QQQE

I’m generally more of a value investor, but I had a little spare cash sitting in our taxable account from interest and dividends that hadn’t yet been reinvested. So, took a little nibble of QQQE at what would appear to be near the 52-week low, on a day when the Nasdaq fell around 3-4%. There’s always the potential for it to fall further, but it is diversified (to a degree), so for time-frames of 3 years+, I think it’s a pretty good bet that it will be somewhat higher before I’m ready to sell. I’ve got a corporate bond maturing in the fall, so depending on the situation and what else catches my eye, I could add funds to the position at that time.

6 Likes

Should work out fairly well for purchases at prices under $70.
Based on historical valuation levels, particularly since around 2005, a “normal” price might be closer to $59-64 or so, but that’s not enough of a difference to make you go broke.
Especially if valuation levels in the next decade [continue to] average higher than those from 20 years ago.

The main cases for QQQE as I see them:

  • Surprisingly diversified in one sense: the largest position is 1% of the capital, so company specific risk is low.
    Very much lower risk than QQQ or the S&P in that way.

  • And the main thing: over very long periods, this group has had a tremendous rate of growth in old fashioned earnings back at least to the 1990s.

I use the median earnings yield among the 100 stocks as a value yardstick.
Since it’s equally weighted, the median and average won’t diverge much except when there is a big outlier.
But the earnings yield outlier will be a small position of capital, so it’s best ignored.
On that median earnings metric, the Nasdaq 100 Equal Weight set earnings have risen about inflation + 8.2%/year for ages, which dwarfs the figure for the S&P 500.
The earnings see transient dips in recessions, but they seem always to return to the prior after-inflation trend.
No matter how speculative some of the issues are and often untethered from earning power,
it seems reasonable to think that their average value does rise over time as fast as their average earnings rise.

Vanishingly few individual firms rise in value with such predictability at a rate approaching inflation + 8%/year.
So if you can get a return that high with so much less risk,
The only argument I can think of against it is you end up buying Facebook stock (the boss is a ***) and Tesla stock (the valuation is a ****).
And of course, I like to buy things when they’re really cheap.

Jim

Data appendix:

These figures are arbitrarily scaled, but they’re worth plotting as a graph for fun.
Each figure is the trailing year inflation-adjusted earnings figure, derived from the median earnings yield as described above.
The figures are snapshots every 6 months since the start of 2005.
First data column is the log of the real earnings figure; right figure is the real earnings figure itself.
The “log” column gives the nice graph of the trend. A linear trend line gives R2 of 0.9406

Ending      Log     Earnings
2004-12    3.357     28.7  (calendar 2004)
2005-06    3.501     33.1
2005-12    3.412     30.3
2006-06    3.523     33.9
2006-12    3.622     37.4
2007-07    3.721     41.3
2007-12    3.662     38.9
2008-06    3.778     43.7
2008-12    3.687     39.9
2009-06    3.554     34.9
2009-12    3.572     35.6
2010-06    3.846     46.8
2010-12    4.024     56.0
2011-07    4.141     62.9
2011-12    4.119     61.5
2012-06    4.170     64.7
2012-12    4.089     59.7
2013-06    4.189     66.0
2013-12    4.190     66.0
2014-06    4.251     70.2
2014-12    4.341     76.8
2015-07    4.346     77.1
2015-12    4.248     69.9
2016-06    4.234     69.0
2016-12    4.249     70.1
2017-06    4.430     83.9
2017-12    4.447     85.4
2018-06    4.405     81.9
2018-12    4.483     88.5
2019-06    4.546     94.2
2019-12    4.619    101.4
2020-06    4.720    112.1
2020-12    4.627    102.2
2021-06    4.720    112.1
2021-12    4.686    108.4  (calendar 2021)
13 Likes

Has anyone ever run a backtest on top 25 or 50 of the QQQs sorted by some metric?

Perhaps by ROE/ROA or sales growth or cash balance …and equally weighted?

Thinking the most profitable/efficient of the group could be a interesting study

Has anyone ever run a backtest on top 25 or 50 of the QQQs sorted by some metric?

Yes, a number of times.
I haven’t found much success, in general.
If you want to create a quant screen, this is one of the worst hunting grounds.
Perhaps it’s a hunting ground that is driven by the actions of the hunters, and they’re all looking or an edge.
Too picked over, in other words. Maybe.

One strategy that’s very simple and not that bad:
Buy only the top 15 stocks ranked by rate of growth of sales per share in the last 5 years.
15 stocks, equally weighted, reconstitution of the portfolio every 2 months, would have beat the S&P 500 by about 2/3 of rolling years since Jan 2000.
Median and average rolling year improved by about 8-9%.

CAGR is the usual metric, but in the case of Nasdaq stocks it’s quite sensitive to the start and end dates chosen.
With that caveat in mind, Jan 2000 through the start of April 2022, CAGR 15.6% versus 7.3% for the S&P, not counting trading costs for the strategy.
Reconstitution every 2 months sounds like a lot, but the sales growth rates don’t change much, so the trading costs aren’t all that bad.
Not much turnover: average 1 or 2 position changes each 2 months.
I estimate trading costs at about 0.29%/year assuming round trip commissions and bid/ask gaps summing to 0.4% per position.

If you like harmless tweaking—
Before you look for the highest sales growth among the 100 stocks, first eliminate the 2 stocks furthest from their 52 week highs.
Skip the things that just fell off their popularity cliffs.
In backtest, this adds another 1.2%/year in annual return.
I call it harmless overtuning because it’s hard to see how it could make the strategy materially worse.
You’re only eliminating 2% of your candidates, affecting at the very most 13% of the portfolio selected.

Jim

7 Likes

Jim, thanks for sharing the historical earnings - they certainly produce an impressive R^2 value with a linear regression. Do you have data on how the equal weight set of N100 stocks performed during the tech crash 2000-2002? With tech seemingly being the biggest bubble sector this time around again, it makes me wonder how badly QQQE might get pulled down with the crowd if QQQ continues to crater. I wonder if the outperformance of the equal weight index vs. the cap weighted in the 2000-2002 time period could be instructive for expectation setting this time around in the event of another tech crash.

Do you have data on how the equal weight set of N100 stocks performed during the tech crash 2000-2002?

Badly!
Both in terms of earnings, and even more in terms of stock returns.
That’s why they called it the tech crash : )
The real total return series for the Nasdaq 100 equal weight peaked 2000-03-27.
The subsequent low was 2002-10-07, real total return loss -81.1%.
There wasn’t a fresh all-time-high real total return level until 2014-11-24.
That tech bubble was a doozy.

But note, that’s mainly because the prices were so extraordinarily high at the top, not because there was no hope of earnings or value.
The median trailing earnings yield was under 1% for quite a while, and bottomed at 0.565%.
Compare that to the median Nas100 earnings yield value at yesterday’s close of 3.154%.
The Nasdaq companies are relatively pricey lately, but a lot cheaper than they were in the tech bubble.

Of course, earnings are cyclical. A better measure would be the trend earnings yield.
Smooth out the earnings squiggles, then relate the on-trend real earnings to the then-current price.
On that metric, the most expensive day during the tech bubble was a median trend earnings yield of 1.3%.
And recent figures have represented a trend earnings yield of around 3.2%. (last month–sorry I don’t have today’s figure)
So, for the Nasdaq 100 firms on that metric, the tech bubble peak was 2.4 times as expensive as it has been recently.

But the interesting thing is that, after the horrible tech wipe-out, the earnings came (nearly) all the way back to the prior trend.
Imagine you were a super-bullish 1990s optimist, and simply took the real earnings level from (say) Q3 1997 through early 2000, and extrapolated it 22 years into the future.
How far off would you have been by now?
A lot…but not nearly as much as you’d think.
The pre-crash tech bubble trend would have led you to believe earnings today would have been 43% higher today than they actually are.
Sounds like a big error, but a forecast based on a short bubble stretch that’s off by only 1.69%/year 22 years later isn’t really that bad.
The possible moral is that the earnings of this crowd, other than transient dips during bears, trend pretty reliably. (And rise quickly)
The prices paid for those firms swing wildly, but their earning ability very much less so.
This gives hope for a strategy: keep an eye on the trend earnings, and the trend earnings yield.
Figure out the likely forward earnings based on the assumption that earnings will rise on trend at a rate comparable to history.
And make the guess that the valuation multiple on those trend earnings will be “typical” in a few years.

Jim

13 Likes

This gives hope for a strategy: keep an eye on the trend earnings, and the trend earnings yield. Figure out the likely forward earnings based on the assumption that earnings will rise on trend at a rate comparable to history.

This post has given great food for thought - thank you. A follow-up question:

The benefit of looking at earnings trend data seems to be that it removes transient/cyclical squiggles and reveals the longer term trajectory, and if that trend is very stable and consistent, we’d expect a higher probability of it continuing.

But it seems like the big challenge in looking at trend data is discerning (in a timely fashion) when the longer term trend has been broken and the past trend data can no longer be relied upon as a predictor for the future.

Have you found effective techniques for timely identification of when the earnings trend has “gone off the rails” and differentiating that from transient/cyclical effects (the squiggles we’d like to ignore)?

1 Like

Have you found effective techniques for timely identification of when the earnings trend has
“gone off the rails” and differentiating that from transient/cyclical effects (the squiggles we’d like to ignore)?

One can only guess.
But since in this case it is the average among 100 companies, there would only be a lasting break in up or down based on some fairly major event.
That’s because no company-specific issue could move it a lot.
Arguably the TCJA tax cuts caused a one-time shift upwards in the earnings level. Which might or might not last.
A small recession should be a blip, but a really major one or a depression could reasonably cause a permanent downwards shift or even a slope change.

The biggest risk is probably this:
Any change in the selection criteria (both formal and conventional) for Nasdaq firms would change the happy situation.
There is no particular reason for the fastest growing firms to show up in this group, year in and year out forever.
The usual rule of thumb is that all broad groups of equities perform the same over long time periods.
If small caps or companies starting with N were consistent outperformers, more people would buy them.
Their price would get bid up to the point that new buyers would get no extra return on their money.
I’m a little startled that the earnings growth of the Nasdaq 100 bunch has been so oddly fast for so long.
Earnings have risen a whole lot faster than GDP, which is not normal in the long run.

With long run earnings yields, and the cost of the equities, I can calculate the median earnings figure among the set.
After inflation adjustment, that was up by a factor of 4.203 between 1997-09-02 and 2022-04-29.
That’s earnings growth of inflation + 6.0%/year for 24.66 years.
Pretty good.

FWIW, median real earnings among the Nas100 are about 10% below the long run trend at the moment.
I consider that size of squiggle to be pretty normal.
There was a dip in 2016, a brief one in 2018, a longer lasting one in 2020, and so forth.

Jim

4 Likes

That’s earnings growth of inflation + 6.0%/year for 24.66 years.
Pretty good.

FWIW, US real GDP rose inflation + 2.05%/year in the 23 years 1997 through 2020.
That shows how remarkable a growth figure of 6% is.
And, being an equally weighted index, it isn’t because of the extraordinary success of a few firms.

Jim

3 Likes

Jim,

Is Nasdaq 100 earnings data publicly available?

For S&P 500 earnings, below website has a spreadsheet anyone can freely download. Anything similar for Nasdaq? If not, which data provider do you subscribe to get your information.

https://www.spglobal.com/spdji/en/indices/equity/sp-500/#ove…

Thanks in advance.

Zz.

1 Like

Is Nasdaq 100 earnings data publicly available?

Yes and no.
It can be gleaned from the GTR1 backtester used by the MI board.
It’s not at all user friendly, but very powerful and backed up by a lot of rigorous data sources.
One of the data fields is membership in the Nasdaq 100.

Note, the “regular” Nasdaq 100 is not particularly predictable by comparison.
It is very concentrated, and is not even cap weight, only “sort of” cap weight.
So its earnings and performance rely very much on the vagaries of just a few firms which are not predictable in macro terms.
My comments are all about the much more obscure Nasdaq 100 Equal Weight index.
It’s not too hard to find historical earnings data for the “regular” Nasdaq 100 index, but IMO it doesn’t do you much good.

Jim

3 Likes