Developing Forex strategy - your thoughts?

[Artamon]

G'day all,

I'm currently playing around with an FX strategy. Without optimising it to the nth degree, it's currently at a stage where the annual return is 55% with a max sys DD of 18%. It's an AUDUSD hourly system, but my current hourly data only goes back to mid-2003. (I'd like someone who has at least 10yrs data to backtest this for me at a later stage)

I've included the Amibroker report and settings below. If you have a moment can you look at it and tell me what you think of the figures in the report. Some figures you'll see are way way out and not possibly correct (net risk adj return %, avg profit %, avg loss %) but this is because of the settings needed for the report to correctly reflect the 100:1 leverage I'm using for this test.

Any thoughts and comments would be greatly appreciated.

Andrew.

[kaveman]

just need the formula

[money tree]

55% is a poor return for such high leverage

positive carry alone with leverage of 20:1 would yield 50% p.a (and now is a great time to do it after a fall from 80c)

does your strategy involve utilising the EURUSD lead? what about the gold/usd lead?

any T.A system on forex using anything other than trendlines, support, resistance or Fibonacci JUST WONT WORK.

Central banks control currencies. They dont use RSI or stochastic. Those LAGGING indicators were designed for stocks, where investor psychology is being measured. There are NO investors in forex.

[Artamon]

Graham,


just need the formula
Yeah I purposely left the code out coz I wanted comment on the resultant figures, not so much the code that got me there. But, i'll post it later today.


55% is a poor return for such high leverage
positive carry alone with leverage of 20:1 would yield 50% p.a (and now is a great time to do it after a fall from 80c)
does your strategy involve utilising the EURUSD lead? what about the gold/usd lead?
any T.A system on forex using anything other than trendlines, support, resistance or Fibonacci JUST WONT WORK.
Central banks control currencies. They dont use RSI or stochastic. Those LAGGING indicators were designed for stocks, where investor psychology is being measured. There are NO investors in forex.

Money Tree,

My strategy does not involve any other currencies. I'm not using any indicators except for a simple MA from which I generate my entry signals to buy if it is an uptrend and sell if it is in a downtrend (amongst other things). But, I do understand that taking the lead from eur or gold can provide guidance in direction, and I do plan on looking into that at a later stage.

It's nowhere near complete or final. I'm yet to code in the scenario for shorting. Incidentally, the results above are based on a 1 lot purchase. Buying 2 lots in my example took the annual return to 90+% and the drawdown into the mid-20's.

What I'm trying to do in this exercise is get a decently profitable strategy based on 1 lot purchases. In my limited experience, if you can get a decent % profit from 1 lot with a 5k bank with a DD that won't have you sweating whenever you enter a position, then with good exits you should be able to trade the trends (both up and down). The 1 hour timeframe is something I picked as workbable for me.

I appreciate the comments are a more experienced head.

Andrew.

[mit]

Artarmon,

A good return without pyramiding and without a too bad a drawdown. Can you plot the equity curve? I noticed that it is a long only system so far and you want to be sure that you haven't just captured the long term trend. If the equity curve still goes up when the AUD is going down that is a good sign.

This is a problem in forex because shares markets tend to go up as the economies grow but forex can trend for extended periods in either direction.

I had a lot of trouble with this with my system (GBPUSD) and didn't trust it until I could trade both directions equally profitably with the same system. 168 is a reasonable number of trades. I'd optimize each year to see how much your parameters move.

MIT

[Artamon]

Hi MIT,



This is a problem in forex because shares markets tend to go up as the economies grow but forex can trend for extended periods in either direction.
I had a lot of trouble with this with my system (GBPUSD) and didn't trust it until I could trade both directions equally profitably with the same system. 168 is a reasonable number of trades. I'd optimize each year to see how much your parameters move.
MIT

Shorting is what I'm having trouble with at the moment :banghead: I'm still hacking away, but nothing I've coded for the short side has worked out. If I can nut this out, theoretically, I can always be in a trade no matter the direction, which is my end goal.

Andrew.

[Milk Man]

Shorting is what I'm having trouble with at the moment :banghead: I'm still hacking away, but nothing I've coded for the short side has worked out. If I can nut this out, theoretically, I can always be in a trade no matter the direction, which is my end goal.

Andrew.

Ditto to that. I'm scanning for ideas on a short side system for stocks so i'll keep you posted. I don't even know what Forex is fully though :o so i'm unsure how much that will help.

[mit]

I've never found a mechanical shorting strategy for stocks. It annoyed me in Guppy's books when he says to short you simply turn the charts upside down. Shares definitely move down in a different way to how they move up.
I have an idea though to do with something I have noticed with ex-dividend stocks.

Forex is different though and is more symmetrical. The problem is when your system gets caught in a general trend. I have a FOREX system now but it took me a long time to find one that was robust enough to trade. I agree with Money Tree most people who make money on FOREX do it using some version of S/R and Fibs which is hard to backtest automatically and I don't have the patience or time at the moment to forward test on a demo account.

keep at it

MIT

[Artamon]

G'day all,

Here's the hourly fx long system. It's only basic at the moment. Haven't made a lot of progress with the short side, but it'll come. Those that have hourly fx data, can you also backtest this code and post your results please. I tested using 5k capital and a SL of 35. Play around with the parameters.

Thanks,
Andrew.

[Artamon]

G'day all,

Has anyone (with hourly historical data) backtested this? I'm curious to see how it performs.

Thanks,
Andrew.

[mit]

I managed to replicate your results. Here's the equity curve.

MIT

[mit]

One thing, to make your coding easier Amibroker has a defined variable TickSize. You can set this up in settings or on a per symbol basis. Also my graph is based on a 4 pip spread.

MIT

[money tree]

am I the only one concerned that the results are:

1. based on only longs
2. during a test period where the market went up and up

??

[Artamon]

I managed to replicate your results. Here's the equity curve.
MIT

Thanks MIT. The hard part is still on the table though - taking short trades.

Andrew.

[mit]

MT

I agree. However, one good point is that the equity curve does not slavishly follow the AUD price action so there is something there.
Another problem is that you have to watch the market around the clock for your buys and sells. I have heard of people who do this. Start Monday morning and trade until they have made their weekly quota. Pretty tough, you would want to get a good return on your money.

MIT

[Artamon]

am I the only one concerned that the results are:
1. based on only longs
2. during a test period where the market went up and up
??

Not at all, money tree. These are concerns of mine as well. But, I have stated in previous posts that I am looking at taking short trades as well - just have to come up with a suitable strategy.

As for test period, it's all i have available at the moment. Looking to rectify that though.

As results stand now, what are your thoughts?

Andrew.

[money tree]

Im a little confused why there would need to be a different strategy for shorting.

I mean, with any forex pair you are always short one currency and long another. Those who trade forex would know that USDCHF is nearly inversely identical to EURUSD. Therefore, any strategy that works for long EURUSD must also work for short USDCHF.

As I said, the results are not adequate. A buy and hold strategy (positive carry) would yield the same results with less risk, and almost zero management.

For all the labour required you are not getting any extra return.

[Artamon]

G'day all,

I made a bit of progress tonight. I want to be able to go long AND short within the same currency pair (alternately). On it's own, taking short trades gave an annual return of 38% and a max sys DD of 28%. Nothing too flash, I agree. However, when I combined both sets of code into the one strategy, backtesting (my limited data) gives an annual return of 82% with a max sys DD of 24%.

Andrew.

[Beachie41]

I backtested over last five years on non-indicitive data. Allowed 3 pip spread. With optimization (SL 0.0017 - PS 0.0050 I managed 656.96 pips for that period. Max Sys drawdown was -5.44% from 341 trades with 29.91% winners. Allowing a trailing SL of 0.0025 dropped profit to less then 300 pips.

[porkpie324]

Forex trading,been there and lost, you're playing with the big boys here, best advice i can give is get out of it, after being sucessful with stocks i guess it's a natural progression, but i kept being stopped out, you need deep pockets for forex not luck. porkpie :(

[monoply]

Agreed. I trade stocks ok, but I always lose on forex. I find it a very tough market.

[mit]

I think the reason is that in the period we all backtest stocks tend to go up so any reasonable system will tend to show profits. Forex is not so of course as it deals with the difference between two countries. So systems that backtest well are more difficult to find, similar to trying to find a mechanical system for shorting stocks. NOT impossible but difficult.

MIT

[Milk Man]

Ive been devoping a formula and it seems to work extremely well. My question is of the practicality of it. It uses a trailing stop of maximum 3% or 2 x ATR. I only have EOD data at the moment though so I cant tell if the trades will be executed at that price. I am assuming there are automatic trailing stop facilities available and no guaranteed stops (as with CFD's). How much slippage am I likely to encounter, on average, between the stop and the next quoted price down?

[Smurf1976]

There are forex brokers that will guarantee your stops. So if you are long AUD/USD (for example) and set your stop at 0.7498 then that's exactly what you will get. It won't slip to 0.7495 or something like that. The broker I use guarantees stops although there is a limit on the value of the trade where they will guarantee the stop ($2 million I think?).

If you trade longer term (as opposed to day trading) and are trading forwards (futures) then there will be slight movement between your spot data and the forward contract but it shouldn't be enough to cause a problem under normal circumstances.

I always set my stops 4 pips above/below the support/resistance level or whatever other trigger point I am using. So if I was short AUD/USD (example only, this is not a present trade) and the previous high (on whatever timeframe I am looking at) was 0.7600 then I would set the stop at 0.7604 to overcome any movement between the forward contract price and my spot price data. Of course the actual price you enter would differ from the calculated stop due to the spot/forwards spread. Doing this stops a lot of false triggering of the stop too but of course it doesn't suit short term trading systems where you are trying to gain 10 pips per trade.

Yes, I took account of this when doing back testing and also I assume a "worst case" buy and sell scenario. So when going long I assumed that I opened the trade at the highest price on that day and closed it at the lowest. Vice versa when shorting. And remember to take into account the changing spread between the forward and spot price over time since that is effectively a cost to you while the trade is running. Doing it this way means that your actual results should exceed those of your testing since you tested under worst case conditions with entry and exits. :)

[Milk Man]

Thanks heaps smurf.
:bier:
You trade currency futures instead of forex? If so why?

[It's Snake Pliskin]

Thanks heaps smurf.
:bier:
You trade currency futures instead of forex? If so why?

Is that milk in those glasses? :D

[Smurf1976]

Thanks heaps smurf.
:bier:
You trade currency futures instead of forex? If so why?
In short, I trade forwards (as opposed to day trading) using a forex broker. The reason being quite simply cost - holding open a day trade for an extended period means paying fees each day (with this broker at least) whereas the only such cost with forwards is the slowly narrowing gap between the forward and the spot price.

For example, the spot price of pair xyz/abc might be .9900 and the six month forward for a long position might be .9960. Over six months that gap will slowly disappear which represents a cost to the trader but it is less of a cost (and doesn't involve any cash flow issues) than paying the daily fee for a rolling day trade. That said, I typically close my positions well before six months although the occasional one runs for quite a while.

The other reason is simplicity in the accounts. If the only thing I have to worry about is whether each trade made a profit or not and don't have to worry about paying any fees to keep positions open etc. then it makes the book work a lot easier. And making that easier gives me more time to spend on the actual trading rather than accounting for it.

I have never traded commodity futures etc. so I really couldn't say if what I'm doing with forex works the same way or not.

[Milk Man]

Is that milk in those glasses? :D

Nup, snake bites. Tobasco and charteuse....