Tantalising, but is it real?

KIMMRUNNER

Member
Messages
58
Likes
2
Attached the results of a EURUSD trading scheme

It makes a single trade (most days) and holds it for 24 hours opening and closing at the same time.

It gives 2600 ticks since Jan 1 making 36 trades of which 26 were successful

Max drawdown 100

It needs tick data for the entire previous day to calculate the trade tho.

Can anyone give me historic tickdata for EURUSD over a couple of years period to verify that it is not an illusion ( I need min 10 minute samples) ?

You guessed it, it is one of my funny moving averages at work
 

Attachments

  • eurscheme.xls
    27.5 KB · Views: 684
KR - have you posted the system details elsewhere?

My interest in FX is fairly new and I may well hgave missed your previous post on it.
 
FX

My interest in FX is brand new too, so no I have not published the scheme before.

However I am adapting ideas for index trading that I have been developing for several years.

WHilst still working, the returns from these schemes have reduced with decreased volatility: so I am looking to FX as an alternative arena.

I would not have got interested in FX but for intraday data posted on this board, and for the apparent success of others in trading FX ( Thanks Johnny T ! the inspiration has been really valuable!!)

I am concerned by some of the schemes that I have seen posted, that imply a correlation between the movements during one part of the day with other parts of the day. Thats not to say the schemes cant work or dont work, just in my experience analysing intraday index movements , such correlations seem to happen only for short periods.

Maybe forex is different: but the old adage is that the market discounts all knowledge: that if it were known that a break out at one time of day gave a consistent gain in that direction or fading it: then serious money would plough into the market lifting it antacipation and nullifying the pattern: which is why some of these trends exist for a while but are then short lived

So most of the trading schemes I have devised I like to back test across several years of data & for that reason have tended to be basing trading decisions on published daily data.

I have spent many months analysing most of what is written in standard texts to discover that a lot of it is complete tosh - backtesting it proving no reliable gains. The phrase " assumptions repeated so often that they gain the status of fact " spring s to mind.

So I have discarded what is out there and ploughed my own furrow, to discover that some forms of weighted moving average seem to deliver the goods consistently: that the parket is say 60% predictable from history and 40% unpredictable on the basis of placing a trade every day.

I got at least one "yeuggh!" reaction to the suggestion that there is anything new to be discovered with moving averages.
All I can say is I am not averaging the stock price!! so it is a whole new ball game with a new set of rules.

Confining to more restricted conditions that occur infrequently you can get trading situations which are nearer to 100% successful (analogy to some candle patterns) , but only occur every (say) 50 days. Iam too impatient for that

So the schemes I have developed tend to use the daily placing pattern: but then use the infrequent patterns to remove situations of likely loss, rather than only trade situtaions of known gain hence nudge up the per trade gains, or % success with index trades

This FOREX algorithm is different. I have been trying to discover how much better a single trade per day FOREX system can be if it uses intraday as well as EOD figures to determine the trade.

The results as you see are startling.

But are they consistent over months or years? - or si this a freak?

One of the biggest questions I have and would like to get answered is what constitutes success? and I have tried to strike up a discussion of a unified set of rules to have a competition for the best scheme.

My thoughts are:

(a) It must only trade single point and the result must be normalised to the size of the instrument. In this case 2660 pips constitutes about 20% of the value of the underlying instrument at 1.2300 : - this allows FTSE and DOW schemes to compare like for like

Alternatively to normalise to capital employed - ie drawdown+ margin NTR: (in this case 2660 pips/ 100 drawdown + 100? NTR of EURUSD) so a factor of 10 to capital employed.

(b) It must trade a maximum of one open and close in any 24 hours. - or maybe relax this to never having more than one trade open.

(c) It must be automatic, and no data prior to the previous (say) 90 days must be used in the decision to trade on any day. (

Sorry, long term tech traders: but the problem is that if you are allowed to use longer history, then more or less a look up table can give perfect results on every day - but will never reproduce on data on which the system was not trained. I also personally seriously doubt the value of long term TA

Using rules like this the answer tohow good is the scheme is reduced to a simple number. The return. I would love to know how my schemes fare against others.

It certainly seems to behave a lot better than the results published on commercial schemes.

In as far as I can see the graph published is a money machine - if true? But is it true?

Only longer intraday data can ever answer that question.
 
Backdata

Hey KR,

EURUSD data ...

There are the daily files on http://www.forexite.com/free_forex_quotes/forex_history_arhiv.html but you have to import them individually into excel,

On http://eurostoxx50.co.uk/cgi-bin/yabb/YaBB.pl?board=FOREX;action=display;num=1063102047;start=30 someone's put together 2001 to Sept 2003 into one file, about the 3rd post on that page has the links to the actual files (I downloaded a few: 3-4Mb per currency pair), so you'd only be missing Sept to Dec, but that could probably be put together in one file download from http://freeserv.dukascopy.com/exp/ or http://www.fin-rus.com/analysis/export/default.asp

Hope that helps.
Hope it works.
Hope you post the details for the rest of us! :)

Cheers
3rddawn
 
Thanks..

Many thanks for the info 3rddawn - if it checks out Ill send you the calculation:

I downloaded from fin-rus but at the mo it isnt reconcililing with the 2004 data: I think there is a 3 hour timing difference, but also the numbers seem different from Johnny Ts too - only by 5 ticks, but that is enough to disturb things.

Still working at it
 
Hey KR,

I found that Jonny's data is US time ... 02:00 in his spreadsheet =07:00 GMT

I've been downloading the individual files from the forexite archive, and they are GMT+1hr. So I was having to take off 6 hours in total to get to the US time to simply add to JT's spreadsheet.

Have downloaded 16th March from forexite and fin-rus.
Forexite 12pm seems to equate to 2pm on fin-rus to me. Do you concur?
As Forexite is GMT+1hr, that means 12pm in their files = 11am, and yes, you take 3 hours off fin-rus to get gmt (It think :( )

As for the variation, yes I noticed it when looking too - it seems a little bit more volatile than the forexite data.

I can tell you that the forexite data is very very close to what I've been simulated trading on capitalspreads.com, day be day since the 27th Feb, so I trust that source.

I wonder how much difference it would make overall though ... wouldn't it be ok if you use the same source for all testing, because generally the ups and downs are still in the same places during the day?

If you're not happy and want to use the forexite, let me know and I'll be happy to help make up some large files with you. The eurostox file goes all the way to Sept 03, so we're only missing some of Sept 03 - end Dec. Shouldn't take too long - I do it at least once a day now anyway, and it IS about time I got round to writing a bit of VBA to do some of the clicking ;)

Let me know if you want some help :)

3rddawn
 
Forexite

How do you drive the forexite site? It is all in russian, and the link seems to give Canadian/US not eurusd?

The link on EUROSTOX only has data to last sep & the first thing I want to do is to confirm that the 2004 results agree with Johnny T data: so an idea of how to download a february day, or alternatively data for 2004 from forexite would be really appreciated

Thanks
 
KR

Which formats can you accept for the data? I have tick data back to December for Eur/USD in TradeStation, if it's any use.

Paul
 
Thanks

Cant do any specialised formats, (ie only csv )

But I was particularly interested in some forexite data for any part of 2004 for validation against Johnny T's on which I know the algorithm works:
 
Hi KR,,

Is this any use ?


Ian
 

Attachments

  • q1_600_500.xls
    31 KB · Views: 436
KIMMRUNNER,

If not averaging price, then what are you averaging ?

Also, on your spreadsheet, I take it the diff column is only used to calc P+L and is not in any way a parameter for the system decision making ? Is that correct ?

Cheers,

The General.
 
Forexite Data

Sorry, been out for a while!

Forexite data
If you click on the forexite archive link

http://www.forexite.com/free_forex_quotes/forex_history_arhiv.html

(it's the first of the links in the earlier post), you get a page with all dates listed since 3/1/2001 in three formats.

Click on the MetaStock format - it's just simple text files. Each file contains the 1min data for a number of currencies, eurusd included. Just import into excel using comma ... when specifying the fields, you can get excel to convert the date field directly to date data by choosing YMD (it will take 20010301 and make it into 03/01/01 in excel), put the time field in as text and let it import the data

Then you need a column heading row 1 if you didn't already import it, then do Alt D F F, which is Data, Filter, Autofilter, select EURUSD and you only see the eurusd data.

Then if your time column is C, put this in the one after the Close column to convert time text to time :-

=timevalue(left(C2,2)&":"&mid(C2,3,2))

Copy it down the column, paste values and formats back into col C and you have timevalues properly placed. Then select date, time and the 4 data fields and paste them where you want.

You should be able to compare with JT's fairly easily. If you want any help, send me a pm and I can do some manipulation for you.

Doing the above is how I added to his "Is this one sopt on" spreadsheet on a day by day basis.

You do have to remember that JT's data is all US eastern time (i.e. 5 hours behind GMT, and these files are 1 hour+, so take off 6 hours. If you need exactly every minute continuous over every 24 hour period, add the date and time fields together (to make a date/time field), then subtract 0.25 to get the date/time to swing back 6 hours. That should sync with Jonny's data.

I seem to remember doing a quick validation myself with this source and although I can't be sure (have slept since then!), I thought it was pretty similar. Anyway, if your system relies on ma's and stuff, even if everything is 20 pips up or down from jonny's data, as long as it's all 20 pips up or down, the differences would all cancel out. It's if the highs are 20 pips higher and the lows 20 pips lower that you might have trouble.

Also bear in mind that the forexite data even if it doesn't track perfectly with the spreadsheet data we got from JT, it DOES track almost exactly with Capitalspreads simulation account - I was within about 1 pip every day after taking off 3 for spread per trade, so we could try a simulation walk forward if it's promising.

If you need any more help you can pm me.
Don't bother doing anything since JT's 20th Feb - I have all that 1min every day since until last Friday and will post hopefully Tuesday evening at the latest - if you're putting data together, concentrate on the Sept to Dec period, and then we have a full 3 1/4 years. Should be enough!


Here's to the future! :)
Cheers
3rddawn
 
Last edited:
Forexite

Thanks 3rd D - just me being stupid - I opened the first of the files in that list , saw it contained USDCAD and thought that was all it contained.

Have already written the VB to read the files, offset the dates, and extract the day info I need.

Regret is time 4 bed , so testing the trading will have to wait.

I am signing up the legals leases and bankers stuff to buy another business tomorrow which will keep me occupied 4 a while, but hope to report by end of week.
 
Hi Fowkesp,

Is it possible to send me a copy of your TS data,

I am currently looking at a reversal strategy with
very tight stops , and would like to use futures data.

thanks Ian
 
Top