Scmesgaccetfgbx Discussion Threads

[doctorbird] 27/3/2001 22:17

I read somewhere that the Coppock is best applied to markets which rise and fall evenly over a regular period, and in my experience this seems to be true also in the case of individual shares. So I should think in the case of Beazer it may simply be an excess of volatility and a generally bearish drift in recent times. I have found the Coppock helpful for example in trading EBQ where the price movement in the last eighteen months or so following a sizeable fall, whilst being erratic in the shorter term has been generally fairly smooth.

A start has been made with some back testing, firstly with the FT30 where this indicator goes back to the beginning of the nineties on my software. It's hard work because there are so many variables when measuring the performance of anything. I have made no allowance for trading costs. I hope you won't mind me giving the following definitions, you will know them but others may not. Official buy signal = The indicator is falling and then rises when in negative territory. Unofficial buy signal = Ditto in positive territory. Early buy signal = As the indicator crosses the baseline falling from positive to negative territory. Sell signal = The indicator is rising in positive territory and then falls.
1 HOLDING THE INDEX Buy the index 07.09.92@1733.4, hold until 22.03.01@3024.6, +74.49% over 103months = CAGR 6.7%

2 EARLY BUY SIGNAL Three buys in three equal tranches 07.09.92@1733.4, 18.01.95@2338.1, 03.07.00@3748.9, all held until 22.03.01@3024.6, +74.49%, +29.36%, -19.32% over 103months = CAGR 4.56%

3 OFFICIAL BUY SIGNAL All as above but buys 03.11.92@2007.2, 03.05.95@2482.7, 17.01.01@3536.3, +50.69%, +21.83%, -14.47% over 101months = CAGR 3.47%

4 EARLY BUY SIGNAL AND UNOFFICIAL BUY SIGNALS AND SELL SIGNALS
+38.34% in 14m, +4.61% in2m, +20.23% in 16m, +32.14% in 15m, +0.18% in 6m, and -19.32% in 9m, cash held 41m, total 103months = CAGR 9.19%

5 ALL BUY AND SELL SIGNALS
+19.47% in 12m, +4.61% in 2m, +13.22% in 13m, +32.14% in 15m, +0.18% in 6m, -14.47% in 2m, cash held 51m, total 101m = CAGR 7.87%

I should add that calculating in months is not as accurate as days but this whole exercise cannot be precise anyway. For example I have allowed a net return of 4% for periods of cash holding. If trading costs were allowed for in examples 4 and 5 the compound annual growth rate would be reduced to say 7.0% and 5.4% respectively. No doubt this level of return will seem very poor to those who have only been investing until recently in good times but of course the results from individual shares should be better.

Good trading.

[indieman] 19/3/2001 21:24

Tonight I ar mainly watchin' Jamie Oliver cookin' chicken. In the interim, though, I have looked at your data and compared it to mine. Your formula seems to give earlier signals than the IC calculation. Sometimes this helps, sometimes not. Why it gave such bad results for Beazer, I've no idea. Some of my own indicators also work badly for this stock for no obvious reason.

The results for Rolls Royce look reasonable but mean you would have missed far the biggest rise (186p to 300p) between Autumn 98 and Summer 99. The buy signal on 10/7/00 came a few weeks before a 30% price decline, but the later buy signal looks good at the moment.

The end of May sell signal for Dixons came at precisely the point you should have bought for the run up to 146p 3 months later. This set up a head and shoulders formation which failed and broke upwards from below 140p to 180p. The next buy signal (30/10/98) was good and coincided with the general market
upturn turn.

The sell signal which followed (20/9/00, 285p) lost out on a rise to 400p. This, in turn was the second of a series of up/down moves between 240p and 350-400p which were very easy to follow and profit from. I bought at 252p and sold at 346p 9 weeks later and I'm a low risk non-trader type of investor. The jury is still out on the last buy signal.

The Fleming Japan signals look good, but still leave you with a 14 month period losing money on the 1/98 buy signal on the IC Coppock.


Date FlemingJap


Price Coppock
02/01 241.64 -45.6
01/01 249.27 23.1
12/00 276.11 105.7
11/00 296.83 203.7
10/00 317.65 307.4
09/00 360.30 403.5
08/00 334.03 495.9
07/00 358.58 593.4
06/00 368.58 687.7
05/00 353.15 750.8
04/00 362.96 805.4 Sell
03/00 396.66 852.1
02/00 412.87 836.2
01/00 397.68 764.9
12/99 402.60 683.9
11/99 360.72 570.2
10/99 294.10 471.1
09/99 288.24 369.1
08/99 254.26 259.2
07/99 248.69 183.6
06/99 208.24 116.9
05/99 191.48 55.4
04/99 191.54 7.8
03/99 164.11 -46.7 Break even from 1/98 purchase.
02/99 146.06 -78.4
01/99 144.13 -93.0
12/98 139.30 -112.3
11/98 139.67 -133.8 Buy
10/98 119.48 -158.9
09/98 120.06 -155.7
08/98 131.41 -150.8
07/98 142.87 -145.3
06/98 130.42 -141.7
05/98 137.66 -128.8
04/98 138.46 -120.4
03/98 147.59 -122.1
02/98 157.45 -129.4
01/98 150.07 -144.8 Buy
12/97 143.05 -155.7
11/97 140.53 -154.4
10/97 168.50 -147.6
09/97 174.39 -147.5
08/97 191.70 -145.5

It seems a bit of a curate's egg, working well sometimes. I think I'll stick to using it for the indices, especially as my computer can't cope with a one cell calculation for daily coppocks.

Let me know how you get on with the alternative buy signal I suggested for the indices.

Regards,

Indie

[indieman] 19/3/2001 17:43

Thanks for the quick response. I've printed these off and will compare them with my data. Since mine only extend back to early to mid '97, it can only be a partial comparison.

If you have data going back to 20 years, try using Coppock going down through zero as a buy indicator, (for All Share & FT30 indices).

Regards,
Indie

[doctorbird] 19/3/2001 17:22

DIXONS GROUP

Ref. Date....Price..Coppock..+/-%..Days..%annualised
BT1 09.06.93@ 54.75 (37.6)
SD1 31.12.93@ 73.00 46.1 29.33 206 51.97
BT2 20.12.94@ 46.50 (135.0)
SD2 31.05.96@ 128.25 671.3 171.81 528 118.77
BT3 30.10.98@ 163.40 (23.8)
SD3 20.09.99@ 285.50 719.5 70.73 326 79.19
BT4 01.03.01@ 257.00(145.8)


FLEMING JAPAN IT

BT1 16.10.95@ 241.00 ( 79.5)
SD1 05.07.96@ 266.00 99.0 6.37 263 8.84
BT2 17.12.97@ 151.00 (176.5)
BT3 07.09.98@ 116.00 (181.0)
SD2 20.03.00@ 393.00 931.4 156.27 826 69.05
SD3 20.03.00@ 393.00 931.4 234.79 561 152.76

[doctorbird] 19/3/2001 16:49

ROLLS ROYCE Individual percentages reduced by 4 to cover trading costs.

Ref. Date.....Price..Coppock +/-% Days %annualised
BT1 01.10.91@ 150.50 (135.5)
BT2 28.11.91@ 127.00 (134.0)
BT3 11.05.93@ 134.00 ( 92.0)
SD1 28.04.94@ 199.00 236.5 28.23 941 10.95
SD2 28.04.94@ 199.00 236.5 52.69 883 21.78
SD3 28.04.94@ 199.00 236.5 44.51 354 45.89
BT4 08.11.95@ 161.50 ( 35.8)
SD4 25.11.96@ 247.00 227.0 48.94 383 46.64
BT5 09.02.98@ 204.00 ( 27.0)
SD5 06.07.98@ 264.00 034.5 25.41 149 62.25
BT6 10.07.00@ 235.00 ( 91.7)
BT7 05.10.00@ 172.00 ( 99.0)


Beazer

BT1 11.09.97@ 176.00 ( 22.2)
BT2 05.02.98@ 195.00 ( 43.0)
SD1 02.07.98@ 180.00 56.7 ( 1.73) 295 ( 2.14)
SD2 02.07.98@ 180.00 56.7 (11.69) 149 ( 28.64)
BT3 24.03.99@ 195.00 ( 44.0)
BT4 21.05.99@ 210.50 ( 38.6)
SD3 16.09.99@ 149.00 28.4 (27.59) 177 ( 56.90)
SD4 16.09.99@ 149.00 28.4 (33.22) 119 (101.89)
BT5 04.08.00@ 117.00 (215.5)
Sell signal possible soon.

[indieman] 19/3/2001 16:40

My calculations gave the following;-

Bear in mind the headers may be off-set to the left but will be in the right order

Date Db 1yr 1yr
IC Coppock 12m 12m
Rolls Royce 12m 11m
Price
201.91 02/01 -56.0 -86.3 -69.8
201.23 01/01 -67.7 -71.3 -102.3 -83.9
196.14 12/00 -72.3 -83.3 -107.7 -89.6
191.17 11/00 -84.2 -86.6 -120.6 -101.7
173.44 10/00 -90.8 -98.7 -127.6 -108.7
174.21 09/00 -90.2 -98.4 -125.1 -107.4
223.30 08/00 -81.3 -94.5 -110.2 -96.6
234.38 07/00 -83.8 -88.2 -107.4 -96.3
245.45 06/00 -84.1 -92 -103.9 -94.1
240.79 05/00 -81.4 -90.5 -98.1 -90.0
221.99 04/00 -74.8 -86.3 -88.9 -81.5
195.52 03/00 -57.5 -78.2 -67.5 -62.7
205.93 02/00 -33.7 -56.6 -38.2 -35.9
225.00 01/00 -12.1 -32.4 -9.8 -12.3

I've got the full Coppock bumf from Robert Anstead at the IC and have talked with him on several occasions when discrepancies have arisen. The last was when a respected UK analyst gave an incorrect buy signal for the Dow Coppock last month. I have also dissected the Coppock and it's failings and have modified it to better suit today's conditions.

Whatever Synergy are doing, minor discrepancies of a few days in taking monthly averages are unlikely to explain the differences. The Coppocks generated using just last day of month data are almost identical to those using monthly averages.

Still flummoxed,

Indie

[doctorbird] 19/3/2001 16:09

The brackets were intended to denote negatives. I'm surprised you can't get near the values because all of the seven variations that I have fed in, come up with more or less similar chart patterns and inasmuch as this is a very long term indicator the differences are not that important. I have applied the 21d,12m,12 to the shares you listed and I hope the figures will be comprehensible. There are a couple of things to say which I hope do not state the obvious. Unlike many of the oscillators the Coppock requires very little interpretation, the curve normally being a smooth one there are few difficult times when there is any doubt as to whether the stance should be buy or sell. On the other hand the drawbacks are that being long term, response is at times very slow, also there will be longish periods when one is entirely out of the market. This assumes that sell signals are used rather than the original intention only to use buy signals. Buy signals are below the line, sell signals above. Another thing to watch out for is that being weighted the chart line is constantly reforming in its recent history and this can be considerable within the month. For example the BZR chart line after a steep climb is evening out this month. However by the time the month is complete, which for this(Synergy) indicator is 27.03.01, the line may have keeled over thus giving a sell signal. Which all goes to confirm that hindsight is easiest. I'll post the figures separately in case I have a problem setting them out.

[indieman] 18/3/2001 06:46

I have looked at your Rolls Royce Coppock values and can't reproduce them by any minor changes to the formula. I've tried using 11 and 12 months for the last time differential and 10 months for the first time differential and none get anywhere near your values. None of them represent an acceptable indicator for Rolls' share price performance either. I've assumed by the way that your values should have been negative.

I'm flummoxed, but have to watch the Malaysian GP right now,

Regards,
Indie

[indieman] 17/3/2001 13:14

DB,
It was the last 12 in the 21d, 12m, 12 that I was unsure of. You mention (above) the use of 10 months as the time differential. This is Coppock's and ICs requirement, but could Synergy be using 12 months instead. Other indicators (including two of mine) use this differential and this would reduce the number of calculations they need to perform if they supply other indicators as well. I'll post some numbers for comparison later,

Regards,

Indie

[salar] 17/3/2001 11:13

Sorry - article posted to wrong thread and now edited out!

[doctorbird] 17/3/2001 00:03

Indieman,
A good idea to start this. Let's get it straight, I am neither a doctor nor a bird, sorry if that disappoints. My name is a real word, used in parts of the West Indies to describe what we call the hummingbird. These birds are delightful to watch, their wingbeats are apparently around fifty to the second, and I like this local name for it.
I shall get back to you with Coppock.

Good trading.

[doctorbird] 16/3/2001 23:09

This afternoon shot a fax off to Synergy requesting formulae. By closer inspection of the charts I have established that with regard to the 21d,12m,12, it does indeed mean a 21day average as you had supposed and I presume the 12 mths to be the period covered for percentage change and 10 to be ten month weighted moving total. With a variation to the number of days in the average it will probably be fairly meaningless trying to make comparisons. Fiddling with the parameters on this software can produce very colourful charts but in the end I find with other indicators that I usually return to the permutations that most people use. In this case the closest match I have been able to get to your Rolls Royce figures was 22d,12.10. The number of trading days in a month is somewhere between 21 and 22 and the extra day seems to make a fair difference.Just to give you a sample of the plot for Rolls Royce at 21d,12m,12, they are : 29.01.01(71.3), 28.12.00(83.8),30.11.00(86.6),31.10.00(98.7),03.10.00(98.4),04.09.00(94.5),04.08.00(88.2),06.07.00(92.0),07.06.00(90.5),09.05.00(86.3),10.04.00(78.2),10.03.00(56.6),10.02.00(32.4). That will do for now, let me know if you need more. When I get a moment I shall take a look at the other shares you listed and see if the Coppock would have told me anything.

Good trading.

[indieman] 16/3/2001 13:31

Well that seems to have confused almost everyone. I'll keep an eye out for your response on the Coppocks,

Regards,

Indie

[bird] 16/3/2001 12:45

Indieman------Is this post addressed to me?

If it is sorry for being dense...do you want me to check out long term buy sigs on above stocks?

regards

M.

[goatbreath] 16/3/2001 07:17

I used to get cop off signals in night clubs when I was a lad. Sadly,nearly every one of them turned out to be false.

[indieman] 16/3/2001 07:04

PRS, Bird-

I know the Coppock is aimed at indices, but Doctorbird has been using it as an aid to normal TA indicators. She (presumably she anyway) and I have generated different values for Coppocks and we're trying to work out why.

We are both aware of the origins, rationale, history and differing formulae involved in the calculation of Coppocks. I calculate my own values using a spreadsheet and Doctorbird uses proprietary software.

In the absence of any other way of contacting each other (security, privacy etc) we're having to try to sort it out here. Hopefully we can find another way and avoid the public arena.

Apologies again for hogging space but anything which can help investment decisions is worth exploring. Now, where's my astrology book; ah Abbey National is a Pisces I see.....

Regards,

Indie

[peterreidsmith] 15/3/2001 22:35

!? Coppock is for markets not individual shares.
It makes no claims for individual shares.
It has a good record on markets.

[taximania] 15/3/2001 22:34

Apologies accepted.

[indieman] 15/3/2001 22:02

I've looked at Coppocks using the Investors Chronicle formula with several shares.

Looking at Rolls Royce, Abbey National, British Airways, Dixons, Beazer and Fleming Japan (IT), I didn't find Coppock useful. In the case of Fleming Japan, one buy signal was so far out it took 14 months to break even again. What's going on?

Could you possibly run the stocks through your system and see what you think?
I would be interested to see examples of it working successfully and to know which formulae your proprietary software uses.

Apologies to those who find this incomprehensible.