### Coppock Guide Update & Forecast For 2009

Published January 8th, 2009 in Technical AnalysisSince the last time we looked at the Coppock Guide indicator, a few readers have been asking me about updates for it. So here it is:

Since this indicator needs monthly data, with the end of December 2008, we have another data point. The latest Coppock level is -286 which is even more extreme than the last. And it is only second to levels which we last saw in the brutal bear market of the 1970’s:

But what does that mean? Well, first, the extremely negative level means that when the signal arrives, it will be powerful. And since the Coppock Guide foretells bull markets by curling up, we are watching for an increase - even if it as small as +1. Although we’d prefer a more decisive signal, of course.

To give you an idea of how the Coppock guide responds to the market, let’s propose that for the next few months we see the following S&P 500 levels:

- November 2008 — 896.24 (actual)
- December 2008 — 903.25 (actual)
- January 2009 — 1000 (hypothetical)
- February 2009 — 1100 (hypothetical)
- March 2009 — 1200 (hypothetical)
- April 2009 — 1300 (hypothetical)

In other words, the S&P 500 goes up by 100 points each month for the next 4 months. In that hypothetical scenario, the Coppock curve would turn up by the end of February 2009 by the minimum. And in March, it would turn up significantly (+20).

You can see this on the chart at the extreme right - light blue.

Put another way, if by the end of this month the S&P 500 by some miraculous device ends at 1128, it will be enough to cause the Coppock curve to turn up the minimum +1 for a signal. That would be an astronomical 25% rise in one month. Crazy, I know, but that’s how much it would take in a short time to move the Coppock guide to a signal.

So that gives you an idea of how slow this indicator works. On the one hand it lags the market significantly but on the other hand, its signals have been very accurate historically.

Of course, this is assuming that the market doesn’t fall or tread water but rise. And it assumes that the signal given as a result is not a false one.

**Enjoyed this? Don't miss the next one, grab the feed or **

The last time you wrote about this it struck me that this indicator was very similar to a monthly MACD with the right paramerets. Have you compared them?

hey babak your email box is full.

I thought as much about the link exchange, just figured i’d ask.

I was never on word press, my last site I made myself html / CSS. It didn’t

really work so well, so I switched to NING. I’m hoping that by using NING the site might become self sufficient.

I’m guessing you are Canadian? And work in the finance industry?

ps. I used RSS feeds from your news site, I hope that is ok? It does link the users to your page. Let me know, if not I will remove them.

Thanks.

Russ, no I haven’t thought about it like that. Can you explain in more detail?

Thanks Darrell, I actually got your email - dunno why it bounced saying the box was full. I’ll have to look into that to see if there is something haywire. The RSS feed is ok. Thanks for the heads up

cheers… thanks babak!

Wanted you to know I just wrote a piece about Coppock and linked back to Trader’s Narrative.

Guru, thanks for the note. What is the MTI? I didn’t find an explanation for it on your site.

The MTI is essentially a momentum tool. I calculated 4 moving averages on 45-years (actually March 12, 1963-December 31, 2007) of S&P 500 Index closing values. There are 24 permutations of 4 moving averages and 5 possible positions of the Index itself relative to these averages. For example, over the 44 years (nearly 11,400 trading days), the Index was above all 4 moving averages (the most bullish alignment) 53% of the time; it was below the 4 moving averages (the most bearish alignment) 18% of the time. Combining all possible moving average arrays and the position of the Index relative to them yields 120 possible combinations. In actual fact, only 93 of those possible combinations actually occurred and some for only a few days.

Rather than solve the question on a theoretical basis, I took a bull-in-a-china-shop approach. A $1000 portfolio of S&P stocks on 3/12/63 would have grown to $22,414 by 12/31/07 (the buy-and-hold strategy). I then back-tested on a binary basis (own or not own) each of the 93 combinations to determine whether, at the end of 45 years, the hypothetical portfolio of the SPY Index would have been higher or lower than the buy-and-hold strategy. Combining all the possibilities of these 93 combinations for every trading day on an “all-in” or “all-out” basis resulted in the $1000 growing to $50,171. And most aggressively, had the “all-out” days been fully-margined (150%)” then the $1000 would have grown to $258,602! Surprisingly, the back-testing didn’t demand overly active trading activity. Over the 45 years, there weren’t very many runs of less than 10 trading days and most ran up to 30-60 days.

Actually, due to the severe and deep collapse of the market in 2008, the MTI signaled an “all-out” cash position December 26, 2007 and hasn’t yet signaled it safe to be back in and looks like it will be some time before the MTI signals it’s safe (low risk of further decline) to fully reenter the market.

When I said that the Coppock Guide is like an MACD on a monthly basis what I had in mind is that both look at the second derivative of the time series (i.e., prices) they are tracking.

The MACD subtracts a shorter term loving average from a longer term moving average. The reason for using moving averages instead of the actual price is simply to smooth out the series so that it doesn’t bounce all over the place. Subtracting a shorter term MA from a longer term MA gives you something like the first derivative of the series.

The first derivative is the rate of change of the series. The faster MA gives you something like the value of the series a few steps back, i.e., close to the present but smoothed out to eliminate random daily fluctuations. The slower MA gives you the value of the series a few more steps back. Subtracting the slower from the faster gives you the change in the series. You would get something similar if you took a single moving average and subtracted a value a number of steps back from the current value. Basically it tells you whether the series is going up or down.

The next step with MACD is to see if that first derivative is changing. That’s the point of the MA of the MACD line, which serves as the signal or trigger line. If the MACD turns around and crosses the trigger it means that the first derivative has changed enough so that it’s worth paying attention to. So basically one is looking at the change in the first derivative, which is the second derivative.

The Coppock Guide does the same thing. Instead of using two MAs. It subtracts previous values from the current value. That gives you the first derivative. Then it looks to see if that series is turning around. (The business of using 14 months and 11 months, etc. is relatively unimportant. What matters is that you are looking at the change over approximately a year.)

In both cases you want the first derivative to be strongly negative, i.e., that the price series itself is going down fast, so that a change from that trend makes a difference. That’s why Coppock and MACD are most important when the signal occurs from a relatively extreme spot.

I said it backwards. MACS subtracts the longer term MA from the shorter, which gives you the current change.

Russ Abbott, great observation. I’ve done a lot of work with both Coppock and the MACD. And you are right: they pretty much do the same thing. In fact, the MACD is so much easier to calculate, I like to use it instead. But when you compare the Coppock and MACD you’ll def see that they give you pretty much the same thing — and I mean no disrespect to Coppock users, it really is a fascinating indicator, IMO.

On your chart - I see about March ‘03, but why was Jan ‘74 such a good signal?

That bear didn’t bottom out until Sept ‘74, at about one-third lower levels…

Roam92, thanks for pointing that out!

The chart is incorrectly labeled. The Coppock guide turned up for Feb 1975 from -316 to -297 so it was a bit “late” but it usually is. Nevertheless it was a very good signal that the bear market had indeed ended.

I’ll fix the chart.

I don’t have the data, but I’d like to see how the Coppock compares to a monthly MACD. Does anyone have the data for that?

Great - glad we caught that… Since January is very likely to be another down month (four out of the past five for the SPX), does that mean that long-termers should continue their holiday for another month?

Or maybe another several months?

Market Outlook: The Golden Cross Won’t Happen for Months

http://seekingalpha.com/article/115024-market-outlook-the-golden-cross-won-t-happen-for-months

Marketwatch quotes Eliades on the Coppock:

…

Stock Market Cycles’ Peter Eliades is another Crash of 2008 star.

Uniquely, he continues to try to be cheerful. He wrote on Friday: “We are looking forward to next week’s market action and a possible resolution of a market bottom of some degree.”

See? That’s unique.

Not for the first time, Eliades is excited about a new technical tool I’ve never heard of, something called a “Coppock Curve”.

But whatever it is, it’s not yet applicable. Eliades concludes his discussion of the Coppock Curve: “We do not believe we have a buy signal to look forward to at the end of February because it would take a reading above 10,212 on the Dow to generate a buy signal at that time. That would entail a rally of almost 28% on the Dow over the next four weeks.”

Oh. That’s too bad.

Eliades is in cash, although, as he did last week, he’s trying a quick trade if conditions are right Monday morning.

If I’m not mistaken we just got a buy signal on the NASDAQ index at the end of April.