Idea for EA - brainstorming required

[JAGfx]

You may have already read this as I previously posted it in the wrong sub forum. I've just transferred all the info over to here. So here it is again:

Hey guys,

I'm always coming up with new ideas on how to trade, especially when it comes to EAs. I'm not a Programmer so I normally have to get my gun Programmer to build the EAs for me. There is no problems with the build, it is normally the idea there is a problem with. I have a habit of having the EA built and then modifying it several times to get it right. Had mixed success to date with these EAs, but not quite happy overall as I would like to reduce my risk somewhat, and hence this thread.

I've got a few ideas bouncing around in my melon at the moment and I want to convert them into a new EA. The members in here seem to be a tad smarter than the average forex forum dwellers, so I would appreciate any feedback or ideas I can get. I'm pretty thick skinned, so if you want to tell me it is a rubbish idea, I can handle that also. Any EA I do have built, I am more than happy to share with other members at no cost, especially with those that contributed.

So let's get into the main idea itself. Probably nothing new here, but it may have a slight twist. I have heard of grid trading systems where there is new trades placed at certain intervals and baskets of trades being closed out when certain profit targets are hit etc. These are pretty common. I'm thinking of a grid here also, but instead of being based on price, it would be based on an actual trading signal to buy or sell. Now I generally use two custom Empty4 indicators, and someone was kind enough to build a simple indicator that places buy/sell arrows on the chart when these 2x indicators are in sync with each other. That just makes it way more visual and easier to spot the entries.

This will only work if you can hedge trades, which means buy and sell at the same time. I am an Aussie so this is not a problem, but it maybe for my US friends. What I was looking at, was something like the 1hr chart and from the start of the week, it takes every buy and sell signal generated by these 2x indicators. All of these trades would be with the same position size, whether long or short. The plan is then to close all of these trades at the end of the week, and hopefully end up with an overall profit, or worst case, a small loss.

I have done some manual backtesting on the AUD/USD 1hr, just picking random weeks. The results seem pretty good with some very good weeks at +400 pips, most around 80-100 pips and the odd small losing week. On average, more winning weeks than losing weeks.

As you can see, this is a pretty broad idea at the moment and will require some refinement. A few things I am considering are:

* Why go from week to week.
* Why not have a set profit target eg 100 pips, and once this is achieved, close all trades for the week, or start a new sequence.
* Use of a filter eg only take short trades if price is below a MA etc, or in sync with a higher timeframe using the same indicators. Unlimited options here.
* Even trading times nominated, especially if trading the smaller timeframes ie 5m charts.
* Safety stop eg when the basket of trades go -100 overall, close all trades and start a new sequence.

There are probably more considerations I'm sure. I will post some screen shots and might even do up a YouTube video showing this idea. Any feedback or ideas would be welcome and appreciated. Cheers. Jim

[JAGfx]

Here is a screenshot of the AUD/USD 1hr chart. I just picked a random 4 week period and did a manual backtest. It is a bit squashed up for my liking as I wanted to get the 4 weeks in the shot. The red and green arrows are the buy and sell signals based on my 2x indicators. These don't repaint and the trade is opened on the open of the following candle. The orange vertical lines are the week blocks. All trades were closed on the very last candle closing price. Spreads weren't taken into consideration either. But it does show the potential just on this basic idea. So just over 400 pips for the month. Cheers. Jim

[JAGfx]

Just playing around with some charts using the same idea as above, but this time used the EUR/JPY 4hr chart with the exact same settings on the indicators. Obviously less trades overall, with about 3-4 trades each week. Did a quick manual backtest over an 8 week period with the following weekly results : +168, +130, -165, +232, +82, +161, -44 and -117. Overall +447 pips for 2 months. Not spectacular, but it is consistent. Without really checking out other pairs to date, I would assume that they would have similar type of results. Now if you were running 4 or more at once, then the results start to look up. More testing required. Cheers. Jim

[Jemook]

Hi JAG,

I've deleted the other thread for you. You can ask us to move a thread if you like in future

Regards,
Jeremy

[JAGfx]

After a bit more manual back testing, I have decided to go ahead and have an EA built by my gun Programmer Christina from Wise-EA.

I was toying with the idea of a profit target and decided on going with two options. One, I just open a new trade on every new signal and close all at a nominated time at the end of the week as previously discussed.

Secondly, have a set target on a sequence of trades, whether this be one, three, five etc trades. For example if I had a profit target of +20, and this was hit on the very first signal, then that trade would be closed and the EA would wait for the next trade signal. Obviously if this target wasn't hit on this trade, then it wouldn't be possible for the target to be hit on the 2nd trade due to the hedging effect, so I would have to wait for at least a 3rd trade to have any chance of the target being hit, and so on. This entire sequence would then be closed when an overall profit of +20 is achieved.

Then I guess there is always the 3rd option where I can just manage every individual trade. This would be much easier on something like the 4hr charts which is my preferred timeframe anyway. I can also add some sort of filter like an MA where I would only take trades in one direction as a result of that filter.

Now looking forward to getting my hands on the first version and checking it out. I'll keep you posted with my results, and if it turns out to be a dog, I've probably wasted everyone's time. Shouldn't be the case as I am basically after the EA just to enter each trade on each new signal anyway, and the other stuff is just a bonus if it works out. Cheers. Jim.

[JAGfx]

Well I have had the EA built and it does everything I have mentioned above. There were a couple of small glitches and I had a couple of mods done, but all good now. I'm probably not using it so much based on my original idea where it just takes every trade for the week and closes them all out at a certain time on Friday. I have done a bit of backtesting on this strategy and it does do okay, but I think I can do better. As I tend to concentrate more on the 4hr charts, there is ample opportunity to manually manage the trades when required.

The EA does work of two custom indicators only, and certain settings do give some pretty good buy and sell signals. Not perfect, but good enough to catch most major trends. I have also had a higher timeframe filter added using these same two indicators. For example I could trade the 4hr chart and use the daily chart for trend direction, so if the daily was heading up, then the EA would only take longs on the 4hr chart.

So I can either leave all trades open and close them all at the end of the week, or I can nominate an overall profit for those trades, or I can control each one individually with a stop, break even point and/or trailling stop. So plenty of options and plenty of testing to do.

I have attached the two indicators you need for this EA and also the handy signal indicator that puts the buy and sell arrows on your chart. I have also attached the EA itself for you to have a play with. It is pretty straightforward, if you don't want to use a higher timeframe filter just put '-1' in that field, and if you wanted to use the daily chart, put '1440' in as it is based on minutes. For target, that is in pips and all times are based on your broker's platform time. If you have any questions about it, feel free to ask away here. I'll post some backtesting settings and results just to give you an idea how I use it shortly. Love this stuff!! Cheers. Jim

[JAGfx]

Here are the settings and results for a quick backtest done on Go Trader. Keep in mind, different brokers have different closing times for their 4hr candles. Also backtesting is not the most reliable indication of how well an EA would work.

This was on the EUR/AUD 4hr chart from the 1st Jan 2011 until now, starting with a balance of $10,000 and trading with 0.1 lot on every trade. I also used pretty aggressive settings to give me plenty of trades. I'll run through the settings in order they are on the EA.

Fast Platinum 3
Slow Platinum 6
Smooth Platinum 2
SF 1
RSI Period 8
WP 3
Filter Time Frame 1440
FTF Fast Platinum 18
FTF Slow Platinum 39
FTF Smooth Platinum 13
FTF SF 2
FTF RSI Period 12
FTF WP 5
Slippage 2
Stop Loss 150
Open Lot 0.1
Basket TPPip 0
Use BE True
BE Point 150
Use Trailing Stop True
Trailing Stop 150
Trailing Step 1
Friday Close Time (leave it blank)

Now the stop loss, B/E and T/stop figures I just plucked out of mid air after eye balling the chart only. There was nil optimization testing on these at all.

The results weren't spectacular, but better than a bank and pretty good considering this was fully automated and nil change in position size. Drawdown was probably a little high for my liking, but the chart wasn't bad looking at all.

Total net profit 10,628.71
Profit factor 2.11
Absolute drawdown 546.97
Total trades 166
Max drawdown 3473.19 (23.77%)
Profitable trades overall 57.23%
Average profit trade 212.93
Average loss trade 135.20

There was also a little concern over the number of consecutive losses in a row as I thought it was high at 13, which may cause some traders grief. But this was just a quick test with some random figures just to give you an idea of what I am looking at, and that you can run the exact same test on your platform and compare. If I do come up with some spectacular settings, I will pass them on. Same goes for anyone else that does play with this EA, I would love to hear about your results or comments. Thanks. Jim

[RedLineFred]

Hi Jim,

We meet again, exchanged previously on another forum.

I will try this out and get back to you.

As for ideas, have you considered keeping the trades open and trailing your profit? With multiple orders open, if you track total profit and then look for a drop below a close out level from ProfitMax, this should keep you in profit longer and greater.

Cheers, Fred

[JAGfx]

Hey Fred,

It's funny how we all bounce around forums. Most of the forums are pretty useless, but every now and then you do come across some good ideas. I try to keep a lookout just for something that maybe a little left field that I can incorporate into my own trading.

This idea for this EA originally started with what exactly you mentioned, where it would just open a new trade on every signal, and then I would manually manage those trades. This is obviously very hard to backtest, but as I normally trade off the 4hr charts, they make it easy to keep an eye on. I also trade multiple pairs, so even if I am taking a loss on one or two pairs, I normally find the others are in profit. I have no problems closing all trades and starting afresh if things get out of hand. Why sit on long term losers I ask. You mention ProfitMax. Not sure what this is. Could you please give me more info. Thanks. Jim

[RedLineFred]

ProfitMax - my shorthand for maximum profit.
The idea being to continually monitor how high the unrealised profit moves and then stop out all trades if profit pulls back from maximum by a set amount or percentage. The idea is to stay in the market as long as profit keeps building but then close all trades as the basket profit starts to drop away.

I've been playing around with this idea, but cant come up with a reliable method of calculating what the pull back should be. It is very dependent on the number of trades open, and the number of lots in the basket and how high the profit value gets. A 10% pull back when you're up $50 is only $5 but 10% on $2k is another story. Also with more lots in play it takes less price movement to swing the profit.

Anyway will keep working on it.