a cyclical adjustment to earnings, but they aren’t far off trend right now)

FWIW, median ROE among the 100 stocks is 23.2
Pretty impressive.

Jim

10 Likes

The P/E of QQQE is 24.52 as of Friday close … Median P/E among the 100 stocks is currently 26.32. The average across all days of that median-within-the-100 is a P/E of 25.6 since 2005.

Thanks, Jim, great info. Slightly above average valuation currently, and in the throes of a steep bear market, currently down 28.5% from peak. Who knows what will happen, but it appears likely the bear will continue. Recent lows have not seen any kind of dramatic volume that looks like a major bottom.

I’ll keep an eye on this one and aim to buy when its valuation is significantly better than average.

2 Likes

Demonstrating is providing numbers, statistics, and scientific evidence for a thesis.

The following is based on the backtester GTR1 and is from March 1, 1957 to present, with rebalance every 3 months using S&P 500 stocks

CAGR for S&P 500 weighted by market cap: 10.2%

http://gtr1.net/2013/?h63::sp500.a:et1:MktCap:tn500:MktCap:b…

CAGR for S&P 500 equally weighted: 12.12%

http://gtr1.net/2013/?h63::sp500.a:et1:MktCap:tn500:MktCap:b…

CARG for S&P 500 remove top 20 market cap and equally weight remaining 480: 12.21%

http://gtr1.net/2013/?h63::sp500.a:et1:MktCap:tn500:MktCap:b…

Regards

Craig

18 Likes

CAGR for S&P 500 equally weighted: 12.12%
CAGR for S&P 500 weighted by market cap: 10.2%

First thought: what about implementation costs?

Actual ETF results from portfoliovisualizer.com May 2003 - May 2022:
RSP: 10.02%
SPY: 9.47%

Not as advantageous but still useful.
RSP a little more volatile.

3 Likes

It’s not just a hunch, it’s based on long-standing research that small stocks and value stocks tend to outperform over the long term

Now you are confusing stock performance with valuation. Out performance happens because they grow faster, but they are still overvalued. Remember, the discussion, the thesis is mega caps’ are overvalued, hence you should do equal weightage.

The best example for you is, take a look at MSFT, WMT or SBUX or any number of firms, while they were growing they had much higher valuation and as they matured in their growth (their market cap raised) their valuation moderated.

I am not sure how exactly that article proves that mega caps were overvalued. Jim is great in deflection, using confusing but superficially convincing arguments that people eat out of his hands. I will never forget when he mentioned debt is not part of market cap and it is irrelevant and got 100 rec’s. He happily indulges in financial slight of hand. But I digress, still show me where in that article you come to the conclusion that mega caps’ are perennially overvalued.

Your question seems a bit unclear
Typically when growth outperforms therefore gets expensive, value doesn’t follow suit. OTOH when megacap value joins the party except few times growth gets further boost. Now, to begin with growth/ lower cap names are typically more expensive than the megacap (except the recent few years, primarily due to AMZN, GOOGL’s of the world and recently Apple and Softy joined the party). Historically the top of the index is not expensive but they carried higher weightage. When the mega caps raise they lift all boats further, but when small caps raise they make no impact on megacap’s.

A quick check on this proves you wrong: per Vanguard’s info, average PE of the S&P 500 is 20.2, while average PE of the Vanguard Small Cap Index is 13.6
Pretty lazy attempt. For one to switch from market-cap weighted index to equal-cap weighted index, you cannot take one point in time, but you have to have demonstrated outperformance, something like multi-decade. Go back and see historically you will see surprising results. I could be wrong, but I think if a firm is not making any profit that is not included in the PE calculation. Check it out.

I have data only from 2012 and clearly QQQE is not outperforming QQQ from that period. Jim, says he has data that goes to 1995, may be he can share it. He has access to some amazing database, I would happily take back my position if there is a data that shows equal cap outperforms mega cap. I haven’t see it yet.

“So, the main case for QQQE being a better asset is simply that it rises in value more quickly.”

How difficult would it be to put together a basket of the companies with the fastest rates of value growth?

Note that the discussion is about stock index fund, which may not necessarily be applicable to general big-cap or small-cap discussion. This is because the constituent in the index fund keeps changing, losers are removed and winners are added.

“THROUGH THE LOOKING GLASS: PREDICTING S&P 500 CONSTITUENT CHANGES”
https://insight.factset.com/through-the-looking-glass-predic….

RSP: 10.02%
SPY: 9.47%
Not as advantageous but still useful.
RSP a little more volatile.

Though I don’t doubt your figures, do remember the endpoint…the cap weight S&P has just had one of its rare great runs, like the late 1990s.
As in 1999, anything but SPY looked silly lately.

The equal weight beat the cap weight for the S&P 500 and cap-weight predecessors 68.9% of rolling five year periods since 1930.
Observation: it’s far from certain that equal weight will win.
Median advantage among those rolling 5-year periods 1.64%/year.
Observation: it’s still the better bet.

Both without fees.

Note, as always: price volatility isn’t a useful proxy for risk.
The S&P 500 is far riskier because it has vastly more company specific risk.
What can that result in?
In those rolling 5 year returns, the worst for RSP was lagging the cap weight by 8.9%/year.
The worst for cap weight was lagging the equal weight by 25.0%/year.
And presumably a greater risk of making less money over time than you might need.

Jim

6 Likes

How difficult would it be to put together a basket of the companies with the fastest rates of value growth?

Interesting question.
It’s easy to build a basket that has an edge in growth rate.
But it’s probably hard to do in a way that gives a big edge in returns over the long run.
So many people are looking for growth at any given time that more often than not those firms are overpriced relative to their eventual realized future values.
For two otherwise similar firms, imagine the faster growing one being worth 50% more but costing twice as much.
Knowing that its true value is growing very quickly might not be sufficient information.

So you might want to add the criterion “while not overpaying for that growth”, which gets tough.

I’ve tried all sorts of things.
Some tests make more sense than others, but empirically the most effective seems to be five year rate of growth of sales per share.
From among the Value Line 1700 set, an equally weighted portfolio of the 40 fastest growers on that metric would have beat the S&P 500 by 8.5%/year in the last 25 years, after trading costs.
68% chance of beating the S&P in any rolling year.

I did add a tweak to cut down on trading:
Each month, if a currently held stock is no longer in the top 45 by growth rate, replace it with the highest growth stock not already owned.

If 40 stocks is too many for you, unsurprisingly narrowing it down by momentum works pretty darned well.
I like to use ratio of stock price to 52 week high.
A 15-stock portfolio done that way adds a further 4.3%/year after trading costs. In backtest, anyway.
(find 40 by sales growth, then among those top 15 by “off high”. Checking monthly, replace anything no longer among the top 25)
24.9 year CAGR 21.9% versus CAGR 9.1% for SPY.

Congratulations, you’re now a pure growth investor. No value criteria at all.

Jim

8 Likes

“Congratulations, you’re now a pure growth investor. No value criteria at all.”

Too funny. OK. Let’s add a value criterion. And maybe a profitability criterion, too. It’s impossible to find the best 20 or 40 companies, for five years or thirty five years, but I think that we can probably do a little bit better than an index.

I apologize for taking the thread off topic, which was index funds.

1 Like

find 40 by sales growth, then among those top 15 by “off high”.

This is highest by sales growth and nearest to 52-week high, right?

This is highest by sales growth and nearest to 52-week high, right?

Yup.
Highest sales growth 40, then among those, the 15 with the highest ratio of price to 52 week high.*

Though it’s only a backtest–reality is never the same–it’s a shocking amount of extra performance for such a stupidly simple screen even without the momentum check.

It wouldn’t be surprising to me if, for example, all the performance were in the last 5 years while sales growth was almost everything.
But, again merely in the backtest, it beat SPY in 79% of rolling years in the version with the momentum check.
It made amazing money in 1998 and 1999 during the tech bull, no surprise, but also made decent money in 2000 and 2001 during the tech bear.

Jim

  • Geek note:
    Value Line calculates their 52 week high only once per month, end of calendar month, released on the first Monday that is at least a few days later.
    So, it’s actually more like the ratio of current price to highest price in the interval 1-53 weeks ago or even 2-54 weeks ago.
    This makes quite a difference: an even more “stale” 52 week high figure works even better, in general.
    It pays to be lazy.
    So bear in mind that the ratio of price to 52-week high can be greater than 1.
    You only need to worry about this obscure detail if you use a data source other than Value Line for the 52 week high field.
2 Likes

I wrote and Kingran rebutted: “It’s not just a hunch, it’s based on long-standing research that small stocks and value stocks tend to outperform over the long term”

Now you are confusing stock performance with valuation. Out performance happens because they grow faster, but they are still overvalued. Remember, the discussion, the thesis is mega caps’ are overvalued, hence you should do equal weightage.

I’m not confusing stock performance with valuation, I’m pointing out that there is a long history of research showing that when you divide the stock market into categories like large cap/small cap and value/growth, small cap stocks tend to deliver better returns than large cap in the long term (that is by itself a reason to prefer equal weight indexing over cap weight), and value stocks tend to deliver better returns than growth stocks in the long term, which also is a reason to lean toward equal share indexing over cap weight, because as a rule growth stocks are more popular and have higher price/cap for the realistic growth you can expect, and that goes right up to the biggest cap stocks.

This chart of the largest cap stocks in the US over time is interesting: https://americanbusinesshistory.org/most-valuable-companies-…

Growth stocks dominate the top spots most of the 25 years covered, with the exception of several years after the tech bust in the early 2000s when Exxon had the lead.

Kingran wrote, incorrectly: take any of your favorite valuation measurement and use it, you will be surprised to see the big cap’s are generally lot cheaper, be is PE, Price to sales, ROE or ROA, risk adjusted return.

I took an easy shortcut to show that was wrong, at least right now, by comparing the PE of Vanguards Small Cap Index (13.6) to their SP500 Index (20.2). Big caps are more expensive by PE than small caps in this example. I also pointed out that every one of the top 5 companies by market cap in the S&P 500 have PE over the index’s average, some grossly so. Here are those PEs: " AAPL: 23, MSFT: 28, GOOG/GOOGL: 21, AMZN: 56, TSLA: 100." And that is after Tesla and Amazon have already dropped by 38-40% within the last year!

9 Likes

five year rate of growth of sales per share … equally weighted portfolio of the 40 fastest growers … 8.5%/year

ratio of stock price to 52 week high. A 15-stock portfolio…

Jim, as you did all kinds of variations maybe you have the results at your fingertips available for:

A) 15 or so stocks instead of 40 for the 5-year-sales-growth criterion only?

B) Rebalancing once per year only (monthly seems to be your standard), for 40 stocks and for 15 or so?

A) 15 or so stocks instead of 40 for the 5-year-sales-growth criterion only?
B) Rebalancing once per year only (monthly seems to be your standard), for 40 stocks and for 15 or so?

As requested, all figures below are purely ranked on five year sales growth, no momentum test.
Some other hidden criteria:

  • It has to have been a stock covered by Value Line;
  • It has to have been listed in the US, or the database I’m checking wouldn’t have the price data;
  • It has to be at least 5 years old at the time (because it’s a 5-year sales growth rate);
  • …and there are no banks, thrifts, or funds allowed.
    The meaning of revenue for a bank is not the same as revenue for a product or service firm: it’s more like gross margin.
    As a reminder to investors of this, Value Line does not publish revenue or revenue growth numbers for banks.
    So they can never appear on this screen.

One other quirk: the rate of sales growth is calculated for each stock only once per year, using the annual statements I believe.
So the figure is often quite stale. Surprisingly, this isn’t a big problem; it cuts down on pointless trading a lot.
I think (?) the figure might also be smoothed a bit.

Period of analysis is 1997-05-05 through 2022-04-04, a hair under 25 years.
All figures include dividends (reinvested in the stock in question on the ex-date)
and trading costs estimated at 0.4% round trip per trade. No provision for tax.

S&P 500 total return 9.11%

40 stocks monthly with trading costs 17.51%
15 stocks monthly with trading costs 15.08%

40 stocks monthly with trading costs, replacing only those ranked below 45 17.56%
15 stocks monthly with trading costs, replacing only those ranked below 25 15.46%

40 stocks quarterly with trading costs 18.74%
15 stocks quarterly with trading costs 16.59%

40 stocks each 6 months with trading costs 19.39%
15 stocks each 6 months with trading costs 17.24%

40 stocks annually with trading costs 19.19%
15 stocks annually with trading costs 19.29%

Note: as the hold period lengthens, the statistical strength of this test falls.
So the progression of returns with hold period length is only indicative, not definitive.
It is one possible set of trades, out of many many possible such sets of trades.
Add grains of salt in proportion to the number of months held : )

If you decided to go with annually, don’t buy them all the same day.
e.g., buy several each month or two or three, and hold those for a year.
So, it’s like a few mini-portfolios, each one with annual holds, but staggered through time.
You don’t want all your trading to take place on a single day each year that might be very atypical,
with (for example) growth stocks wildly out of fashion or right on the cusp of a leadership rotation.

Jim

15 Likes

Jim, thank you so much!!!

15 stocks annually with trading costs 19.29%

With the caveat you mentioned. Nevertheless especially those last “A system for lazy people” numbers lead to the heretical question: Why Berkshire?

Why do I hold it?
(Not having known your findings)

Or you, for that matter?

3 Likes

“A system for lazy people” numbers lead to the heretical question: Why Berkshire?
Why do I hold it?
Or you, for that matter?

Because backtests aren’t reality.
Nothing returns 19%/year over the long haul.

The subtlety is trying to decide how much of any quant approach is “real”–how predictive it will be in future.
There are lots of things one might look at to make an educated guess.
How simple is it? How evenly spread through time is the performance advantage?
It it finding just 3 stocks it says are great, or 30?
Did you test just one set of purchase dates per year, or 12, or 252?

One of the best checks is how well it has done since it was published.
I used to suggest (partly tongue in cheek) that the best way to choose a quant screen is to find the one with the highest product of:
[how much it has beat the market post publication] times [months since then].

Jim

8 Likes

[how much it has beat the market post publication] times [months since then]

Jim, I just had a quick look at my own screen which I created in 2001, backtested it with data from 1994-2000, and used in the following 8 years.

The yearly numbers are roughly (“2001/02” because I rebalanced in the middle of each year):

2001/02: +11.7%
2002/03: +9.8%
2003/04: +60.6%
2004/05: +45.5%
2005/06: +0.7%
2006/07: +12.7%

“Roughly” as I just simply looked at my chronological list of buys/sales over those years, using “Stocks sold in year X” as a proxy for the portfolio performance in that year (not really correct as that stock or part of it might have been bought in one of the previous years already) instead of looking at my yearly huge and complex workbooks (It would take me a while to understand them again).

Another quick look seems to confirm that: Alltogether there were 184 complete transactions (share bought+sold) in the years 2001-2007, resulting in an average of +30%/sale, which SHOULD approximate the yearly returns.

Not too shabby? As I have no comparison (I never used any other screen than my own one) I have no idea how you, Jim, would rate those results (If interested I could email you the buy/sold list; but I won’t post it)???

P.S.: Why did I stop using it? Because in 2008 according to that quick look it resulted in -25% which was devastating for me — and not the full story as I stopped using my system but hoped for a few years WM, Beazer, Radian and others would recover - which they did not, so my system’s 2008 decision actually resulted in worse than -25%.

P.P.S.: Looking at the whole picture now, from a distance — and knowing that NO system could have been in 2008 as successful as before — maybe I should have continued using it. Too late now. Although the principle it’s based on is super-simple the whole automated system is so complex I am not sure I could still work my way into it anymore, and after all those years all the data fetching from websites would have to be coded from scratch anyway, and probably much more.

P.P.P.S.: On the other hand once a year that system then did run for many days, day and night, connected to the Internet via 56k modem. With me first thing in the morning always checking whether it’s still running or did stop with a run time error not yet properly handled.

With now having 100x that speed and more, mhmm, that would be interesting :slight_smile:

1 Like

Not too shabby? As I have no comparison (I never used any other screen than my own one) I have no idea how you, Jim, would rate those results

Outstanding.

They’re impressive for a backtest, but really impressive for a real world result.

The only thing to remember is that the date range was almost entirely one long bull market.
Not losing your shirt is a bear market is a good property to have.

I once proposed this as a method for picking what quant screens to use.

  • Pick a broad range of long-only quant screens that seem doable–a very long list.
  • Hand pick a bunch of market tops and bottoms in the past, dividing history in to good times and bad.
  • Check to see the rate of return of all your candidate screens in all those good and bad periods.
  • For each screen, give it an overall score for rate of return in bull markets, and rate of return in bear markets.
  • Forget the bull markets! Just use a mix of the several screens that did best in bear markets.
    The good times will take care of themselves : )
    And you don’t need a market timing system.

Jim

9 Likes