Old Dog with New Tricks | Page 28 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

Old Dog's Taming of the Beast? 34 replies

Old Dog Fib Trading - EA Development 18 replies

Old Dog Fib Trading - EURUSD 399 replies

5 Tricks of Trend Trading, by Rob Booker 25 replies

How To Work Around A Market Maker's Tricks 0 replies

Attachments: Old Dog with New Tricks
Exit Attachments

Old Dog with New Tricks

  • #541
  • Jul 24, 2010 6:32pm Jul 24, 2010 6:32pm
Quoting lionsden
Sorry Ron,

I was just replying with one of your previous quotes earlier in the thread....Next time I'll use a link or quote....LOL!

L
Damn you, I was getting excited.
  • #542
  • Jul 25, 2010 1:52am Jul 25, 2010 1:52am
Ron,

I think that you can reduce the consumption of resources by using a DLL that will make the calculations based on the data loaded directly by the DLL and not the MT4. I've seen that if you load too much data, the resources are minimizing. So, I think that it's not the EA's fault. The MT4 is forced to load more and more data, by the EA.
Today I BUY, tomorrow I SELL!
  • #543
  • Jul 25, 2010 10:09am Jul 25, 2010 10:09am
Would it make a difference in resource usage if I can shrink the chart and instead access the HST files?
  • #544
  • Jul 25, 2010 12:34pm Jul 25, 2010 12:34pm
Quoting getting

rfking21; what's your perference on broker 4 or 5 digits?
Sorry for the delay Getting, with so many post its been hard to keep up.

It really depends on how much money you have in my opinion. 4 digit (market maker) will allow you to trade micro lots and the high volatility of the market is sometimes smoothed but the spreads are much higher. A 5 digit (ENC) doesn't use a desk so they don't allow micro lots (banks don't like dealing with them), they experience the true volatility of the market, but the spreads are much smaller.

If you are limited to trading micros then stick with a four digit. But if you have the cash to trade mini and need the extra pips (like when scalping) use a 5 digit.

Hope this helps
  • #545
  • Jul 25, 2010 4:32pm Jul 25, 2010 4:32pm
Quoting Ronald Raygun
Would it make a difference in resource usage if I can shrink the chart and instead access the HST files?
I don't think so, because the HST files will be accessed by functions interpreted by the terminal.
Today I BUY, tomorrow I SELL!
  • #546
  • Jul 25, 2010 4:36pm Jul 25, 2010 4:36pm
Hey Ron the EA just opened some trades and I noticed something funny. I have it set for MaxTrades=10 so I expect to see 10 trades when it triggers. But this past week the 10 trades were somewhat spread out over a small range of prices. Today all 10 trades opened simultaneously, all at exactly the same price. Is this normal?
You are in a maze of twisty little passages, all alike.
  • #547
  • Jul 25, 2010 5:19pm Jul 25, 2010 5:19pm
Quoting stevegee58
Hey Ron the EA just opened some trades and I noticed something funny. I have it set for MaxTrades=10 so I expect to see 10 trades when it triggers. But this past week the 10 trades were somewhat spread out over a small range of prices. Today all 10 trades opened simultaneously, all at exactly the same price. Is this normal?
Should be at most 1 trade per bar...

Can you go into the experts tab and e-mail me the data from 30 mins before the first trade and 30 mins after the last trade? (RonaldRaygunForex(AT)gmail(dot)com
  • #548
  • Jul 26, 2010 5:52am Jul 26, 2010 5:52am
Hi to all,

I've started to read this thread last week and had no time to have results to post in here. Anyhow I've noticed that were some difficulties in parameters/program setup.
Anyone can post is/her results from last week or so.

Thanks,

Pedro
  • #549
  • Jul 27, 2010 4:04am Jul 27, 2010 4:04am
Great work ronald i was reading to some off your stuff and it made me wonder if you need some help with programming my friend is very well
capebal of that so drop me a line i think would be no problem and will
take off some work off your shoulders maybey

next i come from amsterdam the netherlands and i am busy with a new website it,s all about trading and robots so we will keep in touch

greetings marc
Amsterdam Holland http://cdn.forexfactory.com/images/icons/icon14.gifhttp://cdn.forexfactory.com/images/icons/icon14.gifhttp://cdn.forexfactory.com/images/icons/icon14.gif
email [email protected]
  • #550
  • Jul 27, 2010 11:45am Jul 27, 2010 11:45am
Quoting mactrader9
Great work ronald i was reading to some off your stuff and it made me wonder if you need some help with programming my friend is very well
capebal of that so drop me a line i think would be no problem and will
take off some work off your shoulders maybey

next i come from amsterdam the netherlands and i am busy with a new website it,s all about trading and robots so we will keep in touch

greetings marc
Amsterdam Holland http://cdn.forexfactory.com/images/icons/icon14.gifhttp://cdn.forexfactory.com/images/icons/icon14.gifhttp://cdn.forexfactory.com/images/icons/icon14.gif
email...
The help would be most appreciated Marc. Do you think your friend can find a way to make this EA less resource-intensive? A number of folks here on FF are having trouble it seems with running multiple instances of the EA.
  • #551
  • Jul 27, 2010 12:44pm Jul 27, 2010 12:44pm
Quoting Ronald Raygun
The help would be most appreciated Marc. Do you think your friend can find a way to make this EA less resource-intensive? A number of folks here on FF are having trouble it seems with running multiple instances of the EA.
Maybe if you make the backtest write the info to a csv file and continue writing with each tick like with DB. Then don't let it erase the file. That way mt4 wont have to test the full history every boot, the info is already saved.
  • #552
  • Jul 27, 2010 12:49pm Jul 27, 2010 12:49pm
Apparently each instance of the EA uses a lot of RAM. When you run too many instances the processor starts to thrash the disk.
You are in a maze of twisty little passages, all alike.
  • #553
  • Jul 27, 2010 4:36pm Jul 27, 2010 4:36pm
Is it possible to offload the backtest /optimization and decison making part to an windows exe? At the begining a script can be run to generate the m1 bar data in a csv which can be fed to the exe for processing. Further tick count can be input into the exe through dde. A script can be run on mt4 which shall act as a order execution server. This server shall recieve the command from the exe through a text file or a windows named pipe. Single exe can be multithreaded to handle different pairs. This exe solution may be particularly helpful for brokers who have their own api such as FXCM.

I shall admit that I have not read the whole thread yet. Please don't take offense if this idea has been floated already. Ronald, your EA is awesome.

Thanks,
pillager.
  • #554
  • Jul 27, 2010 8:17pm Jul 27, 2010 8:17pm
Quoting pillager
Is it possible to offload the backtest /optimization and decison making part to an windows exe? At the begining a script can be run to generate the m1 bar data in a csv which can be fed to the exe for processing. Further tick count can be input into the exe through dde. A script can be run on mt4 which shall act as a order execution server. This server shall recieve the command from the exe through a text file or a windows named pipe. Single exe can be multithreaded to handle different pairs. This exe solution may be particularly helpful for brokers...
I'm in Orlando, FL at Walt Disney World at the moment, and I can't get to my trading platforms or programming computer at the moment (be back August 1), so this is off the top of my head.

The short answer:
Is it possible? Yes. By me? No

The long answer:
Developing an exe significantly increases the complexity of the EA and increases the security risks of people using the EA. I try to keep the code as much as possibe in mql so I can easily post the code online for others to double check and audit--as about half a dozen of you already have. The other major concern I have is liability. Say someone manages to corrupt the .exe file and ends up stealing account information from users. I effectively become the one to take the blame for that, and I'd rather not have to worry about that possibility. I have freely posted the code under a FOSS style license and I invite anyone who can improve or expand the code to come forward and do so. When put in the right hands, this EA is incredibly powerful and can relatively easily double an account within a week.

[rant]

With great power however comes great responsibility. About two dozen people have contacted me off-thread via PMs, e-mail, my support ticket, and one greedy bastard who hounded my phone number before I permanently blocked him, with problems related to the EA. These problems were not the resource-intensive ones, or the occasional bug that someone reports to me. The problems that some people are having are along the lines of:

 

  1. This EA lost me 50%, what are the right settings?
  2. The EA lost all its gains from yesterday.
  3. Yesterday, the EA doubled my account, but today it broke even.

Basically, their expectations of this EA are:

 

  1. Holy grail
  2. Large consistent profits
  3. Bond / Bank interest
  4. Set-and-forget

To everyone reading this now, I will tell you that this EA is not the holy grail. This EA is not consistent. Money from this EA--like most of my EAs--comes in bursts. And most of all, this EA needs to be monitored.

My philosophy on EAs is quite simple really. An EA is a tool to augment a trader's trading capability. An EA will not replace a trader. Simply because you have an EA does not mean you're out of the water yet. EAs are not perfect and thus, you need to be able to take over for the EA when--not if--when it fails.

This type of EA, and the HiLo Trader before it was designed using a different mindset. Instead of writing a program to trade the markets, I looked into writing a program which in turn could write a program to trade the markets. My reasoning was simple really, I don't know what the ideal trading system looks like, but I do know what it does. So, I wrote a program to tell me what the ideal trading system looks like within a given set of parameters (win rate, risk reward, success score, etc).

The self-optimizing concept was useful in that it helped reduce the EA's likelihood to fail catastrophically. The main reason most EAs fail after a few months is largely because the market changed in such a way that the EA didn't work anymore. It seemed--based of my empirical observations--that a mildly curve-fit EA would last about 3-6 months in a forward environment. The curve against which the EA was testing itself is constantly changing, so by making the optimization periods a little more frequent, the EA would at the very least get a chance to optimize itself and realign itself to the curve before the EA becomes outdated.

[/rant]





Okay, glad to get that off my chest as incomprehensible parts of it may be.

When I get back home I am considering hosting a Q&A webinar to explain the EA, answer questions regarding functionality, and basically try to clarify anything that isn't well understood about the EA. Would anyone be interested in something like that? I need a headcount to determine if I need to do multiple sessions.

  • #555
  • Jul 27, 2010 8:57pm Jul 27, 2010 8:57pm
Quoting Ronald Raygun
...
[rant]
...
[/rant]
I like it - awesome

Quote
...
When I get back home I am considering hosting a Q&A webinar to explain the EA, answer questions regarding functionality, and basically try to clarify anything that isn't well understood about the EA. Would anyone be interested in something like that? I need a headcount to determine if I need to do multiple sessions.

Oh yea too right. webinarcount = webinarcount+1

cheers scalpz
Target 2 : Stoploss in the green
  • #556
  • Jul 27, 2010 9:19pm Jul 27, 2010 9:19pm
Quoting Ronald Raygun



When I get back home I am considering hosting a Q&A webinar to explain the EA, answer questions regarding functionality, and basically try to clarify anything that isn't well understood about the EA. Would anyone be interested in something like that? I need a headcount to determine if I need to do multiple sessions.
Count me in. :^)
  • #557
  • Jul 27, 2010 9:35pm Jul 27, 2010 9:35pm
When I get back home I am considering hosting a Q&A webinar to explain the EA, answer questions regarding functionality, and basically try to clarify anything that isn't well understood about the EA. Would anyone be interested in something like that? I need a headcount to determine if I need to do multiple sessions.[/quote]

Count me in on the webinar, Ronald. Thanks.

misterpipster
  • #558
  • Jul 27, 2010 10:03pm Jul 27, 2010 10:03pm
Quoting Ronald Raygun
When I get back home I am considering hosting a Q&A webinar to explain the EA, answer questions regarding functionality, and basically try to clarify anything that isn't well understood about the EA. Would anyone be interested in something like that? I need a headcount to determine if I need to do multiple sessions.
Count me in, as well.
  • #559
  • Jul 27, 2010 10:06pm Jul 27, 2010 10:06pm
Hi Ronald,

Thank You for creating this EA. I have installed the "HiLo Self Optimizing [TD]" version (found on Post #494, in Section 33 of this thread) on five pairs in FXDD.

The EA doesn't seem to be working. Can you tell me why? Here are my settings:

MagicNumber 1234
SignalsOnly false
Alerts true
SignalMail false
PlaySounds false
ECNBroker false
SleepTime 100
EachTickMode true
AnimateOptimization true
MaxSimultaneousTrades 10
Lots 0.1
MoneyManagement false
Risk 1
Slippage 5
UseStopLoss true
StopLoss 200
UseTakeProfit true
TakeProfit 200
UseTrailingStop false
TrailingStop 30
MoveStopOnce false
MoveStopWhen Price 50
MoveStopTo 1
Remark 2
Remark 3 -- Breakout Settings --
BarsToOptimize 10,000
InitialRange 60
MaximumBarShift 1440
MinimumWinRate 50.0
MinimumRiskReward 0.0
MinimumSuccessScore 0.0
MinimumSampleSize 10
ReverseTrades true
Remark 4 -- Optimize Based On --
HighestProfit false
HigestWinRate false
HighestRiskReward false
HighestSuccessScore true

Thank you for your comments -- this is very exciting!

Best Regards,
Jay Allen
Dallas, Texas
  • #560
  • Jul 27, 2010 10:09pm Jul 27, 2010 10:09pm
Quoting Ronald Raygun
...When I get back home I am considering hosting a Q&A webinar to explain the EA, answer questions regarding functionality, and basically try to clarify anything that isn't well understood about the EA. Would anyone be interested in something like that? I need a headcount to determine if I need to do multiple sessions.
... have a good time and enjoy your holidays, after that count me in yours Q&A webinar
(the EA still doesn't wont to open any trades for me ... and I'm starting to suspect it has something to do with my broker-GoMarkets)

Darek
0 traders viewing now