Steve Hopwood's auto-trader for David486's H4 trendline scalping system

Hi Steve,

Just a quick update.

I loaded the ea on a demo last night on 24 pairs. I have set it to open 0.02 lots, tp 50.

I am using mptm to set BE+2 at +25 and close half position.

So far the account is up $27.78

Great start.

Just because you *can* use it on lower timeframes doesn't mean you *should*.

The EA works on lower timeframes in the sense that it creates trendlines and trades them. The system itself may not be profitable on the lower timeframes, however.

If you're using a fixed TP as specified in david486's thread (50 pips), you'll have to change the TP to a smaller value for a lower timeframe. Even then the system may not be profitable.

Hi Steve

Great work on this bot. I think this has a lot of potential.

I have a quick question for you about the USDCAD trade you can see in the image attached. (This is on H4 on an FXDD demo account)

As you can see, there was a third touch, but then for some reason which I do not understand, the trade was not entered until the top of the candle. I would have thought that the trade should have triggered at or very near the trendline, but the EA waited until price was a ways North of that before it entered the trade. The distance is much to far to explain with slippage.

In the manual trading thread by David, he says:

"Look for a bullish trendline on the 4hr timeframe. After the second leg retracement buy when the price hits the line again."

So I was expecting to see the buy at or near the trend line. But in this failed USDCAD trade, and in the successful EURGBP trade as well, actually, the trade was taken quite a distance from the trendline.

Here is something else odd:

 

1. Experts tab - USDCAD trade buy at 14:06 (broker time)
2. Journal tab - USDCAD trade buy at 14:06 (broker time)
3. Account history tab - USDCAD trade buy opened at 17:05 (broker time)

Please take a look at the attached image USDCAD trade 2. This would indicate that the trade was actually entered at 17:05, as it quickly hit SL (at 17:26), and if it really had been entered at 14:06 (the red vertical line), then it would definitely have ended in profit (or at least BE) and not been a losing trade for sure.

So, it seems that somehow the EA tried to place this trade at 14:06 as an instant trade. However, the price it tried to open the trade at was 1.06460. This price is the price that the order ended up opening at (according to the account history) at 17:05. I suppose that from 14:06 to 17:05 it must have kept on hitting the server trying to open a trade at that price, but was only successful at 17:05 - although the Journal contradicts that theory by saying the instant order was accepted at that price at 14:06. The price at 14:06 (the close of that M1 candle) was 1.05800 - a great place to enter.

I wonder why it did not open an instant buy at 14:06 for 1.05800, but it instead tried to open an instant buy at 1.064600 at that time and this was not filled until 17:05? This is very strange to me: Right time, wrong price.

Or perhaps I have not understood the logic of the system correctly, although I really had thought I had "gotten it".

Can't make anything out from your picture.

Coo. Mine hasn't even graduated to its own dedicated demo yet. I am impressed.

Uh oh. I hope I'm not in trouble with Mr. Hopwood....

How should I know what was going on in the deepest, darkest recesses of your platform? I cannot even cope with my own.

Hi Steve

So when the bot takes a trade on your platform, it does so right at the touch of the trendline and not far away from it as you can see in the images I attached to that other post?

Invisible

Not sure. I have been a tad busy elsewhere today. The attached looks ok.

We can start sorting out the details next week.

Hard to see, but I think it might look a few pixels late. Worth keeping an eye on. How about your Journal and Experts tabs as compared with the account history tab? Do these all show the trade taken at the same time, or is there a differential like I spotted with mine?

Idea for safer entry. Feel free to throw this out if you think it is rubbish.

Use the H4 for the trendline drawing as usual. However, as soon as the price touches the H4 trendline for the third time, then drop down to the M15. Wait until the M15 closes, and then place a pending trade X (user defined at 0 or greater) pips beyond the high (for buy) or low (for sell) of that M15 candle. What this should ensure is that if price is just going to crash on through the trendline, the trade is not triggered. Price actually has to be bouncing back to set the trade off. Of course, if the price is already near where the pending trade would be entered anyway (in other words it is already bouncing back) an instant market order can be taken instead).

What do you think?

I like this very much. Today, I have to get the 1.1.3.3 thingy working properly in tandem with the initial trading - you will have read about my struggles with NB. After that, I shall start thinking about how to code this idea.

The latest version is in post 1.

 

1. I have tidied up some of the feedback. What people were seeing on screen was not always an accurate reflection of what the robot was up to. Amongst other things, this explains that there is no trend line because the bot calculates the market as ranging, and therefore cannot trade.
2. 1.1.3.3 recovery is working. I am not sure if this is going to be helpful after all; we shall see.
3. The bot waits until the market has moved slightly in the right direction after touching the trendline.
4. I have removed all the mptm-type management features, for simplicity whilst we develop the robot. I have not removed any code; I have commented it out so that I can re-instate it later if necessary.

Great work, Steve. Look forward to testing this version next week. Thanks!

Hi Steve

Just a quick additional detail to my idea above: Since we are entering a pending trade, the question comes as to when to cancel it. Of course, a crude method would be to use a timed cancelation. However, I think it would be best to have an additional user-defined parameter "OvershootPips". This would be the maximum number of pips that price is allowed to go beyond the trendline before the EA decides that it is not bouncing and so the pending trade should be canceled.

Obviously, the perfect bounce in theory would not even go one pip beyond the trendline, but literally just touch it and then reverse. However, in practice we may want to give it a little leeway, hence the "OvershootPips" factor. I would think that something like 5 would be reasonable, but anything over 10 would be too great. Something that we can experiment with anyway.

The point being that we don't want the pending trade to be hanging around for ages after the signal has long since passed.

I also have an idea for a complementary robot / addition to this robot which would take advantage of trades that fail the third bounce, which I will write up later.

Cheers

Invisible

Sounds like A Plan to me.

Steve

This is an idea for safe entries when the third touch fails.

What we are looking for is a break of the trendline and then a failed retrace move towards it followed by a strong continuation move in the direction of the original trendline break.

So, if you take a look at the image attached it will give you a clearer idea.

 

1. So, first of all, you are on the H4 as with the bot in this thread, and there are two touches. (Or there could be more than two, I guess it does not really matter. But there has to be at least 2.)
2. Price then hits the trendline again, but this time passes through it.
3. Now drop down to the M15.
4. Around point B the 3 period RSI dips below 20 (or perhaps it is already as it passes through the trendline). In any case, cutting the trendline and RSI3 being under 20 is the precondition for this sell (ie at least two bounces on the trendline, then trendline is cut and RSI3 on the M15 is below 20).
5. Now, price then reverses back towards the trendline.
6. However, now what was support is now resistance.
7. As price retraces towards the trendline, RSI3 gradually rises.
8. If RSI3 >= 50 (somewhere around point C on this diagram) and price is still below the trendline, the EA enters a pending sell at the previous low (as you can see).
9. The SL is the level of the trendline at the time the pending sell is entered.
10. The TP is the same distance in pips as the difference between point A (the last high above the trendline) and the location of the pending sell (which was the low after that last high) minus the SL in pips. This is probably most easy to see on the diagram. (This is based on Elliott Wave theory of 2 pushes against the previous trend.)
11. The pending trade would be canceled for one of two reasons: (1) The price retraces back to the trendline (which we need to leave in place in order to check this). A cross of the original trendline will automatically invalidate the retrace and so the trade is canceled. (2) in case price wanders for a while between the pending trade and the trendline, we want to cancel the pending trade if nothing has happened for a while, so we need to have a time-based cancellation also. Since we are trading the M15, I think a time limit of 4 hours would be more than enough.

The weaknesses of this system would be in the case that price simply dives through and never retraces, or in the case that the retrace is so weak that the RSI3 never rises to 50 to allow the pending trade to be placed. However, it is my understanding that one of the safest trades is after a failed retrace after a trendline break, and this is what this logic is aimed at capturing. Also, it should have a good R:R ratio as in general the SL is likely to be smaller than the TP.

I have tried to think how this would be coded and (as a non-programmer) I think the above should be enough to find these trades, along with the trendline logic of the current robot.

What do you think Steve? Obviously you want to focus on the current 3 bounce bot, but perhaps something like this could be a good companion to that one?

Steve,

Help me out please. I am looking for the code snippet that will cancel the current trendline and trigger the generation of a new one when the current trendline is violated without a trade being opened. (Please don't shout.)

G

Trades can be missed if the market just kisses the trend line and the robot cannot send the trade for any of the gazillion reasons that the crap mt platform throws up.

I have put the trade-send in a loop; it will try 20 times with a 5 second break in between each attempt, before finally giving up.

Update in post 1.

Invisible, I am looking at your idea. I am going to code Obnoxious' first as it is easier, but yours also makes a lot of sense.

Great ideas here folks. Keep them coming.

Probably the easiest thing is to simply move the trend line. You can also delete it and let the robot draw a fresh one.