No. of Recommendations: 10
musselmant,
http://boards.fool.com/better-timing-rule-ma200-plus-economi...,
brought GTT timing to our attention.

The author of the post there,
http://www.philosophicaleconomics.com/2016/01/gtt/
suggested a timing system as follows:

first check for signals suggesting a recession;

if no recession is on the horizon, then stay in the market with no use of timing methods based on trend following or price momentum;

when some some evidence of economic weakness or a recession appears, then use regular timing signals, like 200 SMA type signals or others you prefer.

The idea was that something like a 200 day SMA timing method whipsaws too much in a good economy to be profitable, making timing under perform, so save the timing for recessions, where the exit/entrance signals are usually further apart allowing timing to be profitable and limit drawdowns.

He suggested two of the 7-8 signals at the Fed could be used, or all, if you prefer. The signals one has to use are always delayed at least one month, but if you'd like to try them here they are:

https://fred.stlouisfed.org/graph/?g=3asR

This is back on Nov. 1, but is high enough that I doubt it's showing a recession by Dec 1, so it says all is fine with the economy. No timing needed.

The next is not so good

https://fred.stlouisfed.org/graph/?g=3atC

but is close to zero and with the 'Trump bounce' has likely turned positive by Dec 1.

With this signal showing weakness the GTT scheme tells us to use timing for the next month, December.

Two of our signals are related to regular momentum strategies and/or trend following, the SMA is a trend follower and NH/NL probably attempts to predict near term changes in momentum, so I'd say it was a (non-classical) momentum strategy. Entrances in DBE are momentum based, but
exits are based on estimates of the time it takes momentum to disappear, so I won't place it as a momentum strategy of the type
the GTT author used. It's some kind of a hybrid.

rrjjgg
Print the post Back To Top
No. of Recommendations: 11
I've backtested this technique. The rules are a bit subtle (or I'm just slow) and I had to re-read it a bunch of times to get it.
Basically:
1) Buy signal is close above the SMA.
2) Sell signal is close below the SMA *unless* any of the Fed signals are positive. That is, only if *all* of them are negative.

Restated, you only take the sell signal if we are in a recession. If not in a recession, buy/hold and don't sell, no matter what the SMA (or NH/NL or whatever) signal says. Make logical sense.

In followup posts/comments, he said that further study indicated to him that the unemployment (employment?) signal was all that was needed, and was the most useful.
FWIW, I decline to look at that, because there is too much political shenanigans with un/employment figures. They lie, like Red China does, for political reasons.


The signals one has to use are always delayed at least one month,
Which does not matter. True tops are usually rounded and you've usually got plenty of time to get out before the market rolls over.

Remember .. the purpose of timing is not to increase your return, it's to get you out before the full depths of a bear market. Other than these bad bear markets, you want to be always in the market. Using these Fed recession signals to negate a SMA sell signal accomplishes that. My backtests increased the time-in-market from ~75% to ~85%.
Print the post Back To Top
No. of Recommendations: 1
Both Fed recession signals mentioned in the first post were updated in mid-January.
The first remained positive, the second switched to positive on Dec. 1st, as I guessed would happen. To update the graphs, if need be, click on 'max' in the date bar above the graphs.

So timing signals should be ignored for now in this system.

But the BCC signals are all happiness and delight, so who cares :-)

rrjjgg
Print the post Back To Top
No. of Recommendations: 3
Both Fed recession signals mentioned in the first GTT post were updated in mid-February. The graphs may not display the most recent updated data unless you
click on 'max' and, possibly, scroll over to the right on the interactive graph to read the value.

No regular timing methods need to be employed
this month.

This one is .005, so almost 0
https://fred.stlouisfed.org/graph/?g=3atC

https://fred.stlouisfed.org/graph/?g=3asR

rrjjgg
Print the post Back To Top
No. of Recommendations: 1
In case anybody cares....
I have a script that every day grabs a couple of these indicators from FRED, and keeps track of each time it changes. Looking at them you can tell when a change happens and how often it is revised in a month. Useful to tell how much you are exposed to Look-Ahead Bias in the indicator.

Here's the list:

rrsfs-2016-01-20.txt
rrsfs-2016-02-12.txt
rrsfs-2016-02-19.txt <<<<<
rrsfs-2016-03-15.txt
rrsfs-2016-03-16.txt <<<<<
rrsfs-2016-04-13.txt
rrsfs-2016-04-14.txt <<<<<
rrsfs-2016-05-13.txt
rrsfs-2016-05-17.txt <<<<<
rrsfs-2016-06-14.txt
rrsfs-2016-06-16.txt <<<<<
rrsfs-2016-07-15.txt
rrsfs-2016-08-12.txt
rrsfs-2016-08-16.txt <<<<<
rrsfs-2016-09-15.txt
rrsfs-2016-09-16.txt <<<<<
rrsfs-2016-10-14.txt
rrsfs-2016-10-18.txt <<<<<
rrsfs-2016-11-15.txt
rrsfs-2016-11-17.txt <<<<<
rrsfs-2016-12-14.txt
rrsfs-2016-12-15.txt <<<<<
rrsfs-2017-01-13.txt
rrsfs-2017-01-18.txt <<<<<
rrsfs-2017-02-15.txt
rrsfs-2017-02-16.txt <<<<<

We see that rrs is revised almost every month.

indfs-2016-01-15.txt
indfs-2016-02-17.txt
indfs-2016-03-16.txt
indfs-2016-04-01.txt
indfs-2016-04-15.txt <<<<<
indfs-2016-05-17.txt
indfs-2016-06-15.txt
indfs-2016-07-15.txt
indfs-2016-08-16.txt
indfs-2016-09-15.txt
indfs-2016-10-17.txt
indfs-2016-11-16.txt
indfs-2016-12-14.txt
indfs-2017-01-18.txt
indfs-2017-02-15.txt

Print the post Back To Top
No. of Recommendations: 2
One GTT signal was updated by FRED awhile ago, the second on March 17.

Push the MAX button at the top to get the Feb. number

https://fred.stlouisfed.org/graph/?g=3atC

https://fred.stlouisfed.org/graph/?g=3asR

Both are positive, so trend following timing signals can be ignored.

Since unemployment was mentioned as another fundamental timing signal, here's one of three variants, this one seasonally adjusted. Go to Fred to find the other two.


https://fred.stlouisfed.org/series/UNRATE#0

Civilian unemployment has decreased in the past 11 months. Put the cursor over the graph to read different dates and values.

rrjjgg
Print the post Back To Top
No. of Recommendations: 2

Both are positive, so trend following timing signals can be ignored.

Since unemployment was mentioned as another fundamental timing signal, here's one of three variants, this one seasonally adjusted. Go to Fred to find the other two.


According to the paper I read, *any* indicator being positive means to ignore a timing signal that says to sell.

Not "all" or "most" -- "any".


I always have to think about it carefully, 'cause it's easy to get it backwards. (Or maybe I just confuse easily.)
What he said was that you only want to do a timing-sell when we are in a recession, and always be in stocks when we aren't in a recession. And any of the indicators being positive dis-confirms a recession.
Print the post Back To Top
No. of Recommendations: 1
Ray,

Not the way I understood it, but he does state it in an initially confusing way. Here are quotes from the paper mentioned at the top of the thread:

'That’s exactly what Growth-Trend Timing does. It takes various combinations of high quality monthly coincident recession signals, and directs the moving average strategy to turn itself off during periods when those signals are unanimously disconfirming recession, i.e., periods where they are all confirming a positive fundamental economic backdrop.'

[Here I see an 'all' giving positive signals, not just 'any'. Part of the reason is some give an occasional false positive, so we want the others to confirm no
recession as well.]

'The available monthly signals are:

Real Retail Sales Growth (yoy)
Industrial Production Growth (yoy)
Real S&P 500 EPS Growth (yoy), modeled on a total return basis.
Employment Growth (yoy)
Real Personal Income Growth (yoy)
Housing Start Growth (yoy)
The precise timing criterion for GTT is as follows. Take a reliable monthly growth signal, or better, a collection of reliable monthly growth signals that overlap well to describe the total state of the economy:

If, at the close of the month, the growth signals for the prior month are unanimously positive, then go long or stay long for the next month, and ignore the next step.'

[Unanimous usually means 'all' to me.]

'If, at the close of the month, the growth signals for the prior month are not unanimously positive, then if price is above the 10 month moving average, then go long or stay long for the next month. If price is below the 10 month moving average, sell or stay out for the next month.'

Here we can replace the 10 month SMA test by some of our bear catchers.

In the next part he chooses just two signals to follow (those I've been following) and elaborates on why they should be enough to check for recessions.

rrjjgg
Print the post Back To Top
No. of Recommendations: 6
Well, crap! I *said* it was confusing. That's why in software design (programming) it is bad practice to have flags with a negative meaning.

Which is exactly what he did. Darn it.

There's a bunch of research that says the human brain has a hard time processing negatives, so clear communications requires that things be stated in the positive, especially in a stressful situation. The one example that sticks in my memory is about firefighters. You NEVER tell a fireman in a burning building to "Don't open that door." You always say, "Leave that door closed.". Because what the brain hears is "{inaudible} open that door".
That's why it's "Hit the deck!" and not "DOn't keep standing there." Always say "DO {thing}, not "don't do {opposite thing}."
</soapbox>



"when those signals are unanimously disconfirming recession".
That's negative indicator. A postive indicator statement would be, "...any of the signals indicating a possible recession..." [*]

Selling (on the SMA signal) is being disallowed by the state of the indicator(s). That's negative logic. Positive logic would be "allow the SMA signal to sell" or "gate the SMA sell signal". Argh.

Now I have to go back an take ANOTHER look at my code, to see which way I did it.

A problem: the more indicators you use, the more likely it is that one outlier will be negative while all the others are positive. You run the risk of never being able to "disconfirm recession".

Maybe that's why he limited it to just two. It's been a year+ since I read the paper and incorporated it into my timing code -- so I've forgotten the details.
One of the things about "creative laziness" (A.K.A. mental efficiency) is that once you've figured out the reason *why* something it correct, you no longer have to remember the "why", you only need to remember the "what".

Regardless, whatever my coding details were, it worked, in that it improved the backtests performance and risk metrics.

S&P500 w/o dividends, Jan 1950 to Jan 2016, weekly (43 week SMA):
SMA only:
CAGR 9.2% --- Sortino 1.3 --- stdev 10% --- in 70% --- trades 197 --- whipsaws 50
CAGR when IN 13.4% --- when OUT 5.9%
(Note that the CAGR is still positive when OUT, meaning thatyou left money on the table.)

SMA With growth gate (as coded):
CAGR 11.8% --- Sortino 2.0 --- stdev 12% -- in 84% --- trades 59 --- whipsaws 8
CAGR when IN 14.1% --- when OUT -3.9%.
(Note that this avoided actual losses when OUT.)

Either way, the volatility is better with timing. Stdev when IN = 12% --- when OUT 20%.

==================================================
[*] I've been reading too much Scott Adams lately. Because I just realized that what I just did there was to "prove" that I wasn't wrong, I was right. It was that other guy who was wrong --- by making me misread what he meant. Yeah, and it's turtles all the way down.
Print the post Back To Top
No. of Recommendations: 1
Ahhh, just took a look at my spreadsheet.

Each indicator has a flag column titled "Allow sell?" which is YES if the indicator is positive and NO if negative.
Then there is a column titled "Joint allow sell?", which is YES if any indicator is YES.

So my spreadsheet appears to do it the way he said to.

Confusing.

==========
IND allows sell 161 times.
RRS allows sell 192 times.

Joint allow sell is 244 times.

So, yes, the more indicators you use, the more often you allow the SMA signal to sell (that is, the less often you ignore the SMA signal--ugh there's that negative logic again) --- and if you have enough (too many) indicators, the SMA signal is *always* allowed to sell. Which obviates the whole concept of forcing you to stay IN when not in a recession.
Print the post Back To Top
No. of Recommendations: 1
Ray,

You clearly caught the sentence I also thought was confusing in his description of GTT.

However, I'm not sure your spreadsheet is working that way according to your next post. I don't understand exactly what it codes, so you may be correct.

I assume, given fundamental tests A and B, with your signal yes meaning good times expected (or no recession detected by that test) this would be, if A AND B both signal yes, then stay in. Else, if at least one No signal appears, then decide whether to stay in based on the SMA signal.

I guess something like this codes the logic here, If ((A=no) or (B=no)) and (price<SMA), then go to cash; Else,be in the market.

rrjjgg
Print the post Back To Top
No. of Recommendations: 1
I started sweating. ;-(

So I went to my spreadsheet and verified that it does indeed to what he (and you) said, and not what I said. And it did.

That's where modeling in a spreadsheet comes in handy -- it is easy to see what it does and to change the rules, or to include another column and model the rules the other way and then compare the two.

It's a PITA to run your real screen in a spreadsheet, though, and so I have special-purpose programs to run my real-money screens. The ones that can be done, that is.

So once the spreadsheet logic was verified, I turned to the program to see what it did. Luckily, it did it right also.
Although I did take the opportunity to clarify the code, and to rename the function to "allow_sma_sell".

BTW, I don't recall exactly what if anything he said about the buy rule. The spreadsheet has options for both the sell and the buy rule. One way for the buy is "whenever above the SMA". The other way is "whenever not potentially in recession", which is the complement of the gated sell rule.

Turns out that the first is better. When you are out, buy back in when and only when the current price is above the SMA.
That's CAGR of 11.9% vs. 11.4% for the second.

So, happy day! I accidently did it right.
Print the post Back To Top
No. of Recommendations: 2
All of the graphs have been updated to April and all are positive, so one should ignore
bearish momentum timing signals.

See post 265836 for links to the three FRED charts.
Print the post Back To Top
No. of Recommendations: 6
All of the graphs have been updated to May and all are positive, so one should ignore
bearish momentum timing signals in the GTT system.

See post 265836 for links to the three FRED charts used here.

However, the BCC's are all bullish, so no action required by any timing system.

rrjjgg
Print the post Back To Top
No. of Recommendations: 3
All of the graphs have been updated to June 1, and all are bullish, so one should ignore
bearish momentum timing signals in the GTT system.

See post 265836 for links to the three FRED charts used here.

However, the BCC's are all bullish, so no action required by any timing system.

rrjjgg
Print the post Back To Top
No. of Recommendations: 2
All of the graphs have been updated to July 1, and all are bullish, so one should ignore
bearish momentum timing signals if you use the GTT system.

See post 265836 for links to the three FRED charts used here.

However, the new highs-new lows signal is bearish, so if you already trust the BCC signals, it might
be worth ignoring GTT. I'm following GTT here, because it had a good backtest and makes some sense, but I need more evidence going forward to be a fan.

We are also in the Inferior 5-day period of August. I don't recall if
that's especially bad in August compared to other months, but $SPX loses more in August, on average,than in other months.

The last graph to be updated to July 1 was the unemployment graph. It is interesting to note how low it has gotten. Every time the graph had a sharp low as this one seems to be setting up to be, we
had an 'official' recession within 2 years, usually 5 months to 1.5 years. (The graph is interactive
so point your cursor to read dates and values.) If this pattern continues we most likely will be in a recession by the end of next year or the first half of 2019. Of course, the baby boomers are retiring and using more services, so that might well open up jobs and delay the recession suggested by the graph.
https://fred.stlouisfed.org/series/UNRATE#0

rrjjgg
Print the post Back To Top
No. of Recommendations: 2
The last graph to be updated to July 1 was the unemployment graph. It is interesting to note how low it has gotten. Every time the graph had a sharp low as this one seems to be setting up to be, we
had an 'official' recession within 2 years, usually 5 months to 1.5 years. (The graph is interactive
so point your cursor to read dates and values.) If this pattern continues we most likely will be in a recession by the end of next year or the first half of 2019. Of course, the baby boomers are retiring and using more services, so that might well open up jobs and delay the recession suggested by the graph.
https://fred.stlouisfed.org/series/UNRATE#0


Dare I say this time is different? Low unemployment doesn't cause recessions. It's the Fed's response to an overheating economy, which coincides with low unemployment, that brings a recession. Typically, the Fed has responded to "over heating", and an uptick of inflation, by raising interest rates to "cool" the economy, and typically overdoing it, to bring a recession.

Or maybe it isn't really different. Last time, for the 2008-2009 recesion, I don't think it was the Fed. It was a huge financial bubble to which the Fed was in total denial, which brought a financial crisis and meltdown. So maybe we're ripe for one of those. I can sort of buy into that theory, but my crystal ball isn't clear enough to see black swans.

Elan
Print the post Back To Top
No. of Recommendations: 0
The Chicago Fed puts out a monthly CFNAI index which gauges overall economic activity. If you scroll down to the third chart, you will see the "CFNAI-MA3 and Business Cycles" chart. From the Notes below the third chart, it states, "Following a period of economic expansion, an increasing likelihood of a recession has historically been associated with a CFNAI-MA3 value below –0.70." This would provide a more comprehensive set of data to anticipate recessions. You can sign up for free e-mail alerts. The webpage states, "The CFNAI is released at 8:30 a.m. ET on scheduled days, normally toward the end of each calendar month."


https://chicagofed.org/research/data/cfnai/current-data

Any thoughts about using this?

Sandywater
Print the post Back To Top
No. of Recommendations: 0
Based on the yield curve (3 month T-bills now at 1.11%, 10 year T-bonds at 2.27%), the Cleveland Fed model has a 12.9% probability of recession in 1 year. clevelandfed.org/our-research/indicators-and-data/yield-curv...
Print the post Back To Top
No. of Recommendations: 7
All of the FRED graphs have been updated to August 1, and all are bullish, so one should ignore
bearish momentum timing signals, if you use the GTT system.

See post 265836 for links to the three FRED charts used here.

As usual lately, this doesn't matter, since all BCC signals are bullish and the $SPX is
above the 200 day SMA.


rrjjgg
Print the post Back To Top
No. of Recommendations: 1
I was traveling in August, so I didn't see two earlier posts.

The first is Elan's.
I was just observing what seemed like a pattern in the graphs, not trying to explain it. Your idea that the Fed's meddling may indeed be the main reason for this pattern in the graph is probably correct.

Sandywater,
Thanks for the reference. The only reason I will continue (if I do!) looking at the two or three charts
I'm using is that the first two were specifically suggested by the article about GTT which seemed to incorporate backtesting in support of using these as well as some explanation as to why GTT 'makes sense' to help improve results when using classical momentum strategies. The third chart was suggested by, I think, Ray, so I added it to the mix.

If you can download a spreadsheet of the Chicago signal and check to see if it helps momentum methods
as well as GTT seems to, I would be happy to entertain it. Keep in mind these Fed signals don't change as quickly as the market does, so they are not suitable bear catchers themselves. They only suggest when it is safe to ignore more sensitive gauges of market behavior to avoid unnecessary trading.

rrjjgg
Print the post Back To Top
No. of Recommendations: 5
All of the FRED graphs have been updated to September 1, and all are bullish, so one should ignore
bearish momentum timing signals, if you use the GTT system.

See post 265836 for links to the three FRED charts used here.

As usual lately, this doesn't matter, since all BCC signals are bullish and the $SPX is
above the 200 day SMA.

rrjjgg
Print the post Back To Top
No. of Recommendations: 0
Thank you rrjjgg,

Do you or anyone knows how to setup the GTT timing system on the GTR1 backtester?
So, we can test different etfs such as SECTOR FUNDS ETFs etc.

A link for GTR1 would be greatly appreciated.
Print the post Back To Top
No. of Recommendations: 1
All of the FRED graphs have been updated to September 1, and all are bullish, so one should ignore
bearish momentum timing signals, if you use the GTT system.

See post 265836 for links to the three FRED charts used here.

As usual lately, this doesn't matter, since all BCC signals are bullish and the $SPX is
above the 200 day SMA.


Y'know it's kind of funny. Just before reading this post I read an article that seekingalpha{dot}com suggested to me.
Looking To Sell, Not Buy https://seekingalpha.com/article/4114552-looking-sell-buy?up...

Basically a bunch of emotionally-driven fears, with not a lot of actual backtesting. My favorite "market measurement" was "Price Regression To S&P 500 Trend" -- which is not a measurement at all, it's a phenomenon.

Why do I read this guy? Quite some time ago I came across one of his articles that was interesting and seemed informative, so I put him on my "follow" list. He has created a number of spreadsheets to implement a few various screens (he calls them "portfolios"), somewhat similar to our MI screens. Although access to most of them require you to buy a subscription.

Some of the things his spreadsheets do make some reasonable sense, but I had a few questions--primarily if some of the things they did (like looking at 12 days of prices instead of the typical one day) improved the performance, and how much the improvement was.

His answer really floored me. He said he did not backtest *any* of his spreadsheet screens because backtesting was just too hard. He simply assumed that his complex screens work.

Amazing.

Anyway......the majority of small investors that try to do timing evidently do it by gut fear, and they always seem to have one foot out the door. The data-driven approach is much better, we know that it works (historically), and it's sooo much easier on the emotions.
Print the post Back To Top
No. of Recommendations: 1
The main reason I've been following GTT is the theoretical discussion about it in the linked reference. Most of us notice whipsaws are a nuisance with fees,friction and frequent trading, so some have devised ways to limit them (Ray waits a few weeks to act, some people use collars on one side of a sma test, some only act when BCC=0 ...), but the article made a stronger case for limiting them to improve returns and suggested FRED data for non-market based signals. That's the part I like. It's not based on stock prices alone which are subject to all sorts of human emotions in addition to 'real' potential business influences.

A much more basic question is can you statistically test your ideas with stock prices at all? There have only been 7 recessions since 1970 and only 2 during the period of time
SPY has been been in existence and only 1 recession since most other ETFs have been around. If they don't closely follow a sector index that can be extended back as far as the S&P or the DJI, then you have almost no support for any statistical experiments you might try. You can try to decide if your favorite signal, say BCC=0, helped or hurt, but not the FED signals. Since we've been in a bull market since 2009, most timing methods seem to reduce returns during that period. French, Schiller or other economists have extended some sector indexes into the past. You'll have to find these, then check if any current ETFs follow those indexes. I'd start with their websites, references there and Google.

rrjjgg
Print the post Back To Top
No. of Recommendations: 3
re: articles & readings on the internet today.

And just now I was listening to a Meb Faber podcast with an investment advisor guy. http://mebfaber.com/2017/10/18/episode-76-phil-demuth-nothin...

Interesting tidbit:

“Everything looks expensive. It’s just a question of what looks more expensive than others.” That said “Nothing in my global outlook is telling me it’s time to pull up the anchor and set sail,” even though there seems to be 10 articles each day claiming the sky is falling.

This dovetails into an interesting conversation about how nearly no one believes there will be strong U.S. equity returns over the next decade or so. But what is the psychological impact of everyone believing that? Especially in light of how terrible humans tend to be at these kinds of predictions.
Print the post Back To Top
No. of Recommendations: 3
... the article made a stronger case for limiting them to improve returns and suggested FRED data for non-market based signals. That's the part I like. It's not based on stock prices alone which are subject to all sorts of human emotions in addition to 'real' potential business influences.

Just had a thought in this area, and an analogy.

Dividend growth investing (DGI) is popular now. The proponents say that the success of their investments is due to the growing dividends.

However, Larry Swedroe & others looked at it and concluded that the driver of DGI portfolios is actually that they have a large value factor tilt. It's not that these DGI stocks pay a dividend, it's that they are value stocks. He says that if you screen for value factors and ignore dividends, you get the same performance as a DGI portfolio. Value-factor is sufficient, dividend growth factor is not necessary.
Therefore: a DGI portfolio is really just a value-factor portfolio that uses an _outcome_ factor (growing dividends) to inefficiently identify value. Value-factor is the driver, growing dividend is an outcome. (Not *the* outcome, *an* outcome. DGI picks value stocks, but is blind to value stocks that don't have a dividend. Thus: inefficient.)

Similarly, I'm coming to think that the primary driver of US stock prices is a growing economy -- which means an economy that *is not* in a recession.
Thus, looking at price trends -- SMA, etc. -- is an inefficient way to identify periods when stocks won't do well. Not as strong as the value vs. DGI disconnect, though.

Stocks (mostly) go down during a recession, but they don't always go up when we aren't in a recession. The price trend is still a useful factor when not in recession.
And *that's* probably why being slow on taking the price-trend sell signal is better than being fast. Being slow to sell when not in recession gives time for the price to get past the speedbump and resume the upward trend.

Or maybe this is all just confirmation bias. ;-(
Print the post Back To Top
No. of Recommendations: 0
Or maybe this is all just confirmation bias. ;-(

Maybe it's superior intellect at work! :-)

rrjjgg
Print the post Back To Top
No. of Recommendations: 9
All of the FRED graphs have been updated to October 1, and all are bullish, so one should ignore bearish momentum timing signals, if you use the GTT system.

See post 265836 for links to the three FRED charts used here.

As usual lately, this doesn't matter, since all BCC signals are bullish and the $SPX is above the 200 day SMA.

However, both the 9 day and 13 day EMA new high-new low signals have been falling toward 0 lately.

Today was a nice uplift for the signal. I don't recall how the week of Thanksgiving usually does, but it is during the 'Inferior Five' week, so
this signal may still switch bearish next week for those following it as an action signal. (As mentioned, the GTT method will ignore any such change.)

http://stockcharts.com/h-sc/ui?s=%24NAHL&p=D&b=5&...

rrjjgg
Print the post Back To Top
No. of Recommendations: 4
All of the FRED graphs have been updated to November 1, and all are bullish, so one should ignore bearish momentum timing signals, if you use the GTT system.

See post 265836 for links to the three FRED charts used here.

As usual lately, this doesn't matter, since all BCC signals are bullish and the $SPX is above the 200 day SMA.

I read a blurb on a newsletter that an accurate (at least recently) timing indicator is the consumer confidence index. It seems to dive shortly before the market does. Probably worth considering.

rrjjgg
Print the post Back To Top
No. of Recommendations: 1
...I assume, given fundamental tests A and B, with your signal yes meaning good times expected (or no recession detected by that test) this would be, if A AND B both signal yes, then stay in. Else, if at least one No signal appears, then decide whether to stay in based on the SMA signal.

I guess something like this codes the logic here, If ((A=no) or (B=no)) and (price<SMA), then go to cash; Else,be in the market.

rrjjgg


rrjjgg & Rayvt,

In you rules, on the backtest, When is the buy signal, If (A=no) or (B=no)? The BUY signal is not clear when the FRED s gives "recession signal". Does it means, if (A=no) or (B=no) and Price> 200SMA then BUY?

Also, could you explain how your rules gives a signal for A or B become "=no" or "=yes"?

[FRED A=Ind. Production Grt. B=Retail Sales & Food Services.]


Thank you
Print the post Back To Top
No. of Recommendations: 4
In you rules, on the backtest, When is the buy signal, If (A=no) or (B=no)?

Well, it's one of those things where, once you understand it, it is perfectly clear. But until you do understand it, it's very unclear.

The primary buy/sell signal is the SMA of the S&P500. Period.

However... the sell signal is gated by the "economy is doing fine" signal. If the economy is doing fine, the SMA sell signal is ignored. "Gated" in the sense of "the sell signal is blocked by the gate".

It is confusing to understand because the description of the IPG & RSG is "disconfirming a recession". That's really hard to comprehend because human minds don't work to good for negatives.[*] (Which is why firemen don't say "Do not open that door." They say "Leave the door closed." Your mind doesn't hear the "not".)

So.
1) If the SMA says to buy, then buy.
2) If the SMA says to sell, consult the IPG/RSG to see if the sell signal is allowed.

Note that if the SMA says to sell but the FRED signals block the sell, then you don't sell. But while the SMA signal is sitting there mashing the sell button which is blocked, once the FRED signals cease to block the sell signal, then the sell signal is allowed and then you sell.

Colloquially, the way to understand it is: you should be in stocks when anything is looking good, and you should be out of stocks when everything looks bad. It is not symmetric.


----------
[*] That's why I don't do it manually. I have a script that does all these calculations and reports the ultimate BUY/SELL outcome. The script does not get confused by negative logic.
Print the post Back To Top
No. of Recommendations: 3
Colloquially, the way to understand it is: you should be in stocks when anything is looking good, and you should be out of stocks when everything looks bad. It is not symmetric.

No, scratch that. I said it wrong. (Told you this is confusing!)

Correct statement is:
Colloquially, the way to understand it is: you should be in stocks when the market is going up, and you should only be out of stocks when things look really bad. It is not symmetric. The default, preferred, state it to be in stocks. You only get out of stocks when the market is going down _and_ the economy is looking bad. (The economy is looking bad if _either_ IND or RRS is down Y-O-Y.) The bias is to be in stocks.
Print the post Back To Top
No. of Recommendations: 0
Colloquially, the way to understand it is: you should be in stocks when anything is looking good, and you should be out of stocks when everything looks bad. It is not symmetric.

No, scratch that. I said it wrong. (Told you this is confusing!)

Correct statement is:
Colloquially, the way to understand it is: you should be in stocks when the market is going up, and you should only be out of stocks when things look really bad. It is not symmetric. The default, preferred, state it to be in stocks. You only get out of stocks when the market is going down _and_ the economy is looking bad. (The economy is looking bad if _either_ IND or RRS is down Y-O-Y.) The bias is to be in stocks.


I think you said it better the first time. Both statements are saying the same. The first is concise. The second is meandering and redundant.

Elan
Print the post Back To Top
No. of Recommendations: 0
Rayvt,

Thank you for your reply.

I think I understand your backtest rule to "SELL".

However; your backtest rule regarding "BUY" signal:
(i)when (A=no) OR (B=no) AND Price>200SMA then "BUY" ?
(ii)when (A=no) AND (B=no) AND Price>200SMA then "BUY" ?
(iii)when (A=yes) AND (B=yes) AND Price>200SMA then "BUY" .

I think i, ii, iii, covers all the conditions for BUY.
Do I understand your backtest rule correctly regarding "BUY"?

If so, this rule get you back to the market faster and is compatible with the sharp character of the market bottoms.

Thanks
Print the post Back To Top
No. of Recommendations: 9
I think i, ii, iii, covers all the conditions for BUY.
Do I understand your backtest rule correctly regarding "BUY"?


You are making it too complicated.

The buy rule is: "if Price >= 200SMA then "BUY" Period. The end.

Actually I use 43 week SMA, but it comes to the same thing. 200 days == 43 weeks == 10 months.

I've got it all in a spreadsheet, with parameters.
From 1/1/1950 to 1/1/2016 (S&P500 incl dividends), at 0.25% slippage per round-trip trade
The rules I just stated:
59 trades, in 84% of the time, 8 whipsaws, 11.8% CAGR, 2.03 Sortino


If the buy rule is "not in recession" (that is, opposite the sell gate)
75 trades, in 87% of the time, 9 whipsaws, 11.3% CAGR, 1.54 Sortino

With the standard SMA rule only:
197 trades, in 70% of the time, 10 whipsaws, 9.2% CAGR, 1.30 Sortino

With my previous sell rule (sell at 4% below the SMA):
67 trades, in 77% of the time, 0 whipsaws, 10.5% CAGR, 1.76 Sortino

My alternative previous sell rule (4 consecutive week below the SMA):
89 trades, in 76% of the time, 20 whipsaws, 10.2% CAGR, 1.46 Sortino

You can see that being on a hair-trigger to sell is costly. Just the opposite of what most people think. You want to stroll casually to the sell window, not run hastily.

------------
For my own amusement I also count the "successful OUTs". That's when you got back in at a lower price than you got out at.
For the GTT rule it's 13 --45% of them.
For standard SMA it's 32 --33% of them.
Print the post Back To Top
No. of Recommendations: 10
The FRED graphs have been updated to December 1,2017, and all are bullish, so one should ignore bearish momentum timing signals, if you use the GTT system.

See post 265836 for links to three FRED charts. The
original GTT article uses only the first two for the basic GTT system.

As usual lately, this doesn't matter, since all BCC signals are bullish and the $SPX is above the 200 day SMA.

rrjjgg

I read a blurb on a newsletter that an accurate (at least recently) timing indicator is the consumer confidence index. It seems to dive shortly before the market does. Probably worth considering.
Print the post Back To Top
No. of Recommendations: 2
https://finance.yahoo.com/news/goldman-sachs-telling-million...

GS backs up the general thesis behind GTT.

I summarize.....bull markets don't die of old age or over valuation, they die because of recessions. 17% likelihood of a US recession in next two year per GS.

Looks like the bull market still has legs to run.

Peace,
Opihi
Print the post Back To Top
No. of Recommendations: 8
This morning the final FRED graph has been updated to January 1,2018, and all are bullish, so one should ignore bearish momentum timing signals, if you use the GTT system.

See post 265836 for links to three FRED charts. The
original GTT article uses only the first two for the basic GTT system.

Only the new highs-new lows BCC signal is currently bearish and the $SPX is above the 200 day SMA.

rrjjgg
Print the post Back To Top
No. of Recommendations: 12
This morning the final FRED graph has been updated to February,2018, and all are bullish, so one should ignore bearish momentum timing signals, if you use the GTT system.

See post 265836 for links to three FRED charts. The
original GTT article uses only the first two for the basic GTT system.

Of course, it doesn't matter, since all our signals are bullish, suggesting the market is in a correction, not beginning a crash.

The new highs-new lows signals[9 &13 day versions] are now bullish
http://schrts.co/Rkopzy as easily seen using the MACD charts. The slope test is bullish (blue 200 day sma line above the red 200 day sma line from 10 days ago), the dbe 99 day price channel signal is bullish
and the $SPX is above its 200 day SMA. http://schrts.co/LfzYED


rrjjgg
Print the post Back To Top
No. of Recommendations: 0
I have not been able to find it using datahelper.

Please, what is the MDD for the GTT timing on the S&P500?

Thanks a lot!

jc
Print the post Back To Top
No. of Recommendations: 0
One MDD is mentioned on the first page of the originating GTT article:
http://www.philosophicaleconomics.com/2016/01/gtt/

For perfect recession timing only, it is -27 % vs -51% with no timing over the period used. Since GTT is close to this (see graphs) I'd guess that's a start. Near the end of the article he gives information including MDD based on $SPX starting at various times. It's usually a bit more than using Monthly MA (his MAA, I think) momentum timing, but less than Buy &Hold. His method raises the CAGR to about 11.2% or an increase of about 10% more than B&H.

It is certainly one of the most interesting articles I've read.

Thanks for rousing me enough to read it again!

rrjjgg
Print the post Back To Top
No. of Recommendations: 0
Rrjjgg,

Thank you very much indeed!

I will read it through.

jc
Print the post Back To Top
No. of Recommendations: 7
Yesterday the final FRED graph was updated to March, 2018, and all signals are bullish, so one should ignore bearish momentum timing signals, if you use the GTT system.

See post 265836 for links to the three FRED charts. The unemployment chart is optional. The
original GTT paper (http://www.philosophicaleconomics.com/2016/01/gtt/) uses only the first two for the basic GTT system.

All our momentum signals are bullish, suggesting the market was just correcting lately, not beginning a crash.


rrjjgg
Print the post Back To Top
No. of Recommendations: 6
The GTT graphs are now updated to April, 2018, and all signals are bullish, so one should ignore bearish momentum timing signals, if you use the GTT system.

See post 265836 for links to the three FRED charts. The unemployment chart is optional. The
original GTT paper (http://www.philosophicaleconomics.com/2016/01/gtt/) uses only the first two for the basic GTT timing system.

All our usual momentum signals are bullish.

rrjjgg
Print the post Back To Top
No. of Recommendations: 12
The GTT graphs are now updated to May, 2018, and all signals are bullish, so one should ignore bearish momentum timing signals, if you use the GTT system.

See post 265836 for links to the three FRED charts. The unemployment chart is optional. The
original GTT paper (http://www.philosophicaleconomics.com/2016/01/gtt/) uses only the first two for the basic GTT timing system.

All our usual momentum signals are bullish.

The Fred graphs are usually updated about the 15th of the month. I anticipate the 'dying euphoria' or 'no new high in 99 trading days' signal will turn bearish before the next GTT update, but I don't anticipate a change in the slowly changing GTT signals by next month.

rrjjgg
Print the post Back To Top
No. of Recommendations: 0
Does anyone use this system?
Print the post Back To Top
No. of Recommendations: 2
We've all followed it, if we've been in the market! :-)

It makes sense, but it was first discussed here during this last period where new highs minus new lows has triggered, but not BCC=0 (all three signals bearish), which
seems to be the main signal used in GTR1 backtesting. Recently the S&P500 has dropped below
its 200 day moving average for a day or so, but hasn't stayed there long enough to trigger
Ray's signal.

It may be several months or several years before we get an opportunity
to test it going forward.

rrjjgg
Print the post Back To Top
No. of Recommendations: 0
We've all followed it, if we've been in the market! :-)

It makes sense, but it was first discussed here during this last period where new highs minus new lows has triggered, but not BCC=0 (all three signals bearish), which
seems to be the main signal used in GTR1 backtesting. Recently the S&P500 has dropped below
its 200 day moving average for a day or so, but hasn't stayed there long enough to trigger
Ray's signal.

It may be several months or several years before we get an opportunity
to test it going forward.


I thought most people used things like QTAA, or gem. I did not know that they only used qtaa, or gem, when there were indications of a recession. I only very recently got interested in timing, so I am behind in the amount of knowledge I have about it.
Print the post Back To Top
No. of Recommendations: 12
The GTT graphs are now updated to June, 2018, and all signals are bullish, so one should ignore bearish momentum timing signals, if you use the GTT system.

See post 265836 for links to the three FRED charts. The unemployment chart is optional. The original GTT paper (http://www.philosophicaleconomics.com/2016/01/gtt/) uses only the first two for the basic GTT timing system.

All our usual momentum signals are bullish.

The Fred graphs are usually updated about the 15th of the month. The last to update is usually the Industrial Production graph which was posted today, the 17th.

rrjjgg
Print the post Back To Top
No. of Recommendations: 4
I'm using a kindle
so I will just report
the GTT reports are
updated to July. All
are good.

rrjjgg
Print the post Back To Top
No. of Recommendations: 7
The GTT graphs are now updated to August, 2018, and all signals are bullish, so one should ignore bearish momentum timing signals, if you use the GTT system.

See post 265836 for links to the three FRED charts. The unemployment chart is optional. The original GTT paper (http://www.philosophicaleconomics.com/2016/01/gtt/) uses only the first two for the basic GTT timing system.


rrjjgg
Print the post Back To Top
No. of Recommendations: 16
The GTT graphs are now updated to September, 2018, and all signals are bullish.

If you are now using momentum based timing signals to limit drawdowns, the backtests in the GTT paper indicate it will cost you during this period between recessions. The price of insurance?

If at least one of the two basic Fed signals is bearish suggesting a recession may be coming, then timing methods not only control drawdowns, but can be profitable should a recession arrive.

See post 265836 for links to the three FRED charts. The unemployment chart is optional. The original GTT paper (http://www.philosophicaleconomics.com/2016/01/gtt/) uses only the first two for the basic GTT timing system.

rrjjgg
Print the post Back To Top
No. of Recommendations: 3
It seems that Sales by retail stores has opened the gate:

https://fred.stlouisfed.org/graph/?g=2aF7

Last value (Sept 2018) is 2.39, under sept 2018 (2.85)

And SPX is under 43w SMA....
Print the post Back To Top
No. of Recommendations: 0
Sorry,

It should say

"under sept 2017 (2.85)"
Print the post Back To Top
No. of Recommendations: 4
The GTT graphs are now updated to October, 2018, and all signals are bullish.

If you are now using momentum based timing signals to limit drawdowns, the backtests in the GTT paper indicate it will cost you during this period between recessions. The price of insurance?

If at least one of the two basic Fed signals is bearish suggesting a recession may be coming, then timing methods not only control drawdowns, but can be profitable should a recession arrive.

See post 265836 for links to the three FRED charts. The unemployment chart is optional. The original GTT paper (http://www.philosophicaleconomics.com/2016/01/gtt/) uses only the first two for the basic GTT timing system.

rrjjgg
Print the post Back To Top
No. of Recommendations: 1
If I am not wrong, retail sales are lower:

October 2018: 1.98467
October 2017: 3.26767

Source: https://fred.stlouisfed.org/graph/?g=2aF7#0

Click on 1Y graph and insert today's date to be able to get October 18 value.

jcbb
Print the post Back To Top
No. of Recommendations: 2
Hi,jcbb.

Your Fred graph is not one of the GTT signals I've been following here. Check the original post in this thread for the article suggesting the two
I've followed. I think Ray suggested the unemployment graph also be followed,
so that was also added early on.

You may misunderstand which graphs are referred to when I talk about Fred graphs.

Retail sales are lower month to month? So what? If you look at your graph, sales are up and down all the time. Monthly changes are generally indistinguishable from noise.

So what else in your graph could signal a recession is approaching? It appears one
might have the graph drop below 0, but that happens too late in many of the indicated
recessions to be a leading indicator. It also can happen between recessions(false positive) and in an instance or two didn't drop below 0 during the recession(no signal), so that isn't much help either. Your graph does not seem to be very useful for our purposes.


rrjjgg
Print the post Back To Top
No. of Recommendations: 3
This link was posted with number 267414 by Borisnand, who did a thoughtful study on the subject (thanks!).

https://fred.stlouisfed.org/graph/?g=2aF7

Anyway, you can check that figures are exactly the same as:

https://fred.stlouisfed.org/graph/?g=3asR

Maybe I am wrong but let me copy and paste what Rayvt explained on post 267425:

"""The GTT signal I use is:
If *either* of the IND or RRS-G is in a year-over-year decline (that is, the current index is below what is was 12 months ago), then allow the sell signal. Otherwise, block (ignore) the sell signal.
""""

I understand that an October value lower than same month last year, "opens the gate".

I will be very grateful for any explanation about what I have missunderstood.

Thanks

jcbg
Print the post Back To Top
No. of Recommendations: 3
Hi jcbb, Thanks for reminding me of boris' post. The main 'misunderstanding' is about what I'm doing.
I have tried to follow the signals the referenced article mentioned as I understood them and added Ray's suggested graph since it seemed useful. The graph you referenced is % growth from last year, so the yearly change is in the graph and one should be on the lookout for a recession when the graph drops below 0, not merely that the graph dropped in value. In other words, you pointed out growth slowed, but it is still positive,so nothing was signalled by the scheme I'm following. As I noted, the signals are not always timely or perfect, so using more than one test should improve the scheme. (Boris misnamed the method; GTT represents Growth-Trend Timing.)

So by mixing your interpretation of Boris' post with this thread might cause confusion as to the system I'm testing. I guess that's my main quibble.

But I keep learning where I need to improve my ideas about this, so don't be afraid to post.
For example, when I first read the GTT article, I thought it was great, since we could (I thought) ignore timing signals most of the time. But after a few posts about recent articles elsewhere claiming timing methods cost more than they saved, I recalled what I had heard for years. It costs to time during any bull market and is only profitable during bear markets, if at all. Then I recalled why we do it.
Peace of mind and preservation of capital. How can one put a price on such a personal thing? A couple of MI Fools followed
their screens all through the recent 2007-2009 bear market, but most of us probably got out at different times depending on our threshold of pain.

I also changed my standard post to suggest the idea that timing is insurance.

For another example, I've been concentrating on when GTT suggests starting to time, not when it suggests ending timing. Boris
pointed out the recession was over a year before the market bottomed in 2002. I checked the signals. Retail Sales growth turned positive in Sept. 2001, but the Ind. Production graph
didn't go above 0 until May, 2002. Still early, but requiring both signals helped. On the other hand the only BCC signal positive in May was $NAHL which went negative by June, so it's
likely one would hesitate to jump back in until an upturn appeared. Interesting observation
though.

rrjjgg
Print the post Back To Top
No. of Recommendations: 2
Jcbg,

I had originally been making the same mistake as you in looking at FRED graphs. These graphs show the PERCENTAGE CHANGE from last year to this year, not raw numbers. As long as the change is positive (i.e., above the zero line), then any bear timing signals should be ignored. The October 2018 number of 1.98467 indicates that this is a positive change over last year's raw number (whatever that was). The graph is just an easy way to visualize the difference between this year's and last year's raw number.

Hope that helps.
Cathy
Print the post Back To Top
No. of Recommendations: 14
The GTT graphs are now updated to November 1, 2018, and all signals are bullish.

If you are now using momentum based timing signals to limit drawdowns, the backtests in the GTT paper indicate it will cost you during this period between recessions. The price of insurance?

If at least one of the two basic Fed signals is bearish suggesting a recession may be coming, then timing methods not only control drawdowns, but can be profitable should a recession arrive.

See post 265836 for links to the three FRED charts. The unemployment chart is optional. The original GTT paper (http://www.philosophicaleconomics.com/2016/01/gtt/) uses only the first two for the basic GTT timing system.

rrjjgg
Print the post Back To Top
No. of Recommendations: 3
I have been trying to follow this thread. I have created a FRED graph with some assumptions. Am I correct in these assumptions and the percent change graph I created?

The first step, to see if we should leave the market, is to see if the 200 Day Moving Average (of the S&P 500)or some other BCC signal is in a sell mode.

If the signal(s) indicate to get out of the market, then check some other GTT signal to confirm that is is best to get out of the market.

The two most important FRED factors for the GTT timing are:

1) Industrial Production Index (INDPRO)

2) Advance Real Retail and Food Services Sales (RRSFS)

A less important factor is 3) Civilian Unemployment Rate (UNRATE)

My FRED graph of percent change of the the 3 factors is at https://fred.stlouisfed.org/graph/?graph_id=527662

The key is to determine if percent change of a factor is up or down from the previous month.

Since the factors in my FRED graph show all percents below zero, the signal says to get out of the market.

Also, has the maximum drawdown of this GTT approach been determined?
Thanks,
Elliot
Print the post Back To Top
No. of Recommendations: 8
Since the factors in my FRED graph show all percents below zero, the signal says to get out of the market.

No, I don't think this is the proper way to use GTT.

Try this FRED chart, which displays the year over year change in percentage of Retail Sales.
https://fred.stlouisfed.org/graph/?g=mtrV


The first step is to check that the most recent month is greater than 0 (i.e., higher Retail Sales than same time last year). If and only if that percentage is LESS than 0 would you proceed to check your timing measure of choice.

Currently, the FRED chart shows that the percentage change is higher, therefore, take no further timing action.

Cathy
Print the post Back To Top
No. of Recommendations: 2
Two GTT graphs are now updated to December 1, 2018, and neither is predicting a recession.

The Retail Sales graph seems to be a victim of the gov. shutdown and has not been updated. The other two either were updated before the shutdown hit hard or have independent data feeds. Has the shut down impacted enough people that
the Retail Sales graph will show weakness when it is next reported? Probably. We will eventually know.



rrjjgg
Print the post Back To Top
No. of Recommendations: 10
Two GTT graphs are now updated to January 1, 2019, and neither is predicting a recession.

The Retail Sales graph update seems to be lagging behind and has only been updated to December 1. It did drop 1.01 to about 0.72, but recent headlines at Yahoo
Finance claimed the December numbers were out. It is not
clear to me if these are already included in this graph or will be posted when it is next updated. If the latter, I expect it to go negative at the next update.

If you are a timer trying the GTT scheme, it might be prudent to use your favorite timing signals now, until this is clear.

Some signals seem to be as follows: $spx is above its 200sma and is also above its value 10 days ago (SMA signal), both bullish
http://schrts.co/RhEiByUk
Jim's 14 week sma from 16 weeks ago is still above the $SPX, bearish,
and looks like it will be at least 1-3 months before a possible change.

boris noted DBE, the 99 day signal, is now bearish at GTR1. The change at
stockcharts is difficult to see, since it is only about one point, but it happened there, too. It should remain stable for about 7 trading days then rapidly come down to about 2817, so if $SPX can rise about 1.52% in the next few weeks, we would get a new long term dbe bullish signal. Unfortunately we are in the Inferior 5 days, so that might slow down things.

New highs minus new lows is bullish

http://schrts.co/SDxNubNA and the bullish percent index is
bullish
http://schrts.co/iDzMEGQZ

Elan's Feb. 11 list is tied, 2 bulls 2 bears, but $spx sma changed, so 3 bulls 1 bear, if no other changes.
https://boards.fool.com/arezi-ratio-for-feb-11-34130943.aspx...






rrjjgg
Print the post Back To Top
No. of Recommendations: 7
All three GTT graphs are bullish, with no warning of a recession indicated.

When the retail sales graph was updated to Jan 1, the revised Dec. figure was slightly negative, which was my expectation in my last post. The Jan.
value improved and is again positive.

Several posts have noted all bear catchers are positive. I also add that the bullish percent ndx pt.& figure chart reads bull confirmed at stockcharts.com and Jim's 14 week SMA from 16 weeks ago has also just slipped below the S&P weekly close, so that is bullish as well.

http://schrts.co/CKXCDNDE

https://stockcharts.com/freecharts/pnf.php?c=%24BPNDX,PWTADA...



rrjjgg
Print the post Back To Top
No. of Recommendations: 3
All three GTT graphs are bullish and Lohill has reported the bear catchers are
all in bull mode.



rrjjgg
Print the post Back To Top
No. of Recommendations: 6
All three GTT graphs are bullish, updated to at least April 1, with no warning of a recession indicated.

rrjjgg
Print the post Back To Top
No. of Recommendations: 12
All three GTT graphs are bullish, updated to at least May 1, with no warning of a recession indicated.



rrjjgg
Print the post Back To Top