Dow 96 point entry

[jpwone]

Hi mombasa

I'm not trading at the previous close. I'm trading at the previous close plus/minus the trigger factor. The trigger factor needs to be large enough to ensure that the opening gap has a very high probability of being within it. The core of the system is that the Dow should actually trade at or beyond the entry level which is the previous close plus/minus the trigger level.

This is in theory the beauty of the system - it gives you a direction following entry with good probability of success which is not affected by most opening gaps.

If the trigger is set too low then you will not get the entry or you will get the entry at price which is significantly different to the one you wanted.

Convince me

 

[quint]

Surely gap openings of over 75 points are very rare beasts indeed. By the time the index reaches Johns figure the SB price will be roughly in line. (Well, as roughly as they ever get). Or am I being an idiot?

 

[darrren]

John

I think the way to nail this question on stops is to physically look at intra day data, for each day a trade was triggered in your S/S, and see if the stop was triggered AFTER the entry was triggered, or took place BEFORE the entry?

I assume (as a non S/S person) that a S/S analysis would be no good for backtesting the entry/stop scenarios as it wouldn't have any knowledge of the times entries and stops or which order they took place in.

So, if the above is correct, rather than waiting 6 months while you forward test it, how about we get the intra day data for the last 4 years, split it up between several of us, and each analyse each day on an intra day graph to see the confirm which entries would or would not have been stopped.

What do you think?

Darren

The maximum negative could also be noted after entry for the rest of the day to help decide on a 'reasonable' stop for the system.

 

[Bigbusiness]

Good thinking darren. There are days when the dow does strange things that can't be seen in OHLC data.

I did start looking at the 1min data a while ago but haven't found the time to wade through it all. I have attached a spreadsheet that makes charts for each day if anyone wants to work on it. It's not great and I am sure someone with excel skills could do better.

 

[mombasa]

JP - I agree that in most cases a large enough entry trigger will negate the opening gap - the problem is that a high trigger will reduce the profits substantially - we don't seem to be in the kind of market where 200 point gains are the norm - as was the case until 3 months or so ago.

To maximise the profits, need a smaller trigger (based on weight'd averages (whatever we come up with).

Currently working on the futures ss to avoid triggering an entry if the gap is larger than the trigger - will post asap.

Cheers

 

[jpwone]

Hi Darren

my current plan is:

<b>1. Optimise the spreadsheet using EOD data.</b>
We now know that the longer term average for a positive or negative close is in the order of 94 points ie. over the period 9th June 1999 to 2nd Dec 2003 the average positive close has been plus 93.65 points and the average negative close has been minus 93.9 points. I intend to use this as the basis for the entry.

<b>2. Add dynamic entry</b>
Easy enough to add (even in a spreadsheet) but difficult to optimise as there is the problem of weghting and duration. How much weight do you give the more recent data? What period represents recent data?

<b>3. Test against intraday data</b>
We now have some 1 min data to test against. My plan is to database this and then convert to 5 or 10 min data (just to make it faster to process against). Run entries against this data and generate a meaningful report which will give me intraday whips amongst other things.

<b>4. Optimise stops</b>
Optimise stops based on intraday data.

The beauty of getting it into a database is I can then code the tests rather than us having to manually check each day against the relevant data.

John

PS. assuming all this works at what point do you think I should start charging for it

 

[quint]

John
I would make a strongly worded statement denouncing people who charge as frauds...then start charging.

 

[jpwone]

LOL

I really should have got the ideallist sentiment out a bit earlier though.

Maybe next time.

 

[darrren]

John

From you reply I cannot work out whether you are saying the intraday data can be used in a S/S to check you original work. This is probably because do not know the full abilities of a S/S. But I can't see how a S/S would be able to deal with the order throughout the day in which entries and stops took place. Without knowing the order in which they took place it would be impossible for man or machine to accurately know whether the EOD price had given a profit or a loss.

Darren

PS. In answer to your PS - If you wish to have a torrent of abuse heaped on you and have to flee to Spain, then go ahead. As a finishing touch, set up a website and then announce the charge without any warning! LOL

 

[robber]

Bigbusiness, thats a pretty impressive set of data. Ive opened it in word but can only see records to feb 2002. Is this a problem at my end or is that as far as the data goes?

BTW for anyone interested it opens as a 7789 page word document!!!!!!!


Neil

 

[esiotrot]

I go out for an hour or so and come back to find a mountain of posts here!!

A quick question on the data - referring to a post sometime ago - I am using the same EOD data as JPW but also wanted to test this on the futures.

JonnyT - what future data is in your Spreadsheet? I have Dow future for a round a year and it does not tie up with yours!! Not even close. Admittedly mine is the mini Dow, but it should tie up unless I've got hold of the wrong dataset. Where did you get yours? Am I safe in assuming I could just copy and paste yours to correct mine?

My brain hurts.

 

[Bigbusiness]

Bigbusiness, thats a pretty impressive set of data. Ive opened it in word but can only see records to feb 2002. Is this a problem at my end or is that as far as the data goes?

That's as far as it goes but it should keep you busy for a while.

 

[mombasa]

Futures ss attached with trade entries only when trigger + yesterdays close is within the opening price.

Note there r quite a few days when both sell & buy signals are indicated - probably high volatility days.

The April 2003 onwards period looks interesting now

 

[scpope]

Hi, I am unable to download BigBusiness's word data file. Has it been moved or is it a problem with the company firewall!

Rgds Stephen

 

[JonnyT]

But it's lost money this year !!!

 

[sidinuk]

JonnyT,

This system is currently in development and no final rules have been decided so your comments are rather premature and unhelpful. Given your experience in system development perhaps you could offer some advice as to how the system could be made more reliable.

Personally I think it's great to see a constructive thread rather than the usual slagging off that goes on. Even if the system does turn out to be unusable it will have given everyone involved some good ideas and methods to be used in developing their own systems.

 

[robber]

scpope, Its a BIG file, even with a high speed connection it takes time to download. Perhaps thats your problem.

Neil

 

[JonnyT]

The system needs to be done using the futures data. Daily cash spread bets are effectively the futures +- what is normally a fixed amount.

I have developed a brilliant system using the daily cash prices but it loses using the futures data so is actually untradeable in real life.

The described system also is a best case as it doesn't cater for intra-day swings.

Another problem is the points per trade at around 10-20 isn't brill when considering a 5 point spread + bias through a SB. It would be OK through direct futures.

However it has been in a very lengthy drawdown which to me gives a warning sign. Maybe I'm too picky.

JonnyT

 

[Glenn]

Just tried putting some constants in the G9 and G10 boxes instead of the 100 day averages (mechdow 1999-2003 futs data from Mombasa))
10 and 10 seem best so far.
Result is good (£1.3m or (edited) 28,956 points) and equity curve not bad.
Still room for improvement however.


Points edited because I left the 10,000 start position in.
Glenn

 

[Bigbusiness]

Unless the futures data is for the same opening times as the cash data, it might be hard to compare the two. Sometimes the futures have wild swings before the cash market opens. I have seen the futures up over 100 points, then bad news comes out at 1:30pm and the market opens lower. That would be hard to work out on a spreadsheet.