GSB Forums

General support questions.

 Pages:  1  ..  5    7    9  ..  25

admin - 24-5-2019 at 04:59 PM

Quote: Originally posted by appengineer  
Hi Peter,

TS doesn't support using Exchange Time with multiple data stream when the symbols are in different time zone. (NG,CL,HO,RB)

How do you go around this?

The GSB system I am trying to import into TS checks the last second of the trading session to close trades - If (TimeHms >= 133000 And TimeHms <= 133059) Then

Thank you

Always use local time on your pc

appengineer - 24-5-2019 at 06:08 PM

Quote: Originally posted by admin  
Quote: Originally posted by appengineer  
Hi Peter,

TS doesn't support using Exchange Time with multiple data stream when the symbols are in different time zone. (NG,CL,HO,RB)

How do you go around this?

The GSB system I am trying to import into TS checks the last second of the trading session to close trades - If (TimeHms >= 133000 And TimeHms <= 133059) Then

Thank you

Always use local time on your pc


What will the session be for each of the symbols(NG, CL, HO, RB)
0900-1430 or Regular hours.

I have tried a couple of settings and the performance is different from what GSB produced.




Untitled.png - 26kB

admin - 24-5-2019 at 06:11 PM

appenginer, your times are correct but looks like you put nq instead of ng in the data1

also see https://trademaid.info/gsbhelp/13SupplyingDatatoGSB.html
where it says session times.

appengineer - 24-5-2019 at 07:56 PM

Quote: Originally posted by admin  
appenginer, your times are correct but looks like you put nq instead of ng in the data1

also see https://trademaid.info/gsbhelp/13SupplyingDatatoGSB.html
where it says session times.


Got it, good catch

How about the hard coded time in GSB built system
If (TimeHms >= 133000 And TimeHms <= 133059)

should it be
If (TimeHms >= 143000 And TimeHms <= 143059)

Since we are using 0900 - 1430 session time?

admin - 24-5-2019 at 08:06 PM

Quote: Originally posted by appengineer  
Quote: Originally posted by admin  
appenginer, your times are correct but looks like you put nq instead of ng in the data1

also see https://trademaid.info/gsbhelp/13SupplyingDatatoGSB.html
where it says session times.


Got it, good catch

How about the hard coded time in GSB built system
If (TimeHms >= 133000 And TimeHms <= 133059)

should it be
If (TimeHms >= 143000 And TimeHms <= 143059)

Since we are using 0900 - 1430 session time?


Under contracts, check your moc time is correct, should fix this for future systems

appengineer - 25-5-2019 at 12:36 PM

Quote: Originally posted by admin  
Quote: Originally posted by appengineer  
Quote: Originally posted by admin  
appenginer, your times are correct but looks like you put nq instead of ng in the data1

also see https://trademaid.info/gsbhelp/13SupplyingDatatoGSB.html
where it says session times.


Got it, good catch

How about the hard coded time in GSB built system
If (TimeHms >= 133000 And TimeHms <= 133059)

should it be
If (TimeHms >= 143000 And TimeHms <= 143059)

Since we are using 0900 - 1430 session time?


Under contracts, check your moc time is correct, should fix this for future systems


The MOC under contract list is not editable

Also the system built has a variable called startDate which seems to be a keyword in Easy Language

stop loss

fra2019 - 30-5-2019 at 07:08 AM

How do stop-loss work in GSB? I thought 'Max.Stop Loss' was the maximum currency loss per trade so 1000 max loss for 1 ES contract would mean 20 points stop loss. This does not seem to be the case. I generated a few NQ strategies with trailing stop, Max stop Loss = 2000 and the script show slPoints(25) which implies a much tighter stop-loss.

admin - 30-5-2019 at 08:07 PM

A number os users users installed windows update blocker, as reboots on a live trading computer or gsb manager machine is really bad if they are not planned.
https://www.sordum.org/downloads/?st-windows-update-blocker

Well its times to install all your updates
there are some dangerous flaws that need patching.
https://mspoweruser.com/a-million-unpatched-pcs-waiting-to-b...

boothy - 31-5-2019 at 02:40 AM

Hi Peter,

I can't seem to get the macros to work properly. I'm trying to run indicators stats as per video but don't get the same results.

Please refer attached screen shots, I'm not getting any metrics for stats C, and stats E has way higher fitness and N/P.

I have tried macro, IndicatorStats, IndicatorStats+nthNoTrade and also tried making my own and none have seemed to give similar figures to your video.

Thanks.



IndicatorStats.PNG - 133kBIndicatorStats+nthNoTrade.PNG - 148kB

admin - 31-5-2019 at 04:15 AM

Quote: Originally posted by boothy  
Hi Peter,

I can't seem to get the macros to work properly. I'm trying to run indicators stats as per video but don't get the same results.

Please refer attached screen shots, I'm not getting any metrics for stats C, and stats E has way higher fitness and N/P.

I have tried macro, IndicatorStats, IndicatorStats+nthNoTrade and also tried making my own and none have seemed to give similar figures to your video.

Thanks.
send me your teamviewer details. I will look at this sat morning.
should be an easy fix


admin - 31-5-2019 at 05:55 PM

Quote: Originally posted by fra2019  
How do stop-loss work in GSB? I thought 'Max.Stop Loss' was the maximum currency loss per trade so 1000 max loss for 1 ES contract would mean 20 points stop loss. This does not seem to be the case. I generated a few NQ strategies with trailing stop, Max stop Loss = 2000 and the script show slPoints(25) which implies a much tighter stop-loss.

Sorry for the delay in this reply. It slipped my memory as I first say it very late at night.
Check your contract point values are correct in GSB.

I would expect the stop loss to be the extreme max that the trail is allowed to go.
I can re-check this with the programmer early next week if you still have issues

fra2019 - 31-5-2019 at 09:04 PM

pnt Value for NQ and VX seem correct. Curious to hear if anyone else experienced similar issue.

admin - 31-5-2019 at 09:07 PM

Quote: Originally posted by fra2019  
pnt Value for NQ and VX seem correct. Curious to hear if anyone else experienced similar issue.

please send a screen shot of the settings, and the ts code generated
I will look at this on monday

boothy - 16-6-2019 at 09:30 PM

I am trying to do testing on Silver to get best indicators and build systems but I cant seem to get any systems with good metrics.
All systems have under $10 avg trade, low net profit etc.

I have gone through the docs and forum and pretty sure I've got all setting correct for new updates / changes.

Any ideas what might be causing this for SI only? Seems to build systems fine on ES.

admin - 16-6-2019 at 09:33 PM

Check your secondary filter is on GA, then close/closed (from memory)
ga with closed-closed also worked nearly as well (from memory)
if not send me teamviewer details.
GSB works well on soilver

admin - 17-6-2019 at 12:47 AM

Quote: Originally posted by boothy  
I am trying to do testing on Silver to get best indicators and build systems but I cant seem to get any systems with good metrics.
All systems have under $10 avg trade, low net profit etc.

I have gone through the docs and forum and pretty sure I've got all setting correct for new updates / changes.

Any ideas what might be causing this for SI only? Seems to build systems fine on ES.

Very easy fix. Point value was wrong.

saycem - 17-6-2019 at 08:25 PM

I'm unclear when to use 40 vs 1000 for Generations per your recent post. Do we leave on 1000 with 2 indicators to determine best indicators and once we have top indicators change back to 3 and change generations to 40?



Capture2.PNG - 6kB

admin - 17-6-2019 at 08:55 PM

What happens is after 40 or 1000 generations (depends on what you set it to), the random seed will completely rebuild.
New indicator sets are chosen,
With #indicators set to 3 and 16 of them used, I saw the c/us ratio increasing (not good)
However if the generations was reduced to 40, c/us is much lower.
With gen40 on one worker I got 27835 systems, while with gen 1000 got 10768
(The workers both ran at the same time for the same duration)

You can see on the graph shown, that at 15 minutes, c/us is increasing again, and its inverse (Pass score) decreases

Also the amount of duplicates systems increase at around 10 minutes.
This all implies its time to rebuild the random GA see at 15 minutes. The exact time isnt critical, but 1000 generations is likely to be hours away. Much too long.
2 indicators are going to hit this issue much quicker.

Keep in mind, if gsb was as simple as having only say 3 indicators, and there were 16 separate indicators - then there would be 16^3 possible systems. = 4096
This is why c/us increases. The direction GSB development is going is to make duplicate filters tighter, but add to the amount of indicators used + pattern filters etc





gap.png - 82kB

Best Indicators

saycem - 19-6-2019 at 12:08 AM

I have tried to replicate the top indicators for S&P that you have posted in forum. I built on 29 30 31 (30k systms) using new signedPower settings and 40 Generations. My list is quite different from yours. Is this down to difference in settings now?

Also by default I only have 36/40 indicators switched on which did not have CloseLessPrevClose, SS RSI or X Avg. All of these are in your top 1/3. Just want to check that this is an error on my settings and these should be set to true which would give me 39/40.



MyTopIndic.PNG - 44kBPetersTopInd.PNG - 61kBMyOptSettings.PNG - 47kBActiveIndicators.PNG - 100kB

admin - 19-6-2019 at 12:22 AM

Hi Saycem
I did the same test today, and think got different results to my original ones.
However I did similar to you in testing.
My logic was our method has changed a bit since when we first did this, so I should revisit this again.
Resullts using my new set of indicators were not as good as the old. But i need to recheck my results
You should never use roofilter1pole as its redundant by the 2pole

Correct to remove one closed as they are functionaly the same with no data2

Not sure why your removed SS RSI or X Avg if they are in the top 1/3

This whole section is an area to be re-explored

Cloud Workers error messages

Daniel UK1 - 25-6-2019 at 12:17 PM

I seem to get error messages from cloud workers running on my pc while working, since its not from my testing in GSB, i find it rather annoying, shall it be like this or something is wrong ?

CaptureERRORworker.JPG - 73kB

Night session

Daniel UK1 - 25-6-2019 at 01:43 PM

Night session, i was going to test out the night session for ES, for example 1630 EST to 930 morning... created the data, with same time... however it seems impossible to get any systems built, created session times with the same... using 1630 Mon - 930 Tuesday as an example... Used same times in left side of GUI ... i cant find any reason why it should not produce any systems... also lowered filters.. is there any tweaks in order to make night session to work since its between two dates.. that i have missed ?

Any tips or help would be appreciated.

admin - 25-6-2019 at 04:55 PM

Quote: Originally posted by Daniel UK1  
Night session, i was going to test out the night session for ES, for example 1630 EST to 930 morning... created the data, with same time... however it seems impossible to get any systems built, created session times with the same... using 1630 Mon - 930 Tuesday as an example... Used same times in left side of GUI ... i cant find any reason why it should not produce any systems... also lowered filters.. is there any tweaks in order to make night session to work since its between two dates.. that i have missed ?

Any tips or help would be appreciated.

I would guess the issue is related to SF closed. On a worker (not manager) go to the top/diag tab then right click refresh. See whats going on in gsb.
Also check top right in gsb, is fitness zero or something else.
if zero its a drastic problem. ie oyu might have time to enter set 830 to 1500 but have data 1630 to 830 etc

Daniel UK1 - 26-6-2019 at 05:30 AM

Quote: Originally posted by admin  
Quote: Originally posted by Daniel UK1  
Night session, i was going to test out the night session for ES, for example 1630 EST to 930 morning... created the data, with same time... however it seems impossible to get any systems built, created session times with the same... using 1630 Mon - 930 Tuesday as an example... Used same times in left side of GUI ... i cant find any reason why it should not produce any systems... also lowered filters.. is there any tweaks in order to make night session to work since its between two dates.. that i have missed ?

Any tips or help would be appreciated.

I would guess the issue is related to SF closed. On a worker (not manager) go to the top/diag tab then right click refresh. See whats going on in gsb.
Also check top right in gsb, is fitness zero or something else.
if zero its a drastic problem. ie oyu might have time to enter set 830 to 1500 but have data 1630 to 830 etc


Thank you, will check the settings again

Daniel UK1 - 26-6-2019 at 05:33 AM

Quote: Originally posted by Daniel UK1  
I seem to get error messages from cloud workers running on my pc while working, since its not from my testing in GSB, i find it rather annoying, shall it be like this or something is wrong ?



Any comments on these error messages that seems to be coming from cloud workers running on my PC? Shall really the host of the cloud worker be getting error messages like this ?

CaptureGSBERROR2.JPG - 40kB

Capturegsberrors1.JPG - 46kB

admin - 26-6-2019 at 05:35 AM

Quote: Originally posted by Daniel UK1  
Quote: Originally posted by Daniel UK1  
I seem to get error messages from cloud workers running on my pc while working, since its not from my testing in GSB, i find it rather annoying, shall it be like this or something is wrong ?



Any comments on these error messages that seems to be coming from cloud workers running on my PC? Shall really the host of the cloud worker be getting error messages like this ?


the workers need to be running in gsbadmin mode.
Other users dont have this entry type.
That actually from my manager.
Put the config file in all paths of workers and managers.
I will email you it.



Daniel UK1 - 26-6-2019 at 12:58 PM

MC Missmatch between portfolio trader and chart when using GSB Code.. Chart not matching performance from GSB either, trades are taken overnight on chart despite end time is 1430, but this does not happen in portfolio trader.
In the example entry is taken as a second trade during the day at 1430, and exits with a stop loss at open next day?..

Chart example in the left, and portfolio trader to the right, same strategy test code from GBS.
Code used is below.

What is wrong?



// Settings
// Platform: MultiCharts
// ID: 20190626-122208-159579-MkIN8
// Info: File Name Prefix: , Comment: , App Settings: DanielStandard.gsbappset, Opt. Settings: Daniel.CL.ValidationTestt4.7.TopPrm.gsboptset, Workplace Manager's ID: OO1qnLqJMoKAviQvF.20190626-141942-293355, Manager's GSB Version: 1.0.53.24 / 2019-06-20, Worker's Instance ID: nTybj3nfT7CSW2RAM, Worker's GSB Version: 1.0.53.40 / 2019-06-25, Worker's Machine Name: GSB6_E5-2690
// Price Data: Data1: CL1430.1.Minute.0900_1430EasternTime.20190616.txt (Mult.: 15, Session: "0900 - 1430"), Data2: CL1430.1.Minute.0900_1430EasternTime.20190616.txt (Mult.: 30, Session: "0900 - 1430")
// MaxBarsBack: 500
// Profits Mode: Currency
// Quantity: 1
// Quantity Mode: FixedShareContracts
// Trading Dates: 1900-01-01 - 2018-02-28
// Trading Dates Mode: Trd
// Trading Nth Day: 1
// Trading Nth Day Mode: NoTrd
// Exit Stop Loss: 750
// Exit Profit Target: 0
// Exit Minutes: 0
// Exit Bars:
// Fitness Criteria: NetProfit * AvgTrade
// Commission: 0
// Slippage: 0
// Reports Commission: 0
// Reports Slippage: 0
// Commission Mode: TradePerSide
// Slippage Mode: TradePerSide
// Long Trading Allowed: True
// Short Trading Allowed: True
// Max. Entries per Day:

// Performance (full period)
// Fitness: 6,412,016.42
// Net Profit: 46,760
// Commission (in $): 0
// Drawdown: -11,900
// Avg Trade: 137.13
// Percent Profitable: 50.15
// Pearson: 0.968
// Profit Factor: 1.56
// Trades Count: 341
// Net Profit / -Drawdown: 3.93

// Vars
Vars:
i1Data(2),
i2Data(1),
i2shortLength(4),
i2longLength(9),
i3Data(1),
i3length(37),
i1Weight(1),
i2Weight(0.5),
i3Weight(1),
entryParams(20),
iSFData(1),
iSFWeight(1),
sfEntryLevel(80),
stopLossValue(750);

// Vars
Vars:
id("20190626-122208-159579-MkIN8"),
debugScriptPath("C:\GSB\Data\Debugs\20190626-122208-159579-MkIN8.ts.txt"),
_startDate(20070108),
endDate(20180228),
dateYmd(0),
timeHms(0),
nthDay(1),
lastDate(0000101),
daysCount(0),
currentBarDTOHLCV(""),
lastBarDTOHLCV(""),
result(0),
sfResult(0),
decision(0),
sfDecision(0),
flag(0),
zs(0.0000000001);

// If the statement below does not compile, please import the latest script file from C:\GSB\GSB (Managers)\TradeStation Code\GSB_Scripts_2019_06_04.eld
Once
Begin
Value1 = GSB_Scripts_2019_06_04;
End;

// MaxBarsBack check
Once (MaxBarsBack <> 500)
Begin
RaiseRunTimeError("MaxBarsBack (Maximum number of bars strategy will reference) must be set to 500 (from Properties for All button, General tab)");
End;

// Date YMD, Time HMS, and Day of week
dateYmd = Date + 19000000;
timeHms = Time_S; // Compile note: BarDateTime.Format() is supported by TradeStation only while Time_S is supported by MultiCharts only

// SetExitOnClose
SetExitOnClose;

// Exit Stop Loss
SetStopLoss(stopLossValue);

// Trading Dates
If Not ((dateYmd >= 19000101 And dateYmd <= 20180228)) Then
Begin
BuyToCover("SX-TrdDts") this bar on close;
Sell("LX-TrdDts") this bar on close;
End;

If nthDay > 0 And Date >= CalcDate(lastDate, 80) And dateYmd > _startDate And dateYmd <= endDate Then
Begin
lastDate = Date;

daysCount = daysCount + 1;
If daysCount > nthDay Then
Begin
daysCount = 0;
End;

If daysCount = nthDay Then
Begin
BuyToCover("SX-NthDy") this bar on close;
Sell("LX-NthDy") this bar on close;
End;
End;

// Result and decision
result = ((Sign(GSB_Norm3(GSB_RoofingFilter1Pole of Data(i1Data), 12, 100) of Data(i1Data)) * Power(Absvalue(GSB_Norm3(GSB_RoofingFilter1Pole of Data(i1Data), 12, 100) of Data(i1Data)), i1Weight)) * ((Sign(GSB_Norm3(ChaikinOsc(Volume, i2shortLength, i2longLength) of Data(i2Data), 12, 100) of Data(i2Data)) * Power(Absvalue(GSB_Norm3(ChaikinOsc(Volume, i2shortLength, i2longLength) of Data(i2Data), 12, 100) of Data(i2Data)), i2Weight)) * (Sign(GSB_Norm3(GSB_SS_RSI(i3length) of Data(i3Data), 12, 100) of Data(i3Data)) * Power(Absvalue(GSB_Norm3(GSB_SS_RSI(i3length) of Data(i3Data), 12, 100) of Data(i3Data)), i3Weight))));
result = IFF(AbsValue(result) > zs, result, 0);
decision = GSB_Decision2(result, 3, entryParams);

sfResult = GSB_Norm3(GSB_CloseOverPrevCloseD of Data(iSFData), 12, 100) of Data(iSFData) * iSFWeight;
sfResult = IFF(AbsValue(sfResult) > zs, sfResult, 0);
sfDecision = GSB_Decision2(sfResult, 1, sfEntryLevel);

currentBarDTOHLCV = NumToStr(DateTime, 9) + NumToStr(Open, 9) + NumToStr(High, 9) + NumToStr(Low, 9) + NumToStr(Close, 9) + NumToStr(Volume, 9);

// Entry-filter check
flag = 0;

If (dateYmd > _startDate And dateYmd <= endDate)
And (currentBarDTOHLCV <> lastBarDTOHLCV)
And ((dateYmd >= 19000101 And dateYmd <= 20180228))
And ((timeHms >= 090000 And timeHms <= 143000))
And (daysCount <> nthDay)
And (Not (TimeHms >= 235900 And TimeHms <= 235959)) Then
Begin
// Buy/Sell
If decision = 1 And sfDecision = 1 Then
Begin
Buy("Long Entry") 1 contracts this bar on close;
flag = 1;
End
Else If decision = -1 And sfDecision = -1 Then
Begin
SellShort("Short Entry") 1 contracts this bar on close;
flag = -1;
End;
End;

lastBarDTOHLCV = currentBarDTOHLCV;

// Decision Exit (I/SF)
If flag = 0 And CurrentContracts > 0 AND GSB_DecisionExit(decision, sfDecision, 4) = 1 Then
Begin
BuyToCover("SX-IFltRvAndSFRv") this bar on close;
Sell("LX-IFltRvAndSFRv") this bar on close;
End;

// Hash (DO NOT CHANGE)
// In order to restore the system generated this script, save the full script as a .gsbscript file and load it from GSB (just like .gsbsystem files).
{XQAAEABqlgAAAAAAAAA9iIUkQRkui9ssIYieKTlIhkh9PnHJvlp++lpxS3dTQDpyHrwKzdA2RKvQrUhRpkTBmM9wQsJvT7xZKOh5Ebuc5EUO3i688AdtYwcyawSaOkLx/lb04VUMhGBgp6cD/4s6H 6hzkIXKSBuJ
4MHL3tHBCkmKqo1pWsAXGnbZqdxgSRGgygGciid3Zy4NzO3/YBhJrVFU03j6D5IESSTz2iaSiNNjiDC71kSCgPKaJ0pSIy0SfGT15QUW0I6KaOOYyVhtQCMV8VzbYm7SvVZ5aVwG344OAKiRiwkXxx U+eo/YyiAr
aReMrjsAzhxsHQHnFisvmRHh2Qg5bCIyphtKd2IuQjGzg2TkpMgYLzmCJ2VX00ocM9y9zdK61GvAsoSJXfC199sQgjXwnnkuB+72um74BsMhkW3av/qWhdxQXAK1YXC3Up0XugbmsrpTo5rMG5jpKf +k0DKKD0Sh
MZ71knmkztnoBi4M6rVEN5MjSZ0bm/d4s/OFQor02doykqkKLK6fHhDyrlG+5cldYZ1rFJxxT495IBC7jrAcgOixaNSYvjG3l+yxdFNQ+2zOTY6gVHjzdqDj8YUH3eqkLrNrByiUlPXL9rafDTAa8+ GmTXCtoJ0e
GepMDNEPBx9X05P6uks11+uwA8jH52vg7cMdV5fFML65+2x59WPkytH5mdc1Rwun+swjyZFIC2zJRvfKfIzuKH/yPawWDdTG3HjsCoI1NjVXfej1hzJAjtPWeRDws7rqCy4nfY//2jeQx7HfsUMw53 YJDYekilSR
yYnPcWl4U4lDD9PWJSbsN2YmqDabGA3nLf30kPfUI2jowVdX9PaJomWhojKe2uZkP3PUkjAVFKU8WlDfU+wqFzpSFf/ITMDj/MhAJ53k5VyvHnlty7tdJm2WwKGaF96ElO7AInT0BYNb2pcIgTQuWB vmpUfuUat0
ykiJtx+FyXUKCmgxW3WqeHkYr+Hq5Hjx0VgWrdZRrzs/47JFMQG884IqshVTJ6ANvfNCMdd/7RreMZ8/evaws93sDnXLLBgS0XQNyi91zd7t8tm/sqVTA0hU91TdvBqoy7IfU7PdXVyRX7Hrqg+OiT 1pzJkXpb5D
KuGLkH/1KFGw4ryTEyMH2G1bHUD7nIyBRb4YKH+pzsp8Y6lShMtQ7ptLdFVzETV0tqqbsEVN7rhMTvWwEjhk817/PMLHwVzsCuTNHzeUJ7NhlHHO6XY24B15BNs0Jd+qJKqKCY0TEqK6x4O11EhFZ1 GVujcJ1A0e
8suxJCTJobC/+011+3pCVMlFBKL63DU6ebSnrAJ02DKWxpr3qwvNKYxKLeaddHvaRv0rz/cGzuglvJgPILW9Brn1pMqc9+yD6y7S9ofb+f/3+8TkVxPRkv4yQoK34HSDiCpXzFzU00ObVP2jBav946 FWEa6VMxNy
FKZkwRZfW9PpxtrFnVdSiHp0SG3ngK7PszgYFANwHPBLuTbY5GEMAAwv7evCtrvUUTpzG3hlX32m3Z4X8mOXnO3IqcrIwUnP7ctl3SYN4bh9SixaC2N973s1TRXQ5/N4niu4G0h9R91KzxUOND6UPw 6Os4rvVu3A
1Q6STnDyz9jL7XWmd4cD24CgRByiyOfUfcmJQBtzCvS7yTWaPHDIjxCzoufo4Lla4CJyLlGRdOX/eNd9JYQIW8m3zrmyBmweLhLJ+ReDlcDZrh3DL8PaX0K/zepr3XB81OLPFrxqkVXyqLw3g4Qf8/ 2C3TyzZMdj
gvSSYkf0fhAKOMjny2aFv6/sptaXWQqRHTIZCGV7hxDOP/hdEPIs5+V1hq/DGiglQI7n/A8kcVxsk7noRjL0j4h9e4/Z7B04lSVo/C3C+ZFoG7787Ue7vSUxL6m2f9gbpsdXG1aTL4H457xwuZUTSK pFF1dRAp5m
xnf4eOa+gKrcDnhGIrMRm0Qcmv5At91frI8RH9PUaNn8I8+Btl/oZJWolRAywWjiAmRiIpfDSSKv74tN4zZrZqchUYjc01Ph+Np7eMXIyMUUKL24oFKOx+vX7QZYKAx4WgHqtzB8Fiw7hq35BlIxkx 7ZSk6wrkD2
qcS8UujvJTMY7xHoPJI7qqMffAfQ9JeY6UUeNqcrtCKn/7OIfjjNf59evUgiQmGb5Kj1kasza30YxI3QR3igBbFleflKNow7VdJ+m+t43tmXnBaizf6RpiOjPT1sI9TUYrnFGLF645TC4HrIWGqZLc eRLxvU+T8a
DYMRcV/HfUhP53xLa5IX56pQMO1CYd8SggkvJliGh7tZPp/cvjJLKmU8A03Eer1BRryV/IDLWmLOJBRd8RMwAb0mNskVsmUFxTnmGPo4vg5WVnkjY0Ho7I0wCTmtzsQo0+KBgp8azY4fO/J8tIUIcA q/ZTp+4zBM
1bKIisa6QTjxpVm5qE31rYfR9doQU5XGd+cbslF9mMvUqnU+rLU5qOMqt58kwUhsiQ2T+R3eRzcgl6CLOo/i7YWYWzlaadSoe3I55CHRpSPzYY7CRiktgFodEduLeI+Gx3M/LcsFk4sEG/3GWKoOBH UOLz3QmG7q
MZXDegvrkjgUa3G5ckRSwrpE/XG90OCR6QeP2IrYlibhqIQ9GCOI6LofWCD2c03JHtMypCEL6F7Q+9ymV3VVh1PHFrpnkUjtH74qOUykYNaSzNPLmIr7BdP86JW1D6Mmu7QzalsKuT6GlUC5ZwsQXD ZvSc6CDJ4E
fTNo5eoQNSu1qKOFOcToTXJvkfx6g9knOy4A51+Ae3uytNDrXapvRfmrTqRo88rNG28mxucGDGeQp/LglY8YWijkNhnNgV3qZIq7vF+rRZRcFEOlLszLNwyqvvweOTxoDnaoyhi818emiSrxeD+aAO O9fRi6CQkM
UO87GsXhGIJ5eCx6N4aUkIH1//GuBXwYkcZukrfZ/rcEyAUJRyqOd1cQ4OwF3c+DNPZOZAlXiWqpCx4tsgduY3s9eW6bxA4M+VbhOU4owZ/LT1RdX4eS9Or+P78DkAuPD22Eh/k/eDA1iGmCNUWv0x 8bMthIp1hV
Cq5/b+G7YMz1JXu2Blxyjfk/6MbWCtn+4VKbc2/5lMZ8ROGGxaLVyyWIBE6VUGPrjlFX1FgKlD2kLAf8+hD/CzPZ8c3QnK1jyVni4AOMuiozLLhmwGV1EXCSpDnzKVlaEqcNURmSr4wXUqiPhY1HN3 RnmWPiPWe/
dDoT0zTIFHqXenXYejOSfeo1JJqrygB6xNMrobxU66495fjlVimHSk6xvCu11n/cV6b6NCbu41M2SYWBfzHK9P7mUjfhHjBBWMOdF8HexbPt65PFFB3q/Nk3uza6Ldx45ygWrCZrVSWbXi2FKNsDDz Weobqhb43L
57WGGV3tHbNg3JyRtQ0QaOu0qdFWgLeHNifll5nfobdOwUMGAH8pJj/uz/TOa/jcystcjqnwRJy5GMHGIvNZ4gN4nykD0E2R90XRU+o1a8T0tokrZDdsw+R/icgnQVksMnPe08t7GSGleNvmVAFuH3 +ny9ofczi6
SqL/aj18D38nho5859Q0Mrc9nZkslDJJhNX+DL96mpxD+3fBsQOfN2aUiWw9g+02IzfgCbPVvrmS3WSe03jvLE3/VOt8Xc/M5flHB3wl0cQkecnghSU8NBtvfwNdPvTwI3R9Q9+KTVZxhMp/GBhSSV NnML/xlX8p
+vb7ZrS9sDy8QhigqHSDG/QGx0sGYvphcawdGQ3gehw9eN99q9+4RBnAlOBHRlnLKgTc/5d6igvVPJMxnN/Wv4wxa8piLWoC5KvuAi1DesKyuS0T9BiOnCftImaDjDlM8eguWdc0PV/dy28edVmA8O Qm2dMyLqDd
8RI9FabEYvRbqvjvBGNZUqUJBK6aVi4gdvXHaNCfUEYL6Ng9418LgFnW63TufndCVsfRViSnkXYhrE+LvrYMYjhUpfi0YwIaVwgzPFmJEWbWPcO1qGiE433h3/qhkJhmXuJ+dpBSFyOGhIhLx4lot6 mY7naJVTCg
Y9WQVOMw6JbmxR3Tb0WtVkv0m8+uL5rCIC1NytiMwt6xrEIcYq+sZlPR+KmGdDA5lvalFbQGi8y5rjYLOINa10NNSj5tnlL/BjwlNy3S45NmauQCagORMntat8mSn53eg36TqO6XxufD0mvV3e2SLA X/jdNA5VKb
Zp2mDp+7twxj2dH4RFP4C6ls9+GAM4nun54pcSKl4rBiAdjC0VEGUgidrbmqM0jPCAH3kOrlw6fwKCghj7LtsNYu6eXoX1wSy8QXrM8trzqaYB1nSVwAXUTiU1KhcwkqU0Yp45BUFwKouaxsVyIAaZ BYKL0D3TVn
Q/UB0du8WlHIIQnB1yNGT/dtvyVoOltbqZbeBQM6DJOCL5Wvkh8Q5E5BvAabaq/DzHUkMD2Gvs6T+BDsyhaaXSTwJcn3OztceDPz9TJe/sPz8MLOfpBRnrIQk0jYXBYLLMyNJK1TxOfa9X/HGp7QUF A4RhbbySfv
UChIW9HkQWG10Me3Y2NwSwNVCDgZAnH+MIOcGzR/njp6oGTzq9Ez5BIG6N6y/tBWjtrmv8rT0rzTxulh+pXa04bWZd4a0ODJPrJWWfBdBLWqJoyyTc0FQP4RNuaa24lvv7pyJ2Ycq/tN1NS5EqjvY5 1vTatLY5lI
Yv+rcsfTpa9zRT0emXFnnX9tWU3hSIoIz3sEQYtzsmdseX0MmR5HGJ+OZnrftjVN703PhOyziDZqaqeOii4O9uTB9PVMT85JIYRrFYfyFJ8iAw4YQSdt0pBcOlDUJT2QPUetynB1dMWbd5oMxuyFAD 7eidVTx/IX
TMYnDFHK01CBtNGS5UFMee6RA3KEmjnn7isCJ97YIhGu3FAGvUAXIB/BYk9tJczy3vAf9sHteKfvGLP+Rg5ywNYixswMtaRw5f2m3lnWWxUv0rvqaWTwEB0lJbLjsjVAJomQiciQCVDi2lPpNJN/DN Dqf71qu97u
hh4W+XT8KBvQe3Y/Xi7t/fB1TUraSRowUZeSo8HnZBQkvFs1d88g7Uh8jr2BIreEZTou1IfyU+FZME69cjcNvvA57mgNdvO1KLI/ElKNTU2Z7Uuj5hEjoZraU+tTQW6TOOcBRvP+lptwkXte9LCYgy IprYKC8T0w
mbLi+awnAHXdUrR4nkU/BmK02ByPXg==74367a2f949a170f057805ec7c743030}


CaptureMISSMATCH.JPG - 143kB

admin - 26-6-2019 at 04:15 PM

First think I see is "And ((timeHms >= 090000 And timeHms <= 143000))"
so you can enter on your moc bar. Thats not good and is part of the issue
under trading period times on left side of gsb, change times to 900 to 1400

Daniel UK1 - 27-6-2019 at 12:56 AM

Quote: Originally posted by admin  
First think I see is "And ((timeHms >= 090000 And timeHms <= 143000))"
so you can enter on your moc bar. Thats not good and is part of the issue
under trading period times on left side of gsb, change times to 900 to 1400



Thank you Peter, i was under the impression that specifying end time was enough on the left side under Trading period, so i had it at 000000 - 1430 in GUI, i did not know that would result in that i was able to enter at the MOC bar. I will manually change this in this code as your example, and test again.

Shall i take this, as recommended best practise is to set trading time to one bar before real close on chart you are using?

So if your chart is 0800 -1315 and you are using 15 min bars...
You set trading time to 0800- 13000 ?
Resulting in
>= 080000 And timeHms <= 130000))

admin - 27-6-2019 at 01:04 AM

Quote: Originally posted by Daniel UK1  
Quote: Originally posted by admin  
First think I see is "And ((timeHms >= 090000 And timeHms <= 143000))"
so you can enter on your moc bar. Thats not good and is part of the issue
under trading period times on left side of gsb, change times to 900 to 1400



Thank you Peter, i was under the impression that specifying end time was enough on the left side under Trading period, so i had it at 000000 - 1430 in GUI, i did not know that would result in that i was able to enter at the MOC bar. I will manually change this in this code as your example, and test again.

Shall i take this, as recommended best practise is to set trading time to one bar before real close on chart you are using?

So if your chart is 0800 -1315 and you are using 15 min bars...
You set trading time to 0800- 13000 ?
Resulting in
>= 080000 And timeHms <= 130000))


correct. You should get it right before you build systems, but if you didnt, it doesnt matter - as long as fitness was np*at with 0 slippage used.



Bruce - 27-6-2019 at 01:16 AM


Peter,

When you want to introduce random tick data to the underlying price data, do you perform the randomization to the 1 minute data then once complete take that to make the 30 minute file, or do you perform the randomization to the 30 minute file?

admin - 27-6-2019 at 01:18 AM

Quote: Originally posted by Bruce  

Peter,

When you want to introduce random tick data to the underlying price data, do you perform the randomization to the 1 minute data then once complete take that to make the 30 minute file, or do you perform the randomization to the 30 minute file?


I right click the 30 min bar thats made from 1 min bars
This then makes a 30 min bar with ran noise on it

Daniel UK1 - 27-6-2019 at 01:26 AM

Quote: Originally posted by admin  
Quote: Originally posted by Daniel UK1  
Quote: Originally posted by admin  
First think I see is "And ((timeHms >= 090000 And timeHms <= 143000))"
so you can enter on your moc bar. Thats not good and is part of the issue
under trading period times on left side of gsb, change times to 900 to 1400



Thank you Peter, i was under the impression that specifying end time was enough on the left side under Trading period, so i had it at 000000 - 1430 in GUI, i did not know that would result in that i was able to enter at the MOC bar. I will manually change this in this code as your example, and test again.

Shall i take this, as recommended best practise is to set trading time to one bar before real close on chart you are using?

So if your chart is 0800 -1315 and you are using 15 min bars...
You set trading time to 0800- 13000 ?
Resulting in
>= 080000 And timeHms <= 130000))


correct. You should get it right before you build systems, but if you didnt, it doesnt matter - as long as fitness was np*at with 0 slippage used.



Thank you for clarification and confirming to specify, end to one bar before real MOC in trading session times.

However not understanding your comment below,

"You should get it right before you build systems, but if you didnt, it doesnt matter - as long as fitness was np*at with 0 slippage used."


admin - 27-6-2019 at 01:40 AM

"You should get it right before you build systems, but if you didnt, it doesnt matter - as long as fitness was np*at with 0 slippage used."


What that means is in future make the correction on time in GSB. But if you want to use systems made with the wrong time on them, then
just fix it in ts/mc and all should be well.

Daniel UK1 - 28-6-2019 at 05:18 AM

Quote: Originally posted by admin  
Quote: Originally posted by boothy  
I am trying to do testing on Silver to get best indicators and build systems but I cant seem to get any systems with good metrics.
All systems have under $10 avg trade, low net profit etc.

I have gone through the docs and forum and pretty sure I've got all setting correct for new updates / changes.

Any ideas what might be causing this for SI only? Seems to build systems fine on ES.

Very easy fix. Point value was wrong.


Same issue here, however cant see that it shall be different from what is set in GSB as standard for SI.. i have values: Ticks 200 Pnt Pal 5000 Digits 3 ..... is this incorrect for SILVER Futures?

admin - 28-6-2019 at 06:02 AM


"Same issue here, however cant see that it shall be different from what is set in GSB as standard for SI.. i have values: Ticks 200 Pnt Pal 5000 Digits 3 ..... is this incorrect for SILVER Futures?"

this is correct. Im not clear. Is profit per trade too low? Check secondary filter is ga/closed... not closed.
Check fitness = np*at

Daniel UK1 - 28-6-2019 at 06:46 AM

Quote: Originally posted by admin  

"Same issue here, however cant see that it shall be different from what is set in GSB as standard for SI.. i have values: Ticks 200 Pnt Pal 5000 Digits 3 ..... is this incorrect for SILVER Futures?"

this is correct. Im not clear. Is profit per trade too low? Check secondary filter is ga/closed... not closed.
Check fitness = np*at


SF was GA, i also tried CloseLessPrevCloseDBv
Fitness used was Net profit * Trade

Results was around 3k profit and 300 DD, avg trade 12usd, hence i thought some values for contract was incorrect..


boothy - 28-6-2019 at 07:20 PM

Hi, I am having trouble with ts not generating any trades when I copy code across from gsb.
from reading above posts, could be something to do with trading times? I have had 00:00:00 to 23:59:59 on left side GUI.

could you please explain this new logic IsSessonOn, could this have something to do with it?

// Is-Session-On
isSessionOn = (weekDay = 0 And Time > 0000 And Time <= 1430) Or (weekDay = 0 And Time > 1430 And Time <= 2359) Or (weekDay = 1 And Time > 0000 And Time <= 1430) Or (weekDay = 1 And Time > 1430 And Time <= 2359) Or (weekDay = 2 And Time > 0000 And Time <= 1430) Or (weekDay = 2 And Time > 1430 And Time <= 2359) Or (weekDay = 3 And Time > 0000 And Time <= 1430) Or (weekDay = 3 And Time > 1430 And Time <= 2359) Or (weekDay = 4 And Time > 0000 And Time <= 1430) Or (weekDay = 4 And Time > 1430 And Time <= 2359) Or (weekDay = 5 And Time > 0000 And Time <= 1430) Or (weekDay = 5 And Time > 1430 And Time <= 2359) Or (weekDay = 6 And Time > 0000 And Time <= 1430) Or (weekDay = 6 And Time > 1430 And Time <= 2359);


admin - 28-6-2019 at 07:29 PM

Quote: Originally posted by boothy  
Hi, I am having trouble with ts not generating any trades when I copy code across from gsb.
from reading above posts, could be something to do with trading times? I have had 00:00:00 to 23:59:59 on left side GUI.

could you please explain this new logic IsSessonOn, could this have something to do with it?

// Is-Session-On
isSessionOn = (weekDay = 0 And Time > 0000 And Time <= 1430) Or (weekDay = 0 And Time > 1430 And Time <= 2359) Or (weekDay = 1 And Time > 0000 And Time <= 1430) Or (weekDay = 1 And Time > 1430 And Time <= 2359) Or (weekDay = 2 And Time > 0000 And Time <= 1430) Or (weekDay = 2 And Time > 1430 And Time <= 2359) Or (weekDay = 3 And Time > 0000 And Time <= 1430) Or (weekDay = 3 And Time > 1430 And Time <= 2359) Or (weekDay = 4 And Time > 0000 And Time <= 1430) Or (weekDay = 4 And Time > 1430 And Time <= 2359) Or (weekDay = 5 And Time > 0000 And Time <= 1430) Or (weekDay = 5 And Time > 1430 And Time <= 2359) Or (weekDay = 6 And Time > 0000 And Time <= 1430) Or (weekDay = 6 And Time > 1430 And Time <= 2359);



Likely, this is a bug in a recent build fixed by the latest build.
SO try the newest build, and if that doesnt work, just remove some of the date time filters to see whats the offending line.
if stuck i will help

boothy - 28-6-2019 at 09:05 PM

Quote: Originally posted by admin  
Quote: Originally posted by boothy  
Hi, I am having trouble with ts not generating any trades when I copy code across from gsb.
from reading above posts, could be something to do with trading times? I have had 00:00:00 to 23:59:59 on left side GUI.

could you please explain this new logic IsSessonOn, could this have something to do with it?

// Is-Session-On
isSessionOn = (weekDay = 0 And Time > 0000 And Time <= 1430) Or (weekDay = 0 And Time > 1430 And Time <= 2359) Or (weekDay = 1 And Time > 0000 And Time <= 1430) Or (weekDay = 1 And Time > 1430 And Time <= 2359) Or (weekDay = 2 And Time > 0000 And Time <= 1430) Or (weekDay = 2 And Time > 1430 And Time <= 2359) Or (weekDay = 3 And Time > 0000 And Time <= 1430) Or (weekDay = 3 And Time > 1430 And Time <= 2359) Or (weekDay = 4 And Time > 0000 And Time <= 1430) Or (weekDay = 4 And Time > 1430 And Time <= 2359) Or (weekDay = 5 And Time > 0000 And Time <= 1430) Or (weekDay = 5 And Time > 1430 And Time <= 2359) Or (weekDay = 6 And Time > 0000 And Time <= 1430) Or (weekDay = 6 And Time > 1430 And Time <= 2359);



Likely, this is a bug in a recent build fixed by the latest build.
SO try the newest build, and if that doesnt work, just remove some of the date time filters to see whats the offending line.
if stuck i will help



I'm pretty sure I was on 53.48, but I closed GSB down and reopened 53.48, changed the times on L/H GUI and went to try again but now GSB wont even start when I press play and just sits there not responding. I have closed down and restarted computer, no change, GSB doesn't want to do anything and just freezes up.

admin - 28-6-2019 at 09:31 PM

Boothy, send me teamviewer details. Im free in 30 min roughly

saycem - 29-6-2019 at 12:15 AM

was there a solution to this bug.
I have already built systems and WF and for some reason getting an error when I try to save, so would like to be able to tweak the code to make it work?

admin - 29-6-2019 at 12:18 AM

Quote: Originally posted by saycem  
was there a solution to this bug.
I have already built systems and WF and for some reason getting an error when I try to save, so would like to be able to tweak the code to make it work?


thats fixed in 43.48. Older buidl gave error if save with wf full was on.
save with full not full is work around if your on an old build

saycem - 30-6-2019 at 11:26 PM

Attempted to build systems on Silver with GC secondary on 29,30,31 but got very poor results. Is there something off in my settings that you can see?


setting1.JPG - 86kBsetting2.JPG - 58kBsetting3.JPG - 61kBsetting4.JPG - 91kB

admin - 1-7-2019 at 12:32 AM

Quote: Originally posted by saycem  
Attempted to build systems on Silver with GC secondary on 29,30,31 but got very poor results. Is there something off in my settings that you can see?


I think time was 730 to 1330 exchange time
you have $58 r/t slippage. make it zero at this stage.
If thaat doesnt fix it, send me teamviewer details

jptann - 7-7-2019 at 12:36 PM

Hi Peter:

I opened the new RM and it brought up the first worker, 54.10. Unfortunately, the Op settings were not what I want to trade. I changed the Op settings to the data files I want to use and some other parameters. Once I entered the OP settings and checked the App settings, I saved them under a new name. Now the issue is when the RM opens another worker, it may or may not have the same op settings, so I have to re-input them.

In prior versions, you used the manager as a place where we could enter our Op and App settings, and then any subsequent workers would take the manager's value as long as they were the same version. Now they open up with some other Op version and I have to re-enter them again.

Could you please add a way to save the Op and App settings I want to use, as a default for the user (not the system default) and the program would go to that default when we opened a worker or any additional workers.

Another way would be to use the manager and use that as the default setting and each worker would get their App and OP settings from that worker, they way it used to work.

Finally, the new version is not generating any systems and I have the easiest conditions that I can set, minimum -1, minimum PL -1000000, etc.

Of course I may be doing something wrong so I'm off to look at your new documentation to see if I need to do something different.

Thanks in advance

admin - 7-7-2019 at 05:17 PM

Quote: Originally posted by jptann  
Hi Peter:

I opened the new RM and it brought up the first worker, 54.10. Unfortunately, the Op settings were not what I want to trade. I changed the Op settings to the data files I want to use and some other parameters. Once I entered the OP settings and checked the App settings, I saved them under a new name. Now the issue is when the RM opens another worker, it may or may not have the same op settings, so I have to re-input them.

In prior versions, you used the manager as a place where we could enter our Op and App settings, and then any subsequent workers would take the manager's value as long as they were the same version. Now they open up with some other Op version and I have to re-enter them again.

Could you please add a way to save the Op and App settings I want to use, as a default for the user (not the system default) and the program would go to that default when we opened a worker or any additional workers.

Another way would be to use the manager and use that as the default setting and each worker would get their App and OP settings from that worker, they way it used to work.

Finally, the new version is not generating any systems and I have the easiest conditions that I can set, minimum -1, minimum PL -1000000, etc.

Of course I may be doing something wrong so I'm off to look at your new documentation to see if I need to do something different.
Thanks in advance


Hi JPtann
You are working to hard to fix things. Workers will run the correct settings when they talk to your manager. There should never be a need to change data settings etc.
As for no systems, first thing to look at is fitness zero?
see https://trademaid.info/gsbhelp/Optimization2.html


Second thing to do is look on workers, right click top diag - and see what systems are being made.
I will look via teamviewer shortly

Daniel UK1 - 12-7-2019 at 02:48 AM

Peter, Thanks for the new update..the new PSS score, that you write about here, https://trademaid.info/gsbhelp/Systemstatistics-degradation....
Shall i understand it as, in order to see this column numbers, you need to save and load the results to statistics?
And then as in the example if you change to see the dates from 2015 to 2018 OOS, i need to save and loead the statistics again..
BUT this time i get a normalisation of the previous number in the PSS column and the latest one?

Is it possible to do this by Macro?


admin - 12-7-2019 at 05:46 PM

Quote: Originally posted by Daniel UK1  
Peter, Thanks for the new update..the new PSS score, that you write about here, https://trademaid.info/gsbhelp/Systemstatistics-degradation....
Shall i understand it as, in order to see this column numbers, you need to save and load the results to statistics?
And then as in the example if you change to see the dates from 2015 to 2018 OOS, i need to save and loead the statistics again..
BUT this time i get a normalisation of the previous number in the PSS column and the latest one?

Is it possible to do this by Macro?


yes, it all can be run by macro. The feature works fine, but its not yet proven if its helpful.
If (optional) you save the second time, you get the normalized results of both OOS
periods.

Im also going to add OOS fitness, which I suspect is more useful than oos degradation.

My logic is systems with low IS fitness, even if they degrade little - these may be not as good as higher oos fitness that degrade a bit more.

This is all unproven at this stage.

Error message

Daniel UK1 - 16-7-2019 at 09:47 AM

I often get this error which makes it impossible to ave my systems, at lest that is my conclusion.. when i dont get this i can save systems, as soon as i get this, i cant save anything.. nothing happens when i click on save button..
Its not related to a specific market i work on.

CaptureRERROR.JPG - 18kB

admin - 16-7-2019 at 07:07 PM

Quote: Originally posted by Daniel UK1  
I often get this error which makes it impossible to ave my systems, at lest that is my conclusion.. when i dont get this i can save systems, as soon as i get this, i cant save anything.. nothing happens when i click on save button..
Its not related to a specific market i work on.


What build are you on? Can you try with save wf off. Something like this is a known bug fixed a few builds ago.

Daniel UK1 - 11-8-2019 at 08:26 AM

Anyone knows what is causing this error message?
Since about a day i canat get any workers to build or connect to my manager, i just keep getting these error messages..
I have restarted the computer several tiimes, internet works great and everything else to, its just that the managers just keep seraching for workers, and nothing happens.. is this a general problem or issue on my pc only ?

Issue seems related to my RM, since it does not seem to be able to connect and i get these error messages without even starting a manager.


EDIT, I solved the problem myself, was internal network error that caused it..
Captureerrorgsb.JPG - 24kB

saycem - 23-8-2019 at 02:13 AM

I feel like I'm getting a lot of very similar systems... What is the best way to minimise duplicity? I have Max params sig dupl at 0 from your previous recommendation you have given me which I suspect is reason. What would default be? Is 0 best?

duplicates.JPG - 131kB

admin - 23-8-2019 at 02:21 AM

Quote: Originally posted by saycem  
I feel like I'm getting a lot of very similar systems... What is the best way to minimise duplicity? I have Max params sig dupl at 0 from your previous recommendation you have given me which I suspect is reason. What would default be? Is 0 best?


GSB can have very tight control over this, but near duplicates are good.
Reason is if a system is good, you get lots if very similar systems. ie might be same indicators, different oscillators.
Then use the family feature, and see what the family with the most memeber are.
Many members in the same family is good, but likely jsut take the first one (highest fitness)
see this post
https://trademaid.info/gsbhelp/Createupdatefamily.html

saycem - 23-8-2019 at 03:01 AM

thanks where do I turn on the family column?

admin - 23-8-2019 at 03:36 AM

Quote: Originally posted by saycem  
thanks where do I turn on the family column?


Only works from favorites a,b,c,d
select systems, right click, create family.
The coulomb will appear

appengineer - 2-9-2019 at 11:42 AM

Hi Peter,

If I have multiple macros defined on app settings and "on opt completed" set to false, is it possible to run individual macros manually, one at a time?

Thanks

admin - 2-9-2019 at 04:48 PM

Quote: Originally posted by appengineer  
Hi Peter,

If I have multiple macros defined on app settings and "on opt completed" set to false, is it possible to run individual macros manually, one at a time?

Thanks


yes, just link a macro to one of the 6 macro buttons (and there are optional total of 3 groups of macros) then hit the small green run button to the right of m6 button

Daniel UK1 - 5-9-2019 at 07:46 AM

I am trying to develop systems during the night session, i understand it might not be optimal in terms of range.. however i would like perform a few test.. problem is that it is not working at all.... i have provided data from 1645 to 0930 Eastern time.. data file is good and checked... i set the Trading period Time: to 1645 - 0930..

I have set the moc to 0930 in trade data...

BUT no systems generated... i assume its something i am missing in terms of setting...

What setting do i need to allow systmes to be built for overnight sessions?

Thanks for helping out

Daniel UK1 - 5-9-2019 at 07:47 AM

I am trying to develop systems during the night session, i understand it might not be optimal in terms of range.. however i would like perform a few test.. problem is that it is not working at all.... i have provided data from 1645 to 0930 Eastern time.. data file is good and checked... i set the Trading period Time: to 1645 - 0930..

I have set the moc to 0930 in trade data...

BUT no systems generated... i assume its something i am missing in terms of setting...

What setting do i need to allow systmes to be built for overnight sessions?

Thanks for helping out

admin - 5-9-2019 at 04:37 PM

Quote: Originally posted by Daniel UK1  
I am trying to develop systems during the night session, i understand it might not be optimal in terms of range.. however i would like perform a few test.. problem is that it is not working at all.... i have provided data from 1645 to 0930 Eastern time.. data file is good and checked... i set the Trading period Time: to 1645 - 0930..

I have set the moc to 0930 in trade data...

BUT no systems generated... i assume its something i am missing in terms of setting...

What setting do i need to allow systmes to be built for overnight sessions?

Thanks for helping out

email me teamviewer details. Ive never done this before.

appengineer - 8-9-2019 at 02:00 PM

Hi Peter,

I have noticed for ES system a Buy to Open / Sell to Open signal being generated at 15:00 hrs.
When we define session time 08:30 - 15:00, does "Setexitonclose" close all positions at 15:00 hrs or 15:15 hrs?

Thanks

appengineer - 8-9-2019 at 02:18 PM

Quote: Originally posted by appengineer  
Hi Peter,

I have noticed for ES system a Buy to Open / Sell to Open signal being generated at 15:00 hrs.
When we define session time 08:30 - 15:00, does "Setexitonclose" close all positions at 15:00 hrs or 15:15 hrs?

Thanks


Ignore my previous post

appengineer - 8-9-2019 at 08:27 PM

Hi Peter,

When live trading ES system (.i.e for current contract month), do you use @ESU19 or @ESU19.D?

Thank you

admin - 8-9-2019 at 08:33 PM

Quote: Originally posted by appengineer  
Hi Peter,

When live trading ES system (.i.e for current contract month), do you use @ESU19 or @ESU19.D?

Thank you

@ESU19.D is the day seesion only. Thats what i use.
I think @ESU19 might also trade some holidays that still have data

best also stick with what the system was made on. (my data .d)
exceptions may be I think it was @ym where the data goes back another year or so compared to @ym.d

Carl - 23-9-2019 at 10:12 AM

Hi Peter,

Today I used GSB's "top performing indicators" functionality for the first time.

Please explain, how does this work precisely?

I have been running a 50% IS + 50% OOS test.
After this I activated the "top performing indicators" functionality.

But what do the numbers show?
1. What period is taken into account? IS? OOS? Full period? Something else?
2. And what metric is used to calculate the list? NP/DD? Pearson? Fitness function?

Thanks.

appengineer - 30-9-2019 at 08:54 PM

Hi Peter,

Question on market / other time frame verification, what metrics does GSB analyze when it reports 8/8?

Thank you

admin - 30-9-2019 at 08:57 PM

Quote: Originally posted by appengineer  
Hi Peter,

Question on market / other time frame verification, what metrics does GSB analyze when it reports 8/8?

Thank you

From memory is the degration in fitness, but I will check with programmer later

Carl - 1-10-2019 at 01:09 AM

I think I remember that GSB compares per verification item if it passes the verification filter. This is set in the left menu.

admin - 1-10-2019 at 01:18 AM

Quote: Originally posted by Carl  
I think I remember that GSB compares per verification item if it passes the verification filter. This is set in the left menu.


Hi Carl, you are correct. I was thinking of VSS. However my own tests showed VS was not helpful, but VSS was.

VSS is % degradation of fitness, VS is number of systems that pas the filter under VS performance metrics

Daniel UK1 - 1-10-2019 at 04:31 AM

Peter, a question about verification and VS, do you verify straight after the build for example the top 250 when out into D, or.. do you verify after the walkforward has been done?

Perhaps i am stuck in a thougherror... but i am unsure of if a verification done before WF is done, or a verification after WF is done, would be any different ? and if anything would change if i set USE WF Cur Params WFP, to YES for all systems before Verification...

Thanks

admin - 1-10-2019 at 06:11 AM

Quote: Originally posted by Daniel UK1  
Peter, a question about verification and VS, do you verify straight after the build for example the top 250 when out into D, or.. do you verify after the walkforward has been done?

Perhaps i am stuck in a thougherror... but i am unsure of if a verification done before WF is done, or a verification after WF is done, would be any different ? and if anything would change if i set USE WF Cur Params WFP, to YES for all systems before Verification...

Thanks


Good question, im donig verification before wf.
You can test if it works better after, but I felt before did. Let us know your results

Daniel UK1 - 2-10-2019 at 04:58 AM

Quote: Originally posted by admin  
Quote: Originally posted by Daniel UK1  
Peter, a question about verification and VS, do you verify straight after the build for example the top 250 when out into D, or.. do you verify after the walkforward has been done?

Perhaps i am stuck in a thougherror... but i am unsure of if a verification done before WF is done, or a verification after WF is done, would be any different ? and if anything would change if i set USE WF Cur Params WFP, to YES for all systems before Verification...

Thanks


Good question, im donig verification before wf.
You can test if it works better after, but I felt before did. Let us know your results


I will, thank you.

Btw a proposal, after some time, and developing on several markets using noise bars, would it not be great when changing data stream, tp be able to have folders, so one could divide for example data into market folders for example ES, SI etc.. and perhaps even subfolders for timeframes.. it can get quite messy in the list after some time... what you think, good idea or bad?

Cheers

admin - 2-10-2019 at 05:00 AM

Hi Daniel,
Good idea, but we can have folders already :)

Daniel UK1 - 2-10-2019 at 05:08 AM

That sounds awesome, but how in the world do i go about creating a folder when i am in price data popup and create these magic folders :) i cant see any option when i am in the "name" list to create antyhing, when i for example right click i can only CLONE, CLONE + R or DELETE

admin - 2-10-2019 at 04:33 PM

Quote: Originally posted by Daniel UK1  
That sounds awesome, but how in the world do i go about creating a folder when i am in price data popup and create these magic folders :) i cant see any option when i am in the "name" list to create antyhing, when i for example right click i can only CLONE, CLONE + R or DELETE


Enjoy :)


newfolder.png - 30kB

Daniel UK1 - 3-10-2019 at 01:19 AM

Ahh, thats great Peter thanks. But what i actually meant was one step back, since once does not that often chose pricefiles, but more often change files one step back while testing, in the popup when you click on price data in the left side of your main GUI.

admin - 3-10-2019 at 01:40 AM

Quote: Originally posted by Daniel UK1  
Ahh, thats great Peter thanks. But what i actually meant was one step back, since once does not that often chose pricefiles, but more often change files one step back while testing, in the popup when you click on price data in the left side of your main GUI.


You mean like this?


MAP.png - 20kB

Daniel UK1 - 3-10-2019 at 04:58 AM

Quote: Originally posted by admin  
Quote: Originally posted by Daniel UK1  
Ahh, thats great Peter thanks. But what i actually meant was one step back, since once does not that often chose pricefiles, but more often change files one step back while testing, in the popup when you click on price data in the left side of your main GUI.


You mean like this?


YES :) Just imagine how nice it would be to have small folders there for the individual markets, would make things very neat

And then perhaps subfolders behind that to narrow down into barsizes you work with..

Daniel UK1 - 3-10-2019 at 08:47 AM

Quote: Originally posted by admin  
Quote: Originally posted by Daniel UK1  
Peter, a question about verification and VS, do you verify straight after the build for example the top 250 when out into D, or.. do you verify after the walkforward has been done?

Perhaps i am stuck in a thougherror... but i am unsure of if a verification done before WF is done, or a verification after WF is done, would be any different ? and if anything would change if i set USE WF Cur Params WFP, to YES for all systems before Verification...

Thanks


Good question, im donig verification before wf.
You can test if it works better after, but I felt before did. Let us know your results


In regards to Verification, i find that VSS always gets worse when doing verification with (Use WF Cur Params wfp) in YES mode.. compared to doing verification not using (Use WF Cur Params wfp) in YES mode... i cant get the logic for this, could you explain how the verification works please.... it uses either the WF settings or not depending on if you set the above to yes or no... and then it compares those fitness numbers relative to the used dates not used for training during the original build.. correct? and it only uses the dates in use when you built the systems from start....

admin - 3-10-2019 at 10:04 PM

HI Daniel, I have not looked at this closely, but seem to remember your results matched mine in this. I dont understand the reason for this and would love to know.
One think to check, is the wf stats on B,D,F better than a,c,e. (Its not always the case buy generaly they are better - or at least the Ave Trade PF etc are)
So if the WF stats are better, I would expect vss to be better on them. We both feel this may not be the case.

Regardless if WF parameters are used, I think if the stability score is higher on the system, its a better system.
What your doing shows you have a good grasp on the finer details if whats going on very deep down. Well done.

appengineer - 4-10-2019 at 05:00 AM

Hi Peter,

Just to confirm, we run verify on the top 250 systems (ordered by fitness - A column) or is it top 1000?

Thank you

admin - 4-10-2019 at 05:03 AM

Quote: Originally posted by appengineer  
Hi Peter,

Just to confirm, we run verify on the top 250 systems (ordered by fitness - A column) or is it top 1000?

Thank you

not quite. I use the macro (loaded on m1) that also filters pf and pearsons too.
Use the macro defaults

appengineer - 4-10-2019 at 04:33 PM

Quote: Originally posted by admin  
Quote: Originally posted by appengineer  
Hi Peter,

Just to confirm, we run verify on the top 250 systems (ordered by fitness - A column) or is it top 1000?

Thank you

not quite. I use the macro (loaded on m1) that also filters pf and pearsons too.
Use the macro defaults


Hi Peter,

I am talking the manual step where we do have to right click and select verify?
I do use the macros (m1) for selecting top 250,500,1000 systems and WF.

Thank you

admin - 4-10-2019 at 04:43 PM

Hi appengineer
Note the dates are on 2015. (macro2)
So thats done with nth all, oos off, wfparma off (all done in macro2)


all 250 systems in fav a (and wf) are highlighed.
Right click, verify

Then I run macro6, which puts the top 91 systems into fav D
then run macro 3, which does the stats


verify.png - 132kB

appengineer - 4-10-2019 at 04:47 PM

Quote: Originally posted by admin  
Hi appengineer
Note the dates are on 2015. (macro2)
So thats done with nth all, oos off, wfparma off (all done in macro2)


all 250 systems in fav a (and wf) are highlighed.
Right click, verify

Then I run macro6, which puts the top 91 systems into fav D
then run macro 3, which does the stats


Thanks Peter, we are on the same page.

admin - 4-10-2019 at 04:51 PM

Thanks also for the thanks tick. Its a new feature in the forum. In my head, I have enough content now for the next video. I was stuck on a lot of things.
Next video is really over due, but spectacular amount of time on this spent with very slow progress

appengineer - 4-10-2019 at 06:08 PM

Hi Peter,

What does it mean when so many systems are grouped into the same family(duplicates?), I have family with 80 members. see attached.

Thank you

ES_Stats.PNG - 29kBfamilies.PNG - 16kB

admin - 4-10-2019 at 07:20 PM

Quote: Originally posted by appengineer  
Hi Peter,

What does it mean when so many systems are grouped into the same family(duplicates?), I have family with 80 members. see attached.

Thank you

Thats good, it means its popular and so the most robust system that works on many parameters.

I used to think duplicates was bad (but systems 100% the same are removed)
but now I see it as very good. It means much more robust system
GSB has great control over duplicates, and we now have this set to be very loose.
If its tight, you also wont get 50,000 systems.

Take your favorite member from the family. Likely the top few

appengineer - 5-10-2019 at 07:21 AM

A small tweak on the entry mode, enabled CrossCloseD.

Shows positive results too.

closeD.PNG - 12kBES_stats_closeD.PNG - 28kBfamilies_closeD.PNG - 22kB

admin - 6-10-2019 at 05:24 PM

Quote: Originally posted by appengineer  
A small tweak on the entry mode, enabled CrossCloseD.

Shows positive results too.


your doing some good work.

I found ncc (no conflicht cross) and aic (any indicator cross)
gave good results in pre 2015, but singificantly worse post 2015.
Not sure why.

these are important gsb options to fine tune. More in the next video

Daniel UK1 - 8-10-2019 at 02:07 AM

Peter, in the latest build there is a dropdown on the left hand side, where you can chose GSB, EasyL, or Ninja for session close... could you elaborate a bit on this?
Ninja, well thats understandable .... however running MC i am confused.. should i use GSB or EasyL?

Thanks

admin - 8-10-2019 at 02:39 AM

Quote: Originally posted by Daniel UK1  
Peter, in the latest build there is a dropdown on the left hand side, where you can chose GSB, EasyL, or Ninja for session close... could you elaborate a bit on this?
Ninja, well thats understandable .... however running MC i am confused.. should i use GSB or EasyL?

Thanks

GSB IS TS/MC code to exit at moc.

EL is setexitonclose,
Its in the comments in the very bottom help box of gsb

admin - 8-10-2019 at 03:32 AM

Quote: Originally posted by Daniel UK1  
Peter, in the latest build there is a dropdown on the left hand side, where you can chose GSB, EasyL, or Ninja for session close... could you elaborate a bit on this?
Ninja, well thats understandable .... however running MC i am confused.. should i use GSB or EasyL?

Thanks

Here is the help box. All left men options should be documented here.


help.png - 33kB

Daniel UK1 - 8-10-2019 at 05:12 AM

Thank you, i can see that its supported in documentation, overlooked that.. sorry

What would be interesting to understand, is what is the actual difference between GSB logic and EL logic, i understand that its faster to build... but... what is the difference in reality that makes you recommend to only use it in backtest but not in live trading ? been curios about this even before,, so good time to ask

Many thanks

admin - 8-10-2019 at 07:13 PM

Quote: Originally posted by Daniel UK1  
Thank you, i can see that its supported in documentation, overlooked that.. sorry

What would be interesting to understand, is what is the actual difference between GSB logic and EL logic, i understand that its faster to build... but... what is the difference in reality that makes you recommend to only use it in backtest but not in live trading ? been curios about this even before,, so good time to ask

Many thanks

I think the issue is setexitonclose doesnt close your position if you trade with a live ts brokerage account

Daniel UK1 - 9-10-2019 at 03:19 AM

Thank you Peter, so in other words for a user not using TS but MC it is no difference then ?

admin - 9-10-2019 at 03:22 AM

Quote: Originally posted by Daniel UK1  
Thank you Peter, so in other words for a user not using TS but MC it is no difference then ?


I have no idea on MC execution. This issue only matters for live execution. Test on a sim account first. Safest to not use setexitonclose if live trading.


appengineer - 9-10-2019 at 07:35 PM

Quote: Originally posted by admin  
Quote: Originally posted by Daniel UK1  
Thank you, i can see that its supported in documentation, overlooked that.. sorry

What would be interesting to understand, is what is the actual difference between GSB logic and EL logic, i understand that its faster to build... but... what is the difference in reality that makes you recommend to only use it in backtest but not in live trading ? been curios about this even before,, so good time to ask

Many thanks

I think the issue is setexitonclose doesnt close your position if you trade with a live ts brokerage account


JFYI, setexitonclose also seems not to be reliable in SIM TC account, two of my strategies didn't exit on close yesterday.

admin - 9-10-2019 at 07:38 PM

Quote: Originally posted by appengineer  
Quote: Originally posted by admin  
Quote: Originally posted by Daniel UK1  
Thank you, i can see that its supported in documentation, overlooked that.. sorry

What would be interesting to understand, is what is the actual difference between GSB logic and EL logic, i understand that its faster to build... but... what is the difference in reality that makes you recommend to only use it in backtest but not in live trading ? been curios about this even before,, so good time to ask

Many thanks

I think the issue is setexitonclose doesnt close your position if you trade with a live ts brokerage account


Thanks for update.
One issue is, if close occurs say at 1500:00 and thats when the market actually closes, close wont work.
If market closes at 1515:00 then it MIGHT work. (set exit on close)
Other close ie if time >=1500 then sell to close will work if market closes 1515
JFYI, setexitonclose also seems not to be reliable in SIM TC account, two of my strategies didn't exit on close yesterday.

 Pages:  1  ..  5    7    9  ..  25