Lyxor Msci Em Discussion Threads

[davemake] 21/8/2003 10:31

Update HBOS OHLC 20/08/03 727, 732, 711, 717


4 day : Long trade at 728 - (still open)

9 day : Long trade at 721 - (still open)

16 day : Long trade at 714 - (still open)

25 day : no action

[davemake] 20/8/2003 06:21

Update HBOS OHLC 19/08/03 753.5, 754, 724, 729

4 day : Long trade at 728 - (still open)

9 day : no action

16 day : no action

25 day : no action

[davemake] 19/8/2003 10:24

Today is Day 3 of the 4 day projection. Days 1 & 2 did offer limited opportunities (see above) for Short entries, both of which gave a small profit - at £10 a point potentially £70 - £90.

However, a stop-loss money management system is needed.

The longer term projections (9, 16, 25 Days) have not been triggered.

Intraday High/Low todays = 754 - 741, current 745. This is a continuation of the downside from the potential short position taken out yesterday.

It may be worthwhile back testing an Index - UKX or T1X

[davemake] 18/8/2003 16:48

Update HBOS OHLC 18/08/03 757, 757, 746, 747

4 day : possible short trade at 757 - (9-11pts profit excluding spread)

9 day : no action

16 day : no action

25 day : no action

[davemake] 16/8/2003 10:46

Update HBOS OHLC 15/08/03 744.5, 760, 742, 750.

4 day : possible short trade at 758-760 - (8-10pts profit excluding spread)

9 day : no action

16 day : no action

25 day : no action

[clem] 14/8/2003 23:36

Jeroo. Thanks. Its rewarding writing original material but a bit nerve wracking. :)

Goatbreath. There are lots of various levels of randomness. There is not simply random or non random. If you consider white noise as a random signal and say 50hz signal you get down the mains as non random, the signal coming down you mains electricity contains two components, a random noise plus the non random 50hz and its all mixed together.

If you were to digitise this signal it would change randomly but only by the range of the white noise glued into the 50hz signal. The signal would not be totally random and you could probably put borders around the randoness by examining the time series.

A moving average would quickly smooth out the noise.

The way randomness works in a time series, is the high and low of the signal over time expands by the square root of the time base. Thats just how it is mathematically.

Random Walk remains a good model for the market and while it is not the full picture it is a powerful one.

Mr Potato Head. You are spot on with my thinking. Its frustrating sometimes that the only time I get these days for this kind of work is when the markets are closed.... Duh!!!! :) Happily ADVFN is more fun than trading.

[davemake] 14/8/2003 17:36

In addition here are the forecast projections for HBOS over 4, 9, 16 and 25 days

HBOS closed today at 745, High = 750, Low = 736

Figure

4 days go LONG at 729, SHORT at 757 t/r=14 x 2

9 days go LONG at 722, SHORT at 764 t/r=14 x 3

16 days go LONG at 715, SHORT at 771 t/r=14 x 4

25 days go LONG at 708, SHORT at 778 t/r=14 x 5

(Figures are rounded)

Next week I'm on holiday for a while and I'll be back after the 25 day projection. I'm really interested on the outcome of these projections

Looking forward to the results.

[mr potato head] 14/8/2003 16:35

impressive stuff so far ...predicted max range from 4 days ago 991.62 ..todays high 991.75 currently 987 ish ....if thats the high for the day then I am impressed !

[mr potato head] 14/8/2003 16:13

we are probing the top of the 4 day range now.....theory is that this should contain the move up ....

[clem] 14/8/2003 13:21

Goatbreath. Well T/A is the analysis of time series so it would be by definition. :)

Firstly you have to get an idea of the randomness of the instrument you trade. You can crunch a time series and get this number out. I use wavelets, but you could look at the beta, or get the fractal dimension or even run a compression on it to get a value for that.

Once you can gauge the level of random you can use that in whatever trading method you have.

Let me give you a converse example.

In the old days the bucket shop trading dens had a great system to scalp punters. It wins because of randomness.

I give you 50 times margin in VOD, say which is £1. ie 2p margin. You can not top up your margin.

If it goes to 98p you are out and lose the contract and you 2p stake.
Because of the random volatility only 1 time in 50 will it never go down to that 98p and take all your money. When it does not go down to 98p and insead goes straight up, you on average take 20p out

So because of the random component as the bucket shop, I take 2p *50 = 100p and you win 20p. Thats a fabulous return for giving you what seems to be a big favour with a wonderful gearing deal.

All because the natural random shake in the market is beyond what your margin allows for.

This is a good example of the effect of random on trading.

[luthor] 14/8/2003 09:01

Davemake,
I think I've missed something in your calculation in the heading of the thread.
When u calculated the range (26xsq.rt of 4), this seems to have the dimension of TIME. Should the number of days ( in this case 4) be normalised with something like year, for example.

Lex.

[analyst] 13/8/2003 09:58

It does however have statistical probabilities for the degree of expected variation. i.e. you can be 95% certain that the index will lie between x and y over timeframe z. The main problem of assuming a 'normal' distribution is the underestimation of the probability of very large moves in a short time.

[goatbreath] 13/8/2003 09:31

I think that's my question, Mr PH. Does randomness have boundaries? By definition I would have thought not.

[goatbreath] 12/8/2003 20:21

If you can quantify the randomness surely it is no longer random, and all you are doing is another form of TA?

[mr potato head] 12/8/2003 19:40

we are banging on 987 SPX following rate announcement allowing a little latitude might make this the top for the moment ....random walk say these boots were made for walking ...right back in the opposite direction

[mr potato head] 12/8/2003 14:45

Just been doing some fiddling with this using the daily S&P data I keep...at first glance it seemed to me that trading signals using even a short interval i.e 2 days walk were going to be few and far between....but a little closer investigation showed up some interesting things
one was that this is a function of range i.e a 5 point range today is going to give a much tighter random walk range over the next (n) days than if we had say a 22 point range therefore tight range days are more likely to lead to tradeable signals in the following days ...this makes sense with other work like the NR7 days used by Linda Raschke
based on fridays data 8/08/03 SPX hi = 980.57 lo =973.83 range = 6.74
median =977.2 random = range(6.74) * sq rt 2 (1.414) =9.53 = projected max=987 projected min=968..so over the next 2 days from friday the 8/08 the share price should not exceed 987 or drop below 968 ...987 is resistance from other sources too so it will be interesting to see if it has any significance .

[mad4it] 04/8/2003 21:19

Elaborate trading systems ? Indices ? What are they then ?

Buy small caps ( until you shouldn't )...it's the sound of one hand clapping.

[clem] 04/8/2003 20:39

Bonsai. Reality may be fluid, markets may be liquid but money is definately a solid. :)

[bonsai] 04/8/2003 13:06

et suivi

Reality is perception.
Perceptions change.
Reality is fluid.