Post New

With a march close on the S&P 500 of 4109, the elusive Coppock Signal appears to have triggered. Now may be a good time to buy the market for those seeking decent long-term returns.

Love the new board. Many thanks, Manlobbi. Brk POS to zero. :)

With a march close on the S&P 500 of 4109, the elusive Coppock Signal appears to have triggered.

Seems right to me.
They're pretty rare. The last "buy" was end May 2016. The one before was end May 2009. Before that, a pair of signals November 2002/April 2003.

For why anyone might care, this might be a quick backgrounder, from a couple of years ago
http://www.datahelper.com/mi/search.phtml?nofool=y...

Sure, it's impossible to time the market accurately.
But it's not that difficult at all to stack the deck.
That is to say, it's not hard to divide time into two states, one of which has materially lower long run average rates of return than the other, in and out of sample.
There are lots of ways that a statistical edge can be turned into extra money without betting the rent on every signal being perfect.

One fine example would be "has their been a Coppock buy signal in the last year and a half".
Since the Coppock signal was published in late 1962, the CAGR with answer "yes" has been inflation+ 10.9%/year for that 35% of the time.
The CAGR with answer "no" has been inflation+ 3.2%/year for the other 65% of the time.
Not a bad validation for 58 years out of sample.
Using this as a long/cash signal would probably have reduced your long run returns, since a real 3.2% is more than what you'd get on cash.
But it's not hard to imagine constructing a portfolio that used this signal to give better returns without increased risk on any meaningful metric.

For those who like their lunch overcooked, it would have worked better to use 19 months instead of 18 : )
i.e., bullish all of April if, at the end of March, the time since the last buy signal was <=19 months.

Jim

Thanks, never heard of it prior to this thread.

when you say, "Since the Coppock signal was published in late 1962, the CAGR with answer "yes" has been inflation+ 10.9%/year for that 35% of the time."

there is only a 35pct chance of achieving inflation + 10.9pct. Is that not fairly low?

thanks

Also any recommended sites that track this for the S&P or particular stocks as I cant seem to find any

thanks

Also any recommended sites that track this for the S&P or particular stocks as I cant seem to find any

Try https://finance.yahoo.com/ or https://stockcharts.com/, they both have it in the list of indicators.

This curve, definition from Investopedia?


The Coppock Curve can be calculated as follows:
Coppock Curve = WMA10 of (ROC14 + ROC11)
Where:

WMA10 = 10-period weighted moving average
ROC14 = 14-period rate of change
ROC11 = 11-period rate of change
Given this formula, proceed through the following steps:

Calculate ROC14 using the most recent monthly closing price relative to 14 periods (months) ago.
Calculate ROC11 using the most recent monthly closing price relative to 11 periods (months) ago.
Add ROC14 to ROC11. Continue to do this each period going forward.
Once there are at least 10-periods of ROC14 added to ROC11, take a weighted moving average of the last 10 values. Continue to do this each period going forward.

If above is correct:
it's a 10 month weighted average of the sum of 14 month and 11 month returns.

How robust is it to fiddling the 10,14,11 numbers etc etc?
As it stands, it doesn't seem very appetizing.

technical analysis.... voodoo perhaps?

"Is that not fairly low?"
That's not was was meant. It might help if we pull apart the sentence:

"Since the Coppock signal was published in late 1962, the CAGR with answer "yes" has been inflation+ 10.9%/year for that 35% of the time." = "The Coppock signal was published in late 1962. The CAGR for when the Coppock signal has been 'yes' has been inflation +10.9%/year. Since 1962, the Coppock signal has been 'yes' 35% of that time."

thanks for the clarification

Also any recommended sites that track this for the S&P or particular stocks as I cant seem to find any

Though you could easily do the same calculation for a single stock, but by convention and observation it has no meaning.
The signal was designed to work on the broad index, trying to capture the end of the mood of despair around the end of a bear market.

technical analysis.... voodoo perhaps?

Well, as mentioned in the quote up thread: predicting the market reliably isn't possible. That sort of claim would be voodoo.
But divining times of "more likely to be good" and "more likely to be bad" isn't the least bit difficult.

This particular signal was invented about 60 years ago, with the very modest goals of identifying moments that were above-average times to start long term holds.
It has done that extremely well since then, on average. 60 years is a pretty good pile of out-of-sample statistical validation.
What one does with that information, if anything, is up to the observer.

e.g., during months it has been <=4 months since the last "buy" signal, the S&P has risen at a rate of over inflation+20%/year on average since the signal was published.
The overall return from the S&P since that date has been inflation + 6.5%, pretty much spot on the long run average.

Jim

Post New

Berkshire Hathaway FAQ

---

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