8 pips per hour

[FetteredChinos]

I always thought Off the Wall was Michael Jackson's best album...


..anyway, its a new year, time for another completely random strategy, that is probably untradeable manually, but could be an interesting basis for an automated system..

it is essentially a scalping system, but with a difference.

It Trades the EUR/USD currency pair. Will probably work on other pairs too.

assuming a 3 point spread, since jan 2002, it could have generated approximately 76,000 points..

it is counter-trend, so drawdowns can be a problem, but i will try to work how to code stops into it to try to reduce these.

essentially the rules are:-

at the close of each hourly bar set up 2 orders: one to sell at the close + 10 pips, the other to buy at close - 10 pips.

at the end of each hour, cancel any orders that were not hit, and put in new orders. leaving open positions to run.

then come 9pm GMT friday (2300 on my spreadsheet i believe) close all open positions at the market price.

thus several positons are likely to be open. i will try to work out how to calculate the maximum exposure in due course..

It trades 24 hours a day 5 days per week. If left to be automated fully, and with a pretty sizeable capital base to trade off, retirement could only be a short step away..

I will attach the usual details in due course, but because i am working off hourly data, the spreadsheet is completely huge, and probably too large for sharky to deal with..

FC

 

[FetteredChinos]

here is the equity chart...

 

[trendie]

its seems breakouts are the way to go !!

£trader, SIBKIS, and the above are all range breakouts.
The only things that change are time-frames ( and consequently numbers of trades ), and stop-losses.

Effkay on the "Would you trade a coin-toss" is also essentially a range-breakout !

But then, a breakout is just the start of trendiness, isnt it ?

The other thing I have noticed, is that profit-targets seem popular.
It certainly takes the pressure off when you bank the money.

Good going FC - keep it up.

PS: Where on earth can I get historic intra-day data for FX ??

 

[FetteredChinos]

stats from 1/1/03 to 5/11/04 ( i think). not great, if you consider the Wapping Drawdown..

Points 115954
Trades 10505
Wins 7056
Win % 67.2%
Pts per Trade 11.04
Max Win 374
Max Loss -360
Drawdown 8562
Profit Factor 1.35
Pts after spreads 84439

 

[FetteredChinos]

trendie, drop me a PM if you want.

i can send you some via email, if your inbox can take it. though you can always use the hourly data about to be posted on here..

fc

 

[FetteredChinos]

and here is the sheet.

as usual F9 does the recalc..

FC

 

[FetteredChinos]

hmmm, using a 5 pip trigger results in this..

Points 100367
Trades 10969
Wins 9941
Win % 90.6%
Pts per Trade 9.15
Max Win 331
Max Loss -350
Drawdown 3679
Profit Factor 2.04
Pts after spreads 67460


but are these results real? is it possible to automate such a system?

incidentally, slippage shouldnt be an issue as we are going counter-trend, so if anything it should improve things slightly.

FC

 

[TWI]

its seems breakouts are the way to go !!

at the close of each hourly bar set up 2 orders: one to sell at the close + 10 pips, the other to buy at close - 10 pips.

I think this is essentially a range trading system. As SIBKIS etc. go with break this goes against it.

 

[FetteredChinos]

well, mechanical systems are all the range nowadays..

 

[JTrader]

Hi

FC

is there no form of a stop-loss on each trade?

Thanks

jtrader.

 

[FetteredChinos]

not as yet..JTrader. not as yet..

part of my recent systems work has lead me not to view each trade individually, but rather as a bunch of trades.

therefore the individual loss on a particular trade isnt too detrimental to the greater good as it were.

christ, i sound like TV.


in short. No.

its only the beginning of an idea. just wondering whether anyone with Tradestation or similar could verify the results. and then also test it using stops..

exchange of ideas and wotnot.

FC

 

[cd173]

Hi FC

Interesting idea...but how much equity would you need and what sort of time range would you look to trade??

Have you tried this sort of thing on any futures markets?

Cheers
C

 

[FetteredChinos]

not tried it live, no.

the idea came from something i was looking at from the SP500 markets, and fading out-of-hours moves on it.

transfers pretty well to FX by the looks of it.

i am guessing that it would need a gihungous (technical term) capital pot to trade it effectively.

as regards to time range to trade, i did a pivot table to see if there were any periods during which the system fell over consistently, and from what i can see, the profits are spread fairly evenly over the course of the day.

hence the desire to run it on autopilot on the back burner.

FC

 

[cd173]

Cheers FC

I take it you cancel the order that does not get filled on the next bar?

Had a quick look at the little Stoxx50 and at a glance it looks quite promising, providing you can get the stop loss right....have you tried a trail stop type system??

IMHO I think you'd need an account of around £50K min to start....or am I being greedy

Plenty of food for thought :idea:
C

 

[FetteredChinos]

yup, you cancel any orders not filled on that bar. should be simple enough to code to automate, i would have thought.

but i have no experience in that matter.. is it possible to forward test in tradestation/metastock etc?

fc

 

[TWI]

FC I quickly coded this and had a look at the results, they did not verify your own, I am sorry to say ,however I made certain, possibly incorrect assumptions in rules as follows:

Each time a buy or sell limit is hit only one contract is traded at that price irrespective of how many times the price may get there in any given hour. For example, if you come into an hour with 10 long positions from previous hours and a sell level is hit you then end up net 9 long. The only other order that can then trade during that hour is the Buy at close of last order - 10ticks. i.e. only 2 trades per hour max.

Other option was to completely close open position and end up 1 long or short total at each signal but this was a really bad idea for staying solvent.

I attach my equity curve and EL code here for any others who may want to tinker:

{****************************
Name: FC 8pips per Hour
Written By: Tony Walker
*****************************}

{Close at 2100GMT on Friday}
If dayofweek(date) = 5 and time >= 2000 then begin
Buy to cover ("WeekXS") next bar on open;
Sell ("WeekXL") next bar on open;
end
else begin

{If flat buy or sell 1 contract at next signal}
if marketposition = 0 then begin
Buy ("B")1 contract next bar at C - (10/pricescale) on limit;
Sell short ("S") 1 contract next bar at C + (10/pricescale) on limit;
end;

{If long then buy 1 additional lot at next buy signal or close one long at next sell signal}
if marketposition >= 1 then begin
Buy ("+B") 1 contract next bar at C - (10/pricescale) on limit;
Sell ("+S") 1 contract total next bar at C + (10/pricescale) on limit;
end;

{If short then sell 1 additional lot at next short signal or close one short at next buy signal}
if marketposition <= -1 then begin
Buy to cover("-B") 1 contract total next bar at C - (10/pricescale) on limit;
Sell Short("-S") 1 contract next bar at C + (10/pricescale) on limit;
end;
end;

 

[FetteredChinos]

cheers TW, much appreciated. seemed too good to be true. your efforts are appreciated.

perhaps my data is hugely wonky?

might try this on index futures data to see if a similar thing works there.


thanks again,

FC

 

[FetteredChinos]

ok TW, ive got another one for you to **** on my chips...


instead of setting orders based on hourly close.

just fade the previous 4 hours action. ie if the last 4 hours movement is up, then go short, else go long. close all open positions at 9pm friday

looks like about 50k points before charges..

could you verify this one? ive knocked up a quick sheet in the attached file.

FC

 

[juanbyte]

Hi FC TW

I have had a play with the original data and managed to achieve $1,166,545 profit over 2 years.

It seems the reason why it is working is that you have to have multiple contracts running, unfortunatly this means that tha account required in my case was $376K.

This means that it only produces 300% ROC

Most methods I have looked at will have increased profits on multiple contract enties.

Looking at your most recent spreadsheet I noted that for one week there was net 15 short positions, you need a large account to trade that.

It does work if you can afford it.

Keep up the good work

 

[FetteredChinos]

thanks juan. im just tinkering with ideas etc.

if there is one thing that all this has taught me is that being limited to opening and closing one position per day is largely fruitless. holding for several days, owing to the whippiness of markets seems to produce much better returns. i know it goes against all i believe in - holding onto a position in case it turns round, but with a large pot, it seems a fairly solid method...

more tinkering to come.

FC