Re: the one you call price-book-high

From this thread
https://boards.fool.com/i-believe-it-was-called-39mound-of-t...

Lots of variants discussed, but I like best the one in that post, with a bit of a momentum final sort added:
VL stock
Price-to-book highest 165 as in that post from the mound of toast peak
5-year sales growth top 15
Price/52-week-high top 10 (remembering that 52-week high in the VL database is lagged)

Every two months with friction from 1997, 24.9 years:
CAGR 25.0%, beating S&P by 15.4%/year
Only 74% of the risk of the S&P using DDD3 risk metric.
Beat the S&P 9 of the last 10 calendar years, and 84% of the rolling years in the backtest.

November picks include Netflix and Amazon.

Feeling lucky?
Just stick with the top 2 picks and it comes out at 38.9% CAGR, another 13.8%/year!
98% of rolling years positive.
Not for the faint of heart, but maybe add a tiny pinch of those two stocks to an otherwise boring portfolio?
A mix of 90% S&P and 10% these two stocks (5% each) beat SPY in backtest by 3.3%/year with only 3/4 the DDD3 risk of SPY.
No, I'm not serious about suggesting you try this with real money. Two stock backtests are not meaningful.

Jim

BlueCheaps?

BlueCheaps?

Ah. That one.
That is indeed one of my suggestions, but it has not turned out to have been a good one.
Its backtest was up to end 2009.
Backtest era 1986-2009: beat the S&P by 5.9%/year.
Since then: lagged the S&P by -6.9%/year.
Nice safe firms, and it has made you some money, but it really wasn't a useful strategy to have followed.
The market doesn't care about earnings in the modern era, it seems.
Or, to phrase it the fashionable way, it has been a tough time for value stocks.
Let's blame it on that rather than an examples of unsuccessful data mining, shall we?

Interesting side note: the original thread was entitled "A screen for skeptics".
It would appear any skeptics would have been entirely justified in their skepticism.
Now the screen is mostly useful as a memento mori way to keep me humble.

For those wanting something with big solid firms, the LargeCapCash screen and its variants look better to me.
But of course it doesn't have meaningful out-of-sample validation yet. Maybe it will ultimately turn out to have been a mirage too.
Early signs are OK, just not much of it yet.
19 months out of sample for the recommended versions, trade every two months, 40 stocks deep hold-till-drop at 45.
Regular version: CAGR 50.3%, beating S&P by 11.6%/yr rate.
Dividend version: CAGR 51.8%, beating S&P by 13.1%/yr rate.
But 19 months isn't exactly a lot--mostly just one rally.

Those figures are using the "consensus" VL versions later in the thread.
Regular version:
VL stocks with Timeliness positive
ROE latest quarter top 26% (about 364 stocks)
(Cash - Long term debt) top 40

Dividend version:
VL stocks
Dividend and Timeliness positive
ROE latest quarter top 32% (about 263 stocks)
(Cash - Long term debt) top 40

Jim

The reason I brought it up (original screen post is 222475) absolutely doesn't relate to .. way to keep me humble.
Rather, it relates to the discussion on this board recently of many screens failing post-discovery, and to my hobby horse that closely examining very unexpected failures (as uncomfortable as that can be) can be sometimes be very illuminating.
But, most of all, it's a screen you can believe in.
Yikes, something went wrong!
Is it clear what went wrong with the screen (repeated below), and in retrospect was there something that was missed then that could help us up the chances that a newly created screen won't fail post-discovery?


Not everybody believes that mechanical investing works. "Ah, just a lot of navel-gazing, digging out coincidences in the old data again and again", they say.

So, here is a stock-screen based investment approach intended to be acceptable even to the skeptical and hardened value investor.
To that end I've tried to avoid any board-specific jargon in this post!

I think almost anybody would agree that buying companies with lots of earnings (low P/E's) is probably a good idea, and especially so if they have strong balance sheets.
To address the problem with the idea that small cap companies are too flighty and unpredictable, we'll further limit it to only the big ones.

No momentum, no proprietary "Timeliness" ranking, nothing fancy.
Just strong, big companies with low P/E ratios, period.


So, here's the general idea:
- Start with the stocks rated in A++ or A+ for Financial Strength by Value Line (there are about 120 of them, usually).
- Of those, consider only the 75 biggest ones by market cap.
- Of those 75, buy equal dollar amounts of the few (anywhere from 3 to 15) that have the highest current earnings yields (lowest P/E ratios).
- Hold them all for a while (a fixed number of months, 1 or more).
- When the hold period is over, find the new picks, sell anything no longer top of the list, and buy the new top picks.
backtest.org/StrongBigCheap:s.5SBOOfstEQa++QCAOfstEQa+QCCMOp......
<\i>
...

One nice thing about this screen is that it is very robust.
It beats the market with 3 stocks deep or 20 stocks deep.
It beats the S&P if you trade every 1, 2, 3, 4 ,6, 8, 9, or 12 months, though the best returns are with the shorter holds even after trading costs.
It works with "hold till drop" (e.g., buy the top 5 stocks and check every month, but don't sell anything till it drops below rank 12), and
it works with "dozens" (e.g., each 2 months buy the 3 topmost stocks you don't currently hold, and hold them for two months).

The reason I brought it up (original screen post is 222475) absolutely doesn't relate to .. way to keep me humble.
Rather, it relates to the discussion on this board recently of many screens failing post-discovery...

Oh, I didn't think your post was to keep me humble. I think the bad results of things like BlueCheaps keeps me humble.

As for why it went wrong, yes, a post mortem is always good.

Before looking at why it went wrong, first there is the consideration when it went wrong:
2010 adn 2011 were basically market tracking.
It started diverging from what was "expected" around the start of 2012.

The mini universe it's looking at is the 75 biggest firms with strong financials.
I think it's fair to say that the screen was designed starting with the notion that this working set wouldn't diverge a huge amount from the cap weight index over time.
I would have expected a pinch worse, because you tend to pay a bit of a premium for an extra secure stock (better balance sheet).
But basically, it was intended as a sane baseline working set, not a step intended to give part of the outperformance.
Since 2012 this set of 75, equally weighted has underperformed the S&P 500 by 2.0%/year.
That's a bit more than the lag one might expect, but not soul crushingly different.
We all know super big caps have done best (historically a rare occurrence), so the cap weight of teh S&P is probably a part of that too.
So, offhand, I'd say this isn't really where the problem lay.

The more interesting thing is how the screen did within that 75 stock mini-universe.
That 75 stock group, prior to the "low P/E" sort, returned 14.2%/year in the 9.9 years since Jan 2012.
Since we're thinking of that as our starting mini universe, consider that return our benchmark.
The screen (simply the lowest P/E among those), returned 7.1%/year held 15 deep.
The inverse of the screen (the highest P/E among them), returned 16.8%/year held 15 deep.
So, in essence: among the biggest wonderful-balance-sheet firms, not only did the most basic possible "value investing" not work, but it was backwards.
Emphatically so.
Had the market merely not cared about earnings one way or the other, you'd expect all three figures to be about the same, from random chance within the group of 75.
But no: it appears that the market was actively seeking the firms with the lowest earnings, not merely neutral on the subject.
And even more actively avoiding those with high earnings.

This is, needless to say, an unusual situation compared to millennia of history.
We then have to ponder whether that's a cyclical or a permanent change.
If this situation ends some day, the screen can be seen as something that really didn't work for a
substantial stretch because it didn't foresee an odd stretch in the market.
But, by extension, might one day start working.
Conversely if the rejection of earnings is the new permanent normal, it simply means it was a screen
built with hindsight, unsuited to the new regime which was about to start.
Like somebody who looked at decades of great results from Kodak just before (other companies') digital cameras ate their lunch.
When the world changes, as it occasionally does, hindsight is no longer a useful guide to the future.

Who knows what will happen in the next decade, and thereafter? Beats me.
Personally, I find it hard to believe that earnings will never again matter.
After all, pretty much every company folds some day, or is bought out by a firm that ultimately folds.
You get only so much in the way of distributions before then: dividends, spinoffs, maybe a final payout.
Those have to be discounted back to the present to estimate some kind of present value. A buck today is worth more than a buck 100 years from now, even adjusting for inflation.
Those distributions generally come from profits. (sometimes a little from starting assets) No profits, no distributions.
And, by extension, with not profits, by the time the stock price is zero stockholders in aggregate will have lost money.
Meaning all returns between now and then are a negative sum game for shorter hold periods, with more losers than winners.
So...if a stock is to have positive returns over time, even for a finite time, you'd think some earnings would be needed.
Patience is fine, maybe more patience is better, but ultimately you have to eat.
If there is never any prospect of making a buck, a company is nothing but somebody's expensive hobby.
This line of reasoning does not seem to be believed by the market this past decade.

Though I think earnings will matter eventually for the market (as they certainly will, eventually, for each individual company), that doesn't mean this particular screen will ever add value.
But and end to the market actively avoiding earnings would certainly help its chances : )

Jim

The screen (simply the lowest P/E among those), returned 7.1%/year held 15 deep.
The inverse of the screen (the highest P/E among them), returned 16.8%/year held 15 deep.
So, in essence: among the biggest wonderful-balance-sheet firms, not only did the most basic possible "value investing" not work, but it was backwards.


Thanks Jim!
- Would the inverse sort done at the time of creation of the screen using in-sample data show any hint?
- And/or, if one kept doing this sanity check post-discovery, would a trend be apparent suggesting that things were heading south?

Earlier in this thread you suggested
For those wanting something with big solid firms, the LargeCapCash screen and its variants look better to me.

Here's yours original LargeCapCash screen:
Start from the Value Line 1700-- this is old skool!
Of those with a reported ROE, take the top 30%-- around 475 companines on avreage.
For each one, calculate their cash balance in excess of long term debt.
Buy equal dollar mounts of the 40 stocks with the largest net cash balance.
That's it.

Similar questions about the above:
- does it break in-sample if take the lowest 30% of ROE (or some other reasonable fraction)?
- does it break in-sample if buy equal dollar mounts of the 40 stocks with the smallest net cash balance?

It doesn't need to actually break i.e. shatter into pieces.
But if doing the opposite of what one is thinking doesn't make a significant difference in (or even improves!) results,
then this is a strong hint that what one is thinking is something one shouldn't be thinking.

Here's an odd thought for the day about evaluating screens post-discovery:
An issue is that if a screen starts to fail post discovery it can be hard to tell if it's a significant failure, until it's too late.
Perhaps "broken screens" are more sensitive indicators of an evolving mismatch between screen and market conditions than simply observing the "original screen" post-discovery?
More precisely, for BlueCheaps and other screens with more post-discovery data than LargeCapCash:
is observing the behavior of the broken screen (reverse sorts etc) a more sensitive indicator of possible trouble than observing the original screen post-discovery?
Perhaps the original screen kinda sorta starts to fail post-discovery, but the broken screen quite clearly is doing something different post-discovery than it did on in-sample.
IOW, can the broken screen be a more sensitive canary in the coal mine?

- Would the inverse sort done at the time of creation of the screen using in-sample data show any hint?

Yes. A closer examination would have shown have shown a long term overall advantage for low P/E, but with a bout of wildly backwards results.
Expensive stocks led in the tech bull, much as it has lately in similarly exuberant market conditions.

Jan 86-Jan 98: Advantage low P/E 15, gap 4.6%/year
Jan 98-Mar 00: Advantage high P/E 15, gap 39.5%/year [short interval but big rate gap]!
Mar 00-Aug 07: Advantage low P/E 15, gap 26.2%/year
Aug 07-Nov 21: Advantage high P/E 15, gap 7.5%/year

The screen was formulated not that long after part 3 ended, so part 4 corresponds very roughly to the post discovery period.

So, part 2 was our warning flag.
We ignored it because it was only 10% of the interval covered by the first three sections, so it seemed like an unlikely-to-be-repeated anomaly.
But we got it again, and it has lasted longer. A bigger absolute difference, though it hasn't been as extreme on an annualized basis.

The first section is pretty mild.
In effect, the whole of the advantage was section 3, 2000-2007, a long stretch with a big advantage for earnings.

One might speculate that this is a pretty good indicator (and tool for playing) the value fashion cycle.
And that it will swing the other way when the current trend is over.
(or, looked at the other way, the current trend will be over when it swings the other way)
At this point, profitless firms have been so popular for so long that cyclicality is now just a speculation.

if one kept doing this sanity check post-discovery, would a trend be apparent suggesting that things were heading south?

A definite maybe.
These were pretty long, strong, steady trend sections. Only three major changes of direction in 27 years.
The advantage of the low PE 1/5 over the high P/E 15 was particularly steady 2000-2007.
The biggest counter-trend move was about 14 months when low P/E stocks did pretty well, starting around Jan 2016.
Non annualized, up 26% while the backwards screen was up 3%.
So, it's quite possible that one can spot a regime change soon enough to ride it to its end.
When (well, if) the tide turns, jump on old fashioned value and ride it?
This screen might not be such a bad way to do it, though surely there are others.

Seen in this light you could recategorize this as a case of having mistaken the cycle for the trend.
And if you're being generous, think of the post discovery flop as just one stretch of what it looks like in the part of the cycle that it doesn't work.
Because one can still speculate that it will work pretty well some day. For a while. Maybe.

Jim

Here's yours original LargeCapCash screen:...
Similar questions about the above:
- does it break in-sample if take the lowest 30% of ROE (or some other reasonable fraction)?
- does it break in-sample if buy equal dollar mounts of the 40 stocks with the smallest net cash balance?


Bad ROE followed by nothing is cyclically wild, long term net either better or worse depending when you stop the test.

Bad ROE followed by good cash as in the original screen is almost exactly a market tracker.
(loses a bit more in a bear, gains a bit more in a bull, net result a tracker, and relatively smooth)

Bad ROE followed by bad cash is wild.
Also about the same overall return as SPY from 2000, but wild cyclical beta.
Huge underperformance in bears, huge outperformance in some bulls (02-07, 09-14)

I have a screen intended to be a short leg to pair with LargeCapCash.
Three steps: Poor ROE, poor cash-debt balance, low sales growth.
Long run it underperforms SPY by only 6%/year, but more usefully it underperforms relative to LargeCapCash over most short intervals.
I don't run it--I'm letting the pair "age" for a while before considering it with real money.

Jim

Regarding BlueCheaps:

So, part 2 was our warning flag. We ignored it because it was only 10% of the interval covered by the first three sections, so it seemed like an unlikely-to-be-repeated anomaly.

The flag may not have been waving all that long, but two years of a huge gap could get your attention.
One might speculate that this is a pretty good indicator (and tool for playing) the value fashion cycle. And that it will swing the other way when the current trend is over.
Yep.

Regarding LargeCapCash
Bad ROE followed by good cash as in the original screen is almost exactly a market tracker.
This is a partially broken screen from reversing just the ROE sort: the performance didn't get shattered but it got bent i.e. became a 'market tracker'.
Might make one suspicious?
Bad ROE followed by bad cash is wild. Also about the same overall return as SPY from 2000, but wild cyclical beta.
Huge underperformance in bears, huge outperformance in some bulls (02-07, 09-14)
This totally broken screen's performance (all sorts reversed) also didn't shatter, but it seems to get bent:
could be suspicious that it ended up with similar return to SPY (albeit with wild swings) when presumably it should have ended up as an absolute dog.

I'd love to see simple screens that do well in backtest, and when you break them by reversing the sorts, do in fact howl like dogs.
More generally, the lesson may be that separating a screen into its pieces and seeing how all the gears interact, can be useful.
This is another sort of 'robustness test', previously we listed robustness to changing parameters, robustness to noising up the data, etc.

Regarding Long/Short(reversed long):
I have a screen intended to be a short leg to pair with LargeCapCash.
Three steps: Poor ROE, poor cash-debt balance, low sales growth.
Long run it underperforms SPY by only 6%/year, but more usefully it underperforms relative to LargeCapCash over most short intervals.
I don't run it--I'm letting the pair "age" for a while before considering it with real money.

So it's basically the reversed screen with an added low sales growth filter. I assume it did as expected in backtest.

I'm liking the long/short idea for screens where the short is in large part a reverse of the long screen.
Seems like that should be robust.

Thank you, Jim!

Perhaps the issue with low P/E in BlueCheaps is that it doesn't account for future growth or lack thereof. As you (Jim) often suggest, current price relative to forecast earnings 5 to 10 years from now is probably a much better measure of value and can work for low or high growth companies. Obviously, there are various ways to try and estimate the future earnings whether that involves past or estimated measures of growth (sales or earnings) or something else.

StevnFool

One might speculate that this is a pretty good indicator (and tool for playing) the value fashion cycle.
And that it will swing the other way when the current trend is over.
(or, looked at the other way, the current trend will be over when it swings the other way)

So maybe something like a monthly:
- evaluate BlueCheaps and SPY over last month
- if BlueCheaps better, buy/hold for next month
- else buy/hold SPY if above sma 200 (or up in last month, or ...)
- else cash
?

In the etf/fund area, are there any value funds that we think actually use reasonable criteria? Could we just compare one of them against SPY?

SA

So maybe something like a monthly:
- evaluate BlueCheaps and SPY over last month
- if BlueCheaps better, buy/hold for next month
- else buy/hold SPY if above sma 200 (or up in last month, or ...)
- else cash

I think I'd compare it to a growth type universe that doesn't care about earnings, rather than SPY.
But yes, it's possible there is some predictive power in its recent relative results that could last long enough to be useful.

Long ago I made a timing indicator that simply compares the recent performance of all dividend versus all no-dividend stocks.
Not unlike Zee's much fancier speculative/defensive models.
It works pretty well, though on a much shorter time scale than what we're seeing here.
I expected it to work as an indicator of what sort of thing to invest in, but it ends up
being not a bad predictor of the general market, so it worked better as a long/cash signal.
Lots of trading, though, since the signals change pretty often.

Jim

In the etf/fund area, are there any value funds that we think actually use reasonable criteria?

This discussion is looking for a canary, not for direct investment. Most Value funds use PB. Those that don't get classified as Large Blend. So, to find an ETF using reasonable criteria, we may have to look in the Large Blend category.

As as starting point, consider {BlueCheaps}. It looks for very high Financial Strength (about 120 stocks, usually), then 75 largest market cap, then 15 highest current earnings yield. Those sound reasonable to me. As others have suggested, I might add some growth. Even Graham included growth (anything above zero) in his value metrics.

A somewhat similar ETF, First Trust Capital Strength ETF ($FTCS) uses strength of balance sheet, cash balances, long-term debt ratios, and ROE. This ETF has 15 years history and $9B AUM. Performance has matched the market with a 15-year CAGR 10.8 vs SPY CAGR 10.6. Classified as a Large Blend.

The largest Value ETFs all use PB and none look for positive growth. Some look for low growth.

                                           AUM   uses  uses  uses
Ticker                Name                 [$B]   PB     E     G   notes
  VTV           Vanguard Value ETF          88   yes   yes    no   P/B, forward P/E, historic P/E, dividend-to-price, sales-to-price
  IWD     iShares Russell 1000 Value ETF    56   yes    no    no   price/book , low  growth forecast, low  five-year sales per share growth
  VBR     Vanguard Small-Cap Value ETF      26   yes   yes    no   book to price, forward earnings to price, historic EPS, dividend-to-price, sales-to-price
  IVE       iShares S&P 500 Value ETF       23   yes   yes    no   book value/price, earnings/price,  sales/price
 VLUE   iShares MSCI USA Value Factor ETF   16   yes   yes    no   price/book, price/forward earnings,  enterprise value/operating cash flow - all sector adjusted
  IWN     iShares Russell 2000 Value ETF    16   yes    no    no   price/book , low  growth forecast, low  five-year sales per share growth
  EFV      iShares MSCI EAFE Value ETF      15   yes   yes    no   book/price, forward earnings/price,  dividend yield
  VOE       Vanguard Mid-Cap Value ETF      15   yes   yes    no   book to price, forward earnings to price, historic earnings to price, dividend-to-price, sales-to-price
  IWS   iShares Russell Mid-Cap Value ETF   14   yes    no    no   price/book , low  growth forecast, low  five-year sales per share growth
 SPYV   SPDR Portfolio S&P 500 Value ETF    13   yes   yes    no   book value/price, earnings/price, and sales/price
 DFAT   Dimensional US Targeted Value ETF    7   yes   yes    no   P/B, profitability

Another possiblitiy is Parnassus Core Equity Investor ($PRBLX). The prospectus says "Upon initial investment, a company’s stock must be trading below its intrinsic value". That sounds like Value to me. Classified as Large Blend, with a 15-year CAGR 12.3. This fund uses a value oriented investment process, picking stocks that pay dividends, with capital appreciation potential, and the capacity to raise dividends. This fund's focus on dividends is more than I would like.

Is it possible that the value-trap character of P/B (and, to a lesser extent, dividends) has fueled the emergence of Quality as a distinctive factor among fund designers. QUAL selection criteria are "high ROE, stable earnings growth and low debt/equity, relative to peers in each sector" (ETF.com). Its MSCI Value factor rating is slightly negative, at least at the moment.

I have long looked around for a value fund to hold in the LTBH part of my portfolio, but I have come up empty. My large cap LTBH holding is SPGP, which seeks "growth at reasonable prices."

SPGP has a brand mate, SPVM, which defines value in terms of P/B, P/E, and P/S, but then includes momentum as a selection criteria. Weighting is based on the value score, however.

Baltassar

Maybe try GVAL or a US-ian version of it? Flat the last 6 months, but up 12% YTD (tapering off). Interesting, specific criteria:
The Cambria Global Value ETF identifies the cheapest stock markets in the world using a composite of cyclically adjusted price ratios. The ETF begins with a universe of 45 countries, located in developed and emerging markets. GVAL then selects the top 25% cheapest country stock markets as measured by Cambria’s proprietary long-term valuation metrics based on relative and absolute valuation. Cambria then uses a valuation composite across traditional metrics such as trailing P/E, P/B, P/S, P/FCF, and EV/EBITDA to select the 10 most undervalued stocks out of the top 30 largest stocks by market capitalization within each country.

This discussion is looking for a canary, not for direct investment.

Yes, I understand that. I'm wondering if there's a canary I can just look up on Yahoo every so often (by comparing with SPY or QQQ or whatever), rather than needing to look at a screen in gtr1 or something. I could look at some value vs growth funds / indexes, but the general opinion is that these are constructed poorly.

SA

SPGP has a brand mate, SPVM, which defines value in terms of P/B, P/E, and P/S, but then includes momentum as a selection criteria. Weighting is based on the value score, however.

Ok, so that uses p/b and p/s, but at least it's a blend of factors and does include momentum (well, not sure we want that?). Anyway for a simple visual impression of value vs growth in the market we can look at spvm:qqq, e.g. https://schrts.co/tyTfzyTG

SA

Jim wrote:
Lots of variants discussed, but I like best the one in that post, with a bit of a momentum final sort added:
VL stock
Price-to-book highest 165 as in that post from the mound of toast peak
5-year sales growth top 15
Price/52-week-high top 10 (remembering that 52-week high in the VL database is lagged).

Should the Bank and/or Investment Bank Industries be filtered out of these results, as I've seen it mentioned the concept of sales doesn't apply?

Should the Bank and/or Investment Bank Industries be filtered out of these results, as I've seen it mentioned the concept of sales doesn't apply?

This is a VL screen, so in that context it is automatic...no bank has the 5 year sales growth field populated.

If you want to construct an analogous screen in a different universe, then it's a different screen and you can do what you like : )
But yes, to match as closely as posssible, you'd have to screen out banks.
There will inevitably be other differences.

Jim

1 month RS switch

The simplicity of this screen is so intriguing. Could you share how it backtests

This is a VL screen, so in that context it is automatic...no bank has the 5 year sales growth field populated.

I'm using Value Line Stock Screener, which I have access to from my public library...not sure if its the same as your data from Value Line.

On the Stock Screener I filter on:
1) Timeliness Ranks (1-5), which returns 1,507 Tickers
2) % Price to Book value, limit results to 165 with lowest values by selecting range from 1.88% - 95.0%
3) Sales Growth 5 Year (% Relative to Market 80-100%) results in 18 Companies

Here are the results, Exported (ordering on Growth Rate desc).

Company Name			Ticker	Timeliness™	Price to Book	Sales 5 Yr Growth Rate	Industry
Golar LNG Limited		GLNG	3		66.61%		99.00%			Maritime
Diamondback Energy		FANG	3		80.22%		96.00%			Petroleum (Producing)
United Natural Foods		UNFI	5		52.84%		95.00%			Retail/Wholesale Food
Gray Television			GTN	3		60.69%		95.00%			Entertainment
MarineMax Inc			HZO	2		91.69%		93.00%			Retail (Hardlines)	
Molson Coors Beverage		TAP	4		73.65%		93.00%			Beverage
Beacon Roofing Supply Inc	BECN	2		90.78%		92.00%			Building Materials
TRI Pointe Homes		TPH	2		83.81%		92.00%			Homebuilding
Lincoln National Capita VI	LNC	4		34.00%		92.00%			Insurance (Life)
Antero Resources		AR	1		13.61%		91.00%			Natural Gas (Div.)
Taylor Morrison Home Corporat	TMHC	3		79.62%		89.00%			Homebuilding
State Street Corporation	STT	3		88.47%		85.00%			Bank
Principal Financial Group Inc	PFG	5		71.00%		85.00%			Financial Svcs. (Div.)
Qurate Retail			QRTEA	2		73.21%		80.00%			Retail (Hardlines)
Signet Jewelers Ltd		SIG	4		53.03%		80.00%			Retail (Hardlines)
Everest Re Group Ltd		RE	4		91.00%		80.00%			Reinsurance
DCP Midstream LP		DCP	4		47.88%		80.00%			Pipeline MLPs:
Goldman Sachs Group Inc		GS	2		74.14%		80.00%			Investment Banking

The Industry is from Value Line data. I was referring to GS and STT.

Thanks,

1 month RS switch

The simplicity of this screen is so intriguing. Could you share how it backtests

I wondered the same thing... here's what I get if trading at the end of each month (and no friction):

           EOM_RS_Switch   SPY
2006-12-29          12.2%  13.1%
2007-12-31          14.9    5.1
2008-12-31         -41.9  -36.8
2009-12-31          67.0   26.4
2010-12-31          14.6   15.1
2011-12-30           3.9    1.9
2012-12-31          16.7   16.0
2013-12-31          42.5   32.3
2014-12-31          17.4   13.5
2015-12-31           3.6    1.2
2016-12-30           9.9   12.0
2017-12-29          21.9   21.7
2018-12-31          -6.8   -4.6
2019-12-31          31.4   31.2
2020-12-31          50.4   18.3
2021-11-26          20.8   24.0

During this time, SPY has a CAGR of 10.6% and a GSD of 15.0%
while the RS_Switch has a CAGR of 14.9% and a GSD of 16.8%.

heink

while the RS_Switch has a CAGR of 14.9%

That's ... interesting.

I re-ran my backtest for that screen, although my dates are a bit different. 3/2006 thru 10/2021. And I also got 14.9% CAGR.

Not sure about the figures you showed for 2008/2009, -41.9% and +67.0%
I get -6.2% & 44.0%

I showed being in cash 6 months in 2008 & 3 months in 2009.

Here's the counts of # of months for each:

SPY   :    7  ( 4%)
QQQ   :   91  (48%)
PRF   :   46  (24%)
IWF   :   21  (11%)
cash  :   24  (13%)

Now if we could somehow figure a way to avoid the negative years

Ray. So if all four are negative you are in cash ?

I'm using Value Line Stock Screener, which I have access to from my public library...not sure if its the same as your data from Value Line.

No, it sounds right.
There's a difference (at least in their eyes) between a bank and an investment bank.
Goldman Sachs shows a sales growth figure, Bank of America doesn't.

I looked at firms that were definitely 5 years old, by checking for a value in the 5 year total return field.
These are their industries with no firms with 5-year sales growth data populated:
BANK
BANKMID
CANNABIS
INSLIFE
INSPRPTY
INVEST2
INVEST6
REINSUR
THRIFT

I think Cannabis is a quirk...there are only two, and maybe they're old enough but had zero revenue.

The category FINSERV has sales growth figures for about half firms that are old enough.

Maybe you can figure out the distinction.
These FINSERV companies have sales growth figures:
FIS AON GPN WU HRB AJG L LAZ MMC SC BRO EEFT FCFS WEX MA AER FLT V CACC

...and these don't:
AXP AIG AIZ COF CIT DFS FNF HIG KMPR MTG NAVI PYPL PFG SLM SLF.TO SYF VIRT WLTW

Since I'm not filtering by industry, only on sales growth, the first group would be eligible picks in my backtest and the second group wouldn't.

Jim

Maybe you can figure out the distinction.

I also see a drop (187 -> 100) when I add the Sales Growth 5 YR filter. But, the only tickers from the above list that don't show up for me are SYF and WLTW. In any case, what I was most interested in is verifying the data I see when accessing from the library interface matches what you get from your subscription...

Also, I noticed my first sort was wrong. The firms that passed my filter were ones with lowest P/B. Your rule is to start with firms with the highest P/B.

Thanks again for all your insights!

-Pete

Your rule is to start with firms with the highest P/B.

Yup...counter-intuitive, isn't it?

A high price-to-book is usually seen as a sign of an overvaluation by academics, but that's only for a boring middle-of-the-pack company with no competitive advantage or special pricing power.
Who wants to invest in those?
The few really best businesses need almost no assets to generate their earnings, so they tend to be the very highest P/B firms in a sort.
So if you're taking just a small number of firms, you want high P/B, not low.
Incidentally, this is one reason so-called "value" indexes have done badly, as so many of them use low P/B as one of their main selection criteria.
That makes sure they are eliminating the best businesses, which might be conventionally cheap on other metrics like forward P/E.

Some great firms even have an effectively infinite P/B because they have no positive book value.
It's in a good way: positive earnings with negative book value is usually good.

Jim

So if all four are negative you are in cash ?

Yes.

Here's what I get for 2008, Date is the pick for *that* month, based on the numbers of the previous month.

12/1/2007       IWF
1/1/2008        cash*
2/1/2008        cash
3/1/2008        QQQ *
4/1/2008        QQQ
5/1/2008        QQQ
6/1/2008        cash*
7/1/2008        QQQ *
8/1/2008        PRF *
9/1/2008        cash*
10/1/2008       cash
11/1/2008       cash
12/1/2008       QQQ *
1/1/2009        cash*
2/1/2009        cash
3/1/2009        QQQ *

Here is a link to the PV backtest. For some reason, they get lower returns than I get.
https://www.portfoliovisualizer.com/test-market-timing-model...

Remove CASHX for never going to cash.

PV gets for # of months. (What I get in parens):
cash: 49 (24)
QQQ: 80 (91)
IWF: 12 (21)
PRF: 40 (46)
SPY: 8 (7)