Hanover: an auto-trader using Recent Strength

On a much lighter note, pathumracled sent me this link today: http://www.forexfactory.com/showthread.php?t=307052

I have subscribed.

you're not the only one...

Hehe. Comforting to know.

No, between the current price and 19 ticks ago.

The concept of modulo groups is confusing at first. But once you get it, it's quite simple, really.

I've whipped up a small document that might help explain how this modulo stuff works. It's attached to this post.

Skunny eh? I read that thread from start to finish like a novel. Now I cringe when I think that I actually took it seriously.

Maybe the "Epsilon Megastar" gets a revival!? ;-)

"I hope you now have a better understanding of these concepts."

You bet I do. Fantastic again. Many, many thanks. This makes things clear.

I might have worked out the circular thingy eventually, but 'eventually' can be a long time in my case. And I would have written several thousand lines of code to achieve it.

Cheers

I am building the CBI module; should be ready later today.

Steve I had a chfjpy long trade end in small profit last night. But immediately Hanover opened a new long trade. I have min hours between trades set to 12. I'm using the D1 chart with a slope confirm of 1.

Just curious if the min hours check is working for you? That will tell me it's on my end.

No biggie as I am on demo.

Thanks!

Sandy

Steve I looked at the code for the min check. With cc = OrdersHistoryTotal() - 1; controling the loop, if you only have 1 order open wouldn't this resolve to CC = 1 -1 or 0? So the min hours check would never happen?

Thanks,

Sandy

Code snippet ----->

//MinHoursBetweenTrades check.
//Looks at the most recent closed trade in the history tab, and disallows a new trade if the old one closed too recently
if (MinHoursBetweenTrades > 0 && OrdersHistoryTotal() > 0)
{
for (cc = OrdersHistoryTotal() - 1; cc >= 0; cc--)
{
if (!OrderSelect(cc, SELECT_BY_POS, MODE_HISTORY) ) continue;
if (OrderSymbol() != Symbol() ) continue;
if (OrderMagicNumber() != MagicNumber) continue;
if (OrderCloseTime() == 0) continue;

double TargetTime = OrderCloseTime() + (MinHoursBetweenTrades * 60) * 60;
if (TimeCurrent() < TargetTime) return(false);
break;//No need to go further through the loop
}//for (int cc = OrdersTotal() - 1; cc >= 0; cc--)
}//if (MinHoursBetweenTrades > 0)


return(true);

No, it's a bloop. I will sort it out later.

Cool...thanks Steve

I must be missing some sort of scale factor. When I compare ATR(1) to ATR(60) the ATR(1) exceeds ATR(60) very often. See image. Please enlighten me how you got the image in your post.
Thanks
MC

The CBI module is born. Please see The Central Bank Intervention Module in post 1.

I'm playing around with the new CBI function, trying to scalp some quick pips when there's a move in the market (threshold 5, TP 10, SL 20). Pretty much a crash and burn experiment to see what's going on with the new module.

What I've noticed is that the EA opens several trades on one pair if it the conditions fill again. Could a Maxtrades input be added?

DIego

Hello all programmers, traders and sharp minds

I've talked to Steve about the possibility of creating a RecentStrength Oscillator (or other indicator) that can be used for visual backtesting of a specific currency pair.
I know that this presents a challenge since You have to hold several thousand RS values for just one weeks trading on lower timeframes. But maybe the new indi could reference the RS output file, do its calculations as time passes by and create a new file would hold its values as long as the chart is open.

What I think could be gained by this is a valuable knowledge about price reacts to the different strength and slope relations. And hopefully result in a more appropriate and specific settings for the Hanover filter-module to keep us out of bad trades, rather than going through months of forward testing on several different threshold, strength and slope settings. I think it could be a handy tool to determine the RS values needed as a trade validation for minimizing drawdawns and maximizing the profits on some of the EAs.
Maybe I'm totally on the wrong track here, or maybe You've done similar thoughts... Feedback is appreciated (and needed - I know).

I picture an indicator like this:
It has two currency parameters; ccy1 (base currency) and ccy2 (counter currency).
I references the RecentStrength output file to gather the latest strength readings for the chart timeframe, and calculates ccy1-ccy2 and plots this value as a histogram in a sub-window.
I figure that the indicator would have to write these values to another file for that currency pair, since it has to be able to plot at least one weeks M1 or M5 values (7200 or 1440 candles) to be of any use for the visual backtesting.
Maybe it can even be coded as a MTF oscillator - for example using M15 or M30 RS values on a M5 chart.

If for example ccy1 is EUR (current value eg. 22) and ccy2 is USD (current value is 90) then ccy1-ccy2= -68. If EUR is now strengthening and USD is weakening then the histogram will cross the 0-line soon and the histogram value will continue to rise.
If this is combined with a colorcoded slope confirmation then the visual backtesting would be even easier (4 different colors - one for below 0 and falling, below 0 and rising, above 0 and rising and above 0 and falling).

Unfortunately my programming abilities are very limited, so assistance is required. I know there are some very proficient programmers and traders in this forum, and hopefully some of You find this idea interesting and are able for find more ways to utilize the indicator - rather than just use it for visual backtesting and fine tune entry/exit in the EA development.

I haven't tested any of the EA's that are being worked on in this thread. But I've followed this thread with great interest and have started reading about the Spider EA. Let me know if I can help anyone in forward testing specific settings. I can run a few on a VPS.

Cheers
Tomas
___
What I'm trying to accomplish with the indicator idea presented above is best illustrated with the balance graph. The graph is from a backtest of last weeks trading on EURUSD, M5, with Avoitenkos published EA from the ATC 2010. If the first 13 trades could be filtered out by Hanover this EA would be a monster. Hopefully it's "just" a matter of finding the approximate RS values that will filter out the most loosing trades.
But maybe some of the other EA's in play are just as powerful?

Really? Gosh. Let me just check........ I believe I mentioned in my post..... Yes. I did.
No. This is not about scalping a few pips. If you think it is the you have not been reading what has been going on.

Along with quite a lot of people here. That annoys me. I am quite surprised people have not noticed this regrettable characteristic in my personality.

Be very careful what you post here tonight guys. I have spent a very frustrating day coding this stuff. What should have been easy turned out to be very difficult for me to achieve.

Anybody else ask a question as stupid as this one will get blasted to perdition. I am not in the mood for dealing with stupidity posted by idiots.

If you have nothing intelligent to say, stay quiet.

Very quiet.

Once again, great work Steve.

Scoobs

Nice one Steve.....Stay at it ......You're going to win.....

Thanks. Hehe. The guts came from Carsten. All I really did today was thrash around shredding spaghetti until some mystical combination of thingies I was trying finally decided I was not going to leave them alone until they worked.

If the stuff holds together reasonably well through NFP tomorrow, I shall isolate it all into a module that coders can use more easily.

For testing and on the grounds that you-never-know, I have Generic already loaded onto a GJ chart. You couldn't give your buddies at BOJ a call could you, and tell them we need a decent intervention for testing? They don't need to go the whole hog; a couple of hundred pips would do nicely.

Or maybe that nuclear sub captain pal of yours could accidentally land a nuke somewhere harmless? I have no idea where 'somewhere harmless' might be, but I would expect nuclear sub commanders to have this info at their fingertips. I expect an unexplained nuclear explosion would cause the odd moment of turmoil in the markets.

Damn. I just remembered. He only surfaces every six months. Hey ho.