Reply

Problem With Settings Resulting in Strange Initial Results

6 replies

qattack

Customer, bbp_participant, community, 44 replies.

Visit profile

7 years ago #116608

I am following along with Threshold’s “Settings” video for the third time. (https://www.youtube.com/watch?v=ji4HArEPEjg&t=323s)

 

Each time that I do so, I encounter strange results in the initial Generation that persist into subsequent generations.

 

I am using 10 years of Dukascopy data, 2003-2013, divided equally between IS and OOS.

 

If it matters, I am using Test Precision of Selected Timeframe Only.

 

Many of the results (around 10-20% or so) show very few #/trades IS (From 1-75 or so), with the OOS #/trades always being zero (no data). The trades executed IS are nearly always within the first partial year of 2003, and the equity graph stops at that point.

 

Subsequent generations inherit these traits and several of the second generation results show zero trades IS.

 

I’ve tried to carefully monitor what settings I was changing the past couple times I watched this videos, but I cannot eliminate this problem without reloading a settings file that came with SQ.

 

I messed around with “Min distance of order from price”, ATR parameters, and unchecked a few Indicators in Building Blocks.

 

What causes this?

0

qattack

Customer, bbp_participant, community, 44 replies.

Visit profile

7 years ago #142457

I’ve discovered the cause of this problem, but I don’t know why…

 

When I check “SL Is Required” the problems begin.

 

As an example, if I go to the HOME tab and load “Moving Average Strategies Evolution” and begin to build strategies, it works fine. But when I change only the “SL Is Required” checkbox, many of the strategies are generated with the above problem.

 

It doesn’t matter what the ATR multiple is or whether SL is based on ATR or fixed pips.

 

I cannot find anything in common with these problem strategies that correctly-generating strategies do not have.

 

Here is one such strategy:

 

**************

 

——————————————————————–
Pseudo Source Code of Strategy 1.5
  with parameter names.
 
  Generated by StrategyQuant version 3.8.2
  Generated at Tue Apr 11 01:39:00 GMT 2017
                                                          
  Tested on EURUSD, M15, 05.05.2003 – 05.04.2013
  Spread: 4.0, Slippage: 1.0, Min distance of stop from price: 0.0
——————————————————————–
====================================================================
== Entry conditions
==================================================================== 
LongEntryCondition = (ATR(93) > ATR(57))
ShortEntryCondition = (ATR(93) < ATR(57))
 
 
====================================================================
== Entry orders
====================================================================
— Long entry
if LongEntryCondition is true {
   if No position is open then Buy on open at Market;
   Stop Loss = 241 pips;
}
 
— Short entry
if ShortEntryCondition is true {
   if No position is open then Sell on open at Market;
   Stop Loss = 241 pips;
}
 
**************
I am Newb to SQ, so perhaps there is a setting that I am missing?
 
I discovered that this problem occurs only when “SL Is Required” is checked, but “PT Is Required” is UNchecked. Does this have to do with symmetrical rules?

0

qattack

Customer, bbp_participant, community, 44 replies.

Visit profile

7 years ago #142462

ThanX Notch, will do that today.

0

qattack

Customer, bbp_participant, community, 44 replies.

Visit profile

7 years ago #142489

I have completely re-downloaded and re-imported all my tick data and this problem persists.

 

I am using a different set of criteria this time — using the roboforexlabs settings downloaded directly from the lessons. The only parameter I change is to UNtick “PT Is Required.” I then get many results in the first generation that trade only for a short time period (usually ending in 2003) and produce no more trades after that time period.

0

qattack

Customer, bbp_participant, community, 44 replies.

Visit profile

7 years ago #142503

I’ve uploaded my settings file. The settings were taken directly from roboforexlab and the only change I’ve made was unticking the “PT is Required.”

 

Nearly all offending results exhibit the same pattern:

 

1. #/trades: ~10-45

2. %Wins: <10% (most <5%) [EDIT: actually, all contain only ONE winning trade…see #4 below]

3. Profit Factor: ~1.5-3.0

4. The Equity Curve is straight down until the very last trade, which is always a HUGE win.

5. Most Equity Graphs end in 2003, thought I found one that continues into my OOS period (2008)

0

qattack

Customer, bbp_participant, community, 44 replies.

Visit profile

7 years ago #142512

Max #/trades is set at 10,000. You mean in “Strategy Options”? I verified by downloading the file I uploaded. Or is there another parameter that limits #/trades?

 

I generate plenty of strategies that perform 10,000 trades.

0

qattack

Customer, bbp_participant, community, 44 replies.

Visit profile

7 years ago #142527

What I was saying was that I DOWNloaded the file that I uploaded above and then loaded those settings into my SQ.

 

In the “Strategy Options” section, after I load those settings into my SQ, it reads: Maximum Total Trades: 10000.

 

I just now realized where you got this from: <mMaxTrades>1000</maxTrades>

 

However, I assure you that when I saved these settings, my MaxTrades was set at 10,000 AND when I reload these SAME settings (even though it SAYS 1000 in the XML file) into my SQ, it again reads Max Trades: 10,000.

 

When I open the XML file, it does state 1000 maxTrades. This is true for all my saved settings in the XML files, even though they are all saved with 10,000 max trades and all produce up to 10,000 trades when reloaded into my SQ. So I don’t know what’s going here?

0

Viewing 6 replies - 1 through 6 (of 6 total)