Indicators development for T101's not very simple method | Page 117 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

SIMPLE EA for a simple method needed!! 2 replies

Simple System By T101 - My Thought 18 replies

Continuation of T101 method discussions 24 replies

My first attempt at T101's basket method... 2 replies

Attachments: Indicators development for T101's not very simple method
Exit Attachments

Indicators development for T101's not very simple method

  • #2,321
  • Jun 9, 2010 4:20pm Jun 9, 2010 4:20pm
@ Don and MrPip

geee, this is 4th time or 5th time that i have to post about the bug in
void T101_CloseAll(). i refrain myself posting another of my own edition, cos none test it although it correct this closeall() problem.

1. suppose we have 20 pos opened by EA-charvo, 14 pos opened by EA-D&M (if this EA wants to go LIVE, we need to make it ok with other trades/EA.)

2. let's see below:

while(OpenOrders > 0) {
~~~~~~14~~~~~~


for (int i = OpenOrders - 1; i >= 0; i--) {
~~~~~13~~~~~~~~~

OrderSelect(i, SELECT_BY_POS);

if (OrderMagicNumber()==Magicnumber)
{ // do the closing stuff // }


so this for loop will search from 13th to 0th pos, now say, if EA-charvo's pos are number/index from 0~19 (20 pos), then the while-->for loop above will be searching in vain for EA-D&M's pos that are very likely indexed from 20~33.

To correct this problem:

add another internal var:

int allpos = orderstotal();

then in above loop, replace "for (int i = OpenOrders - 1; i >= 0; i--)" with " for (int i = allpos - 1; i >= 0; i--)"

......

never thought a closeall() could be so time-consuming...... my solution is pretty simple, COPY steve hopwood's MPTM's closeall() to this EA.
  • #2,322
  • Jun 9, 2010 4:24pm Jun 9, 2010 4:24pm
True Don,

but when i had my version on, a newer version came out. but since it was still trading. i did not want to quit yet, since other bugs might come out of that version and evntually it crashed.

when i was younger i was actually a gametester for epic. yes i helped testing the very first unreal game so if there were still bugs in that you can blame me.

but would it not be better if we make a bugtesting criteria, on the criteria you or mr pip or KOS want to test it?

lay out the rule lines how you want to have it tested. we test it and add the results here.

for example you want to have it tested with LBFX pairs yes or no.

we test it and add results here.

that way you can do selective bug testing and fixing. and try to finnish problem by problem.

you and kos created this beast. but to tame it, i think it would be better to have guidelines.

what do you think?
  • #2,323
  • Jun 9, 2010 4:59pm Jun 9, 2010 4:59pm
trade4fun,

Just as charvo said, I too didn't think a CloseAll() problem would have taken this much time to fix. Although I made it clear to everyone that this EA was for Kos' strategy, Robert still put stuff in that is not part of Kos' strategy. I didn't even put my own strategy in there. If I had, I would not be pleased with things that had nothing to do with my strategy being added. If you'll note, in the EA I uploaded, the only things the user had to choose from was IBFX pairs or not, and if they wanted to trade some pairs opposite the rest of basket. That was about it.

Having said that, I will let it go that things have been added that were not asked for, simply saying that Robert could have done the adding after the EA was corrected like is needed, and made that EA Robert's strategy. Then we'd have two correct EAs - one with Kos' strategy and one with Robert's strategy. Instead, we don't even have one EA that's finished.

That's why I asked Robert to refrain from putting anything else that is not part of Kos' strategy into the EA. The only other thing I can do is go back to the last EA I posted and make the 2 or 3 corrections I now think it needed, and test it if Robert's latest version still isn't working right. Lord knows I don't need to deal with any more code than necessary when trying to squash bugs. It's tough enough as it is.

I am hoping that Robert's latest version works, so we can keep moving forward and not go backward.

Regards,

Don


Quoting trade4fun
True Don,

but when i had my version on, a newer version came out. but since it was still trading. i did not want to quit yet, since other bugs might come out of that version and evntually it crashed.

when i was younger i was actually a gametester for epic. yes i helped testing the very first unreal game so if there were still bugs in that you can blame me.

but would it not be better if we make a bugtesting criteria, on the criteria you or mr pip or KOS want to test it?

lay out the rule lines how you want to have it tested. we test it and...
  • #2,324
  • Edited 6:06pm Jun 9, 2010 5:52pm | Edited 6:06pm
Quoting charvo
@ Don and MrPip

geee, this is 4th time or 5th time that i have to post about the bug in
void T101_CloseAll(). i refrain myself posting another of my own edition, cos none test it although it correct this closeall() problem.

1. suppose we have 20 pos opened by EA-charvo, 14 pos opened by EA-D&M (if this EA wants to go LIVE, we need to make it ok with other trades/EA.)

2. let's see below:

while(OpenOrders > 0) {
~~~~~~14~~~~~~


for (int i = OpenOrders - 1; i >= 0; i--) {
~~~~~13~~~~~~~~~

OrderSelect(i, SELECT_BY_POS);

if (OrderMagicNumber()==Magicnumber)...

I though I had already done that change.
The new code does just that, checks AllPos for trades by this EA for closing.

I will now only focus on fixing bugs only.
One current know bug is for the profit lock code.
A possible fix is to modify the compares to not check for exact match of numbers or to round the values for more accuracy. This has to do with errors in floating point.

If you compare a calculated value of 5.000000001 with 5.00 they are not equal. Rounding to 2 digits will fix the problem.

NOTE : This first bug code has been done and is ready for testing.

The other bug is the MACD exit. I will look into the code for that as well but what I have seen so far appears OK.


The addition of LSMA for determining trend does not change the actual logic. It is an indicator that does not repaint and might even get a little earlier change in trend.

The trading hours filter is to prevent trades near the open of a new session.

The rest of the mods were to make the code cleaner and more logical.

Robert
Attached File(s)
File Type: mq4 T101_EA_byDon_JustKos_06-09-10d.mq4   41 KB | 601 downloads
  • #2,325
  • Jun 9, 2010 6:20pm Jun 9, 2010 6:20pm
Great, Robert, I'm replacing with this version now.

Let's stick to this format and logic flow, i'm getting familiar with the code after i read a few times now. Please note me/us if any major change/editing is done in future.

as to MACD bug, I met once, but it was due to T101_closeall(). Is there anyone else who reported to you about this issue? maybe, i guess, it is fine now.

let's watch......

Quoting MrPip
I though I had already done that change.
The new code does just that, checks AllPos for trades by this EA for closing.

I will now only focus on fixing bugs only.
One current know bug is for the profit lock code.
  • #2,326
  • Jun 9, 2010 7:45pm Jun 9, 2010 7:45pm
Thank you, Robert. Your ideas and persistence are appreciated.
Best regards, Glenn
  • #2,327
  • Jun 9, 2010 9:10pm Jun 9, 2010 9:10pm
Quoting wheelbarrow
are you using this strategy with 7bit's anti-grid ea?
no, i trade chaos and have a method for fractal breakouts that I'm messing around with

fractals are for counting Elliot waves after you've already been in a trade. it's really for adding on positions but the original chaos method is totally discretionary on the first entry. the second, (AO indicator) and 3rd (fractal breakouts/Elliot waves) are for adding even more to a trade
  • #2,328
  • Jun 10, 2010 2:17am Jun 10, 2010 2:17am
There is one more potential bug that is easy to fix.

When there is a disconnect from the trade server for any reason an EA will re-run the init function. This will cause all variables to reset.

If there has been a profit lock and the profit retraces below the lock trigger amount, the profit lock is lossed.

The fix is to use MT4 type global variables that are stored on the local hard drive. The code will be easy to add and I will do so once other bugs have been reported.

I will be busy Thurs, June 10 at the LA Money Show.

If there are any bugs to fix I will do so when I return, The next version will also include the Global variables. Just need to determine what needs to be saved besides LockedProfit.

I also noticed 2 hard coded values in the following lines of code.
Would it be better to make the hard coded numbers (-56 and 600) inputs to the EA?

ProfitStopLoss = -56.00*Multiplier;
TakeProfit = 600.00*Multiplier;

I ask because my TakeProfit target is 6000 trading 1 minilot per pair.
That would require an average profit of 428 per pair. I realize the trailing will usually be hit but it would be nice to have a closer Basket TakeProfit, especially when the largest lock is 401 when 500 is reached.

Robert
  • #2,329
  • Jun 10, 2010 2:22am Jun 10, 2010 2:22am
goodmorning all.

just added the new version. will let you guys know how it will go.

also this version still has the moment you attach it to chart and the 4 signals are already lined up it will do direct buys. it will not first wait for a new trend. just boom you in there.

good or bad. we shall see how it turns out. hope everybody is feeling fine, i am out for a while. helping a friend building a wooden house for his children.
  • #2,330
  • Edited 8:23am Jun 10, 2010 7:55am | Edited 8:23am
Robert, your consideration on profit-lock value kept by GlobalVar is very needed if we go LIVE with the EA. That's assuming the reconnection time interval is pretty long, like a few minutes.

Also, there should be a subroutine put on a hard stoploss for each postion say, like 150 pips stoploss, to overcome the scenario of every system is down (server, broker, internet provider...) These of coz don't need to be done now.

Original TP/SL is 6000$ and 560$ for 0.1 lot each pair (1.4 lot total) using standard account (1 pip 1 $, you know). so if it is micro (1 pip 1 dime), it is 600 and 56 as TP/SL. To be honest, I never see 6000$ hit once since 2010 re-following this thread, don't mention this trailing steps are pretty tight. On the other hand, 560$ may not be reasonable, it should be adjusted according to different timeframe to trade (15 min or 1H). Originally, these two numbers were used by the author without strong reasoning, like two random numbers. Of coz, set them EXT Var is the right move.

There is one bug, or incomplete reflection of reality (@ Don, too) in my mind, which caused continue opening baskets. To spot BI_t101's signal, Don used:

if(BI_Red > BI_Blue) BuyCondition3 = True;

However, I think it is better to be:

delta_BI_red = BI_red(0) - BI_red(1);
delta_BI_Blue = BI_Blue(0) - BI_Blue(1);
(delta_BI_Red>delta_BA_Blue && delta_BA_Blue>0)

in fact, this way is STILL NOT Kos way that is about BI's cross!!!

To catch BI's cross, it should be something like this:

delta_BI_red = BI_red(0) - BI_red(1);
delta_BI_Blue = BI_Blue(0) - BI_Blue(1);
(BI_red(0)>BI_blue(0) && BI_red(1)<=BI_blue(1) && delta_BI_Red>delta_BA_Blue)

I believe that this way is not only more accurate, but more profitable and tradable, since i've already seen the results trading this way


Quoting MrPip
If there are any bugs to fix I will do so when I return, The next version will also include the Global variables. Just need to determine what needs to be saved besides LockedProfit.

ProfitStopLoss = -56.00*Multiplier;
TakeProfit = 600.00*Multiplier;
  • #2,331
  • Jun 10, 2010 8:35am Jun 10, 2010 8:35am
trade4fun,

Since all conditions are not met all on same candle/bar, then I had to use conditions 1,2,3,4, and when they align up, the EA does take a basket trade.

I really don't know of a way off the top of my head to do this differently. I think messing with the profit-locking numbers to keep us in the trend longer, or maybe not even use profit-locking, but maybe another indicator to get us out, would help. Making sure price is going in direction of trend would also be a good condition to add.

To all:

But at this point, all that matters to me is: Does it trade the way Kos would? Does it follow his strategy? Only he can tell us for sure. Until that happens, I wouldn't feel comfortable trying to over-analyze or make changes too quickly.

With one reliable, consistent EA, many differently named EAs can be made, and we can try the things we want. I would rather have one good EA with a good strategy on my live account than testing 10 different EAs on demo which may or may not have good strategies. Get one good one first, then we may have to make another thread called "T101 EA Explosion!!!", where that thread will be littered and confusing for everyone since so many EAs will be floating around with so many strategies. Ummm, fun, huh?

So let's stay the course, and it will pay off in due time. Then if we feel we must, we can do a "T101 EA Explosion" thread. K? At least in that thread we would know that all EAs used the same trading logic code, so that the only things that would be different between the EAs are global variables for indicators used, using those indicators in variables using iCustom, then making buy, sell, and close conditions using those variables, custom strings in alerts/emails, and comments.

Regards,

Don


Quoting trade4fun
goodmorning all.

just added the new version. will let you guys know how it will go.

also this version still has the moment you attach it to chart and the 4 signals are already lined up it will do direct buys. it will not first wait for a new trend. just boom you in there.

good or bad. we shall see how it turns out. hope everybody is feeling fine, i am out for a while. helping a friend building a wooden house for his children.
  • #2,332
  • Edited 9:42am Jun 10, 2010 8:41am | Edited 9:42am
charvo,

Just like I mentioned to trade4fun, all conditions are not met on same bar, so the only way I know of to do this is set conditions, which when all true, make the trades. For instance, Kos seemed to me to be showing that Trend Magic was at the "heart" of his system. And so to me, that was first condition. No matter what else was done. So I use "crossing" code for that indicator alone. Once that is checked off the list, then really all I need to know is AT THAT EXACT MOMENT, are the other conditions lining up? And if BI Red is above Blue then the cross has happened on BI, so it's good for buy, if MACD is above signal, then it's good for a buy, and if using TF for trend, then it's good for a buy if Long Strong. So only ONE variable needs the "cross" code. The rest don't, and actually I don't see how it would help them having the cross code.

Having said that, I will test Robert's latest version: d. Results from version c: I can't say if all things lined up (didn't take screen shots), but I got buy trades and sell trades I think. Big losing on both accounts. Looks like I need to make new demo account to test version d to make sure no trades are from earlier version, and maybe put on screenshot indicator to take pictures at beginning of every bar.


Kos,

Also, I see that on H4 IBFX should be having BuyAll signal at beginning of next H4 bar. MBT demo (which I just made to replace old drained one) won't be available to make trades (stupid) until after 4:30 this afternoon, although I'm getting signals from that at beginning of current bar I would have gotten in a BuyAll. (For charvo and trade4fun: Guess who was the last (almost all the time) to align? Trend Magic)

Actually we would already have had one BuyAll on H4 recently. On the bar we would have taken the trade (IBFX 2010.06.09 16:00, because the recommended condition I made is not there to check that this high is higher than last bar's high), we would have gotten into that trade, and anyone can see that price went down, stoploss would have been hit probably, would be a loser. If you look on IBFX 2010.06.09 16:00, all conditions were met, it looks like there wasn't enough profit to even lock, so that BuyAll would either a. be in drawdown if it didn't hit the stoploss, because MACD hasn't crossed back to exit the trade, b. more likely it hit the stoploss. Kos, got any input on this? Would this be a trade you got in? If so, looks like a loser to me, does it to you? (In fact, I see it would have been a loss unless you use different rules than you told us, because just looking at the BuyAll line only, and no other lines or MAs, in my newest BuysVsSells indcator, shows just close to close was $776 assuming trading one mini lot per pair, so definitely went more than -$560. 560 is way too small a stoploss on H4 IMHO, and I know hazelj80 had agreed with that, which is why I had planned on making Stoploss variable depending on an indicator, and not static, but got fed up when I couldn't get my EAs to work, mainly because of the CloseAll problem which has been shown).

Regards,

Don
  • #2,333
  • Jun 10, 2010 9:18am Jun 10, 2010 9:18am
Don, yes, i agree that we are not changing the rules.

what i suggest is definitely following Kos way.

As to 4 conditions lined-up together, there could be another way to code it.

1. imho, BI cross is the major indicator, well, if it is not, doesn't matter really;

2. when human(Kos) eyes read a chart, they always consider the direction instead of one specific bar's value. Do you agree? again, human eyes identify the direction instead of the value of one bar ( bar(0) or bar(1) ). so the code need to consider the change from at least last 1 bar to current bar, how much is the change? NOT how much is current bar. that's about a more reliable way to read indicators. In other words, you don't care about if EU is 1.10 today, but if yesterday EU is 1.2, and today it is 1.10, you will short it, right? it is not "because EU today is 1.1, and 1.1 is a low number so you short it; it is because the change from yesterday to today is huge drop......

3. all 4 conditions are often NOT lined-up together. But think this: was there any trades that Kos show with 4 conditions met taking a few hours? I don't remember so, usually 4 conditions are met within 2~3 bars at most. so a check like this should do:

check if trend magic met within last 2 bars, --- 1st true or false;
check if BI cross in last 2 bars, --- 2nd true or false;
check if MACD in last 2 bars, ---- 3rd true or false;
check if 4th cond met in last 2 bars, --- 4th true or false;

if four true are get, trade, if NOT, don't trade.

The current way of the EA is if four true are get for ANY number of bars, it will trade. If 1st condition is 10 hours ago, and the other 3 condition are met 10 hours later, the EA will still trade......Which is not correct of coz.


Quoting disbellj
charvo,

Just like I mentioned to trade4fun, all conditions are not met on same bar, so the only way I know of to do this is set conditions, which when all true, make the trades.
  • #2,334
  • Jun 10, 2010 9:32am Jun 10, 2010 9:32am
Hi Folks,

I've attached a simple little indicator which shows the net profit of the basket over time. I just attached it to a one minute chart so I could have a record of the state of the basket since I can't watch it all the time.

I hope you find it useful.

Regards, Ari
Attached Image (click to enlarge)
Click to Enlarge Name: Account Profit Indicator.gif Size: 27 KB
Attached File(s)
File Type: mq4 Account Profit Indicator.mq4   2 KB | 573 downloads
  • #2,335
  • Jun 10, 2010 9:53am Jun 10, 2010 9:53am
charvo,

How can you say it's not correct?

If you read what Kos said, he said if all were still aligned and he got knocked out with profit-lock, he would re-enter. That's what the EA does. He didn't mention gap between lines, just that they had crossed. If I check BI and make sure one is above another, that's as good as a cross. They all don't cross on the same line. But when the most lagging indicator, IMHO TrendMagic, crosses (actually price bar closes/crosses TM), then check other conditions. I would say that my eyes move from top to bottom of screen, left to right. Would you agree that your path of vision seems to travel that way also? If so, then first indicator my eyes would check is correct would be TrendMagic, before BI. If TrendMagic hasn't crossed, as I said before, it doesn't matter what other conditions have been met, the trade isn't taken. So GBPJPY has to have bar close above TrendMagic before other conditions are even considered. That's my take. That's how I coded it. Kos has said it trades right so far. I guess I'll wait for his input on if it does not.

I'm not saying your method can't be used, I have used similar myself. But Kos didn't say "if all conditions not met within 2 bar span, don't do the trade". At least I never saw that at all. Also, I'm not looking at just one value of current bar, but two values of current bar (BI Red and BI Blue) and comparing them. It's valid for this procedure. I could compare last bar's value to this bar's value, but in this situation, I think I did the correct thing. Each person has their own take on this "simple" method. God help us if we ever get to a "complicated" strategy.

Regards,

Don


Quoting charvo
Don, yes, i agree that we are not changing the rules.

what i suggest is definitely following Kos way.

As to 4 conditions lined-up together, there could be another way to code it.

1. imho, BI cross is the major indicator, well, if it is not, doesn't matter really;

2. when human(Kos) eyes read a chart, they always consider the direction instead of one specific bar's value....
  • #2,336
  • Jun 10, 2010 10:54am Jun 10, 2010 10:54am
at this point, you do need Kos to clarify these issues. i still believe that re-entry logic after a few hours is not what Kos did or intended.

Quoting disbellj
How can you say it's not correct?

If you read what Kos said, he said if all were still aligned and he got knocked out with profit-lock, he would re-enter.
  • #2,337
  • Jun 10, 2010 11:08am Jun 10, 2010 11:08am
I just got in BuyAll on IBFX GBPJPY H4 using Robert's latest Kos EA version (the defaults, except I will not use the RiskPercent or ridiculous StopLossPerPair that got put in there and that I have mentioned I dislike being done, and wanted changed back to what I had it multiple times).

We'll see how this long-term trade goes. Of course it will re-enter just like Kos said he did if all things still aligned.

Will keep you updated how it goes. Waiting on Kos to chime in saying what his testing of latest revision is trading for him.

Regards,

Don
  • #2,338
  • Jun 10, 2010 12:27pm Jun 10, 2010 12:27pm
alright guys. i am back home. never knew building a small wooden house takes so much work. although it has to be said. reading instructions will always help.

but then again. REAL man dont read instruction. at least that is what i learned when i watch tim talylors toolmen show

anyway back to Ea business.

i tested the latest version of the EA as far as i know, 06-09-010D on the 1 hour chart.

from this morning 8 am till now 7 pm. it has opened 4 trades. after every trade was finnished it started a trade again.

trade 1 till 3 were all succesfull, however everyone of those trades had a -200, -300 pip loss until they became succesfull. i know this because i backchecked the charts of every currency that was open.

talking about that. aint there an indicator which logs the process of the trades so i can check later how the charts went all time long. because what i did now is i print out the results and back check them for every trade on the charts. costed me around 90 minutes, to do this.

not complaining, but if there is an indicator like that. it would be more easy to read back pip data.

thanks.

also at this point of time of writing. i am again at a piploss of -250, -200. it is hovering in between.

also if we ever need testing on a real account. count me in as well. i am far from being rich. but i like to do my share as well for backtesting on a real account as well. just need to make sure we use different brokers. because test reulsts might vary between brokers.
  • #2,339
  • Jun 10, 2010 1:05pm Jun 10, 2010 1:05pm
trade4fun,

Aristarchus just posted an indicator that looks to do what you want (record basket profit during trades).

In fact, it was looking at another indicator like that years ago that made me figure if I could keep profit of demo or real trades by using an indicator that displayed it as a line over time, I could put on "faux" trades doing soimething in code different, but display it as a line. This is why I call my lines in BuyAll or BuysVsSells "profit lines" (because they track profit just like you were putting on the real trades or demo trades).

In fact, after a little digging, here you go. Had to go back over a year and a half to find this one.

Also, the basket was down like others have said hundreds of pips, was in profit hundreds of pips, got stopped out at $198 profit (if I was on MBT mini account, since Lots traded were .18, I would do 198/.18, but since this trade was on IBFX standard account, I'll do 198/1.8, so I multiply Lots x 10 and then divide dollar gain by that to see approx. how many pips I got. So 198/1.8 = 110 pips approx. Then new BuyAll got put on. I don't see anything hanging. This looks good as far as trades being opened, closed, more trades opened (and not automatically closed right away), etc. Robert, thank you.

Regards,

Don


Quoting trade4fun
alright guys. i am back home. never knew building a small wooden house takes so much work. although it has to be said. reading instructions will always help.

but then again. REAL man dont read instruction. at least that is what i learned when i watch tim talylors toolmen show

anyway back to Ea business.

i tested the latest version of the EA as far as i know, 06-09-010D on the 1 hour chart.

from this morning 8 am till now 7 pm. it has opened 4 trades. after every trade was finnished it started a trade again.

trade 1 till 3 were...
Attached File(s)
File Type: mq4 i-BalanceEquity.mq4   8 KB | 497 downloads
  • #2,340
  • Jun 10, 2010 1:41pm Jun 10, 2010 1:41pm
thanks Don for the explanation and indicator. i checked the explanation of Aristarchus his indicator. but i didnt understand actually what it was. however after your explantion now i understand so thank you for it.

also a small update. eventually the fourth trade ended in +50 pips. it opened up again a new trade and so far it is hovering between -50 and -100 pips.

yesterday the crash happened at closing of fourth trade. so i am looking forward on what will happen after this trade closes.

and Robert thank you for your fast work/updates of this EA as well.
3 traders viewing now