General build settings for different instruments and other questiones

Hi all!

I am new to SQ and have just completed the video course, building a strategy that was profitable in OOS, in 2019. Unfortunately, I forgot that we are in 2021, and after retesting said strategy, it turned to be unprofitable. So I will need to try again. (I’ve actually done it, a lot, already).

I don’t really know where to begin with my questions, so I’ll just shoot.

1. When generating strategies, with genetic algorithms as in the video tutorial – are the accepted strategies getting gradually better and better? Is it worth the extra wait, to go from say 1500 accepted strategies, to 3K? or will the strategies in that range be basically of the same quality as the ones that came before?

2. The tutorial is 2 years old. is everything in it still applicable?

3. Does anybody have any specific values to go after, when it comes to default

spreads,

slippages,

settings for pips to SL and TP in different markets and timeframes

settings for amount of trades t oexpect on the different timeframes (both in general and for specific instruments perhaps?)

settings to use in the Ranking section, depending on instrument and TF?

Stuff like that.

If you have more experience than a week or so, like me, you probably have more to go by. CAn you help me?

I mean, I could just try all the different combos randomly, but I am not SQ, and cannot come up with 340000 combinations in minutes. Also, my computer isn’t the newest one, so I get frustrated when trying something out that results in nothing :p

Thanks in advance for all the help I can get!

Hello, to answer you questions

1. “are the accepted strategies getting gradually better and better?” – basically yes, that’s the whole purpose of genetic programming. The results should improve in time as the build progresses. Especially when you enable additional verification tests (cross-checks and other metrics)
2. yes, a slight modification of the user-interface can exists but all the functions and principles will be valid. If you find significant differences let us know please
3. as for various settings such as spreads, slippages, settings for pips to SL and TP in different markets and time-frames – these values are strictly dependent on the market and timeframe used. It is also dependent on the current market regime and cycle. Is there a period of high volatility? Low volatility? This is what matters now but can be different in a year or two. What market you focus on now primarily?

US30 , SP500  ? any hints?

Cheers

Kris

I am sorry for this slow ass response, but I’ve been in bed, sick as an animal for like a week now…

anyway… I believe I need to ask something different.

I guess I could figure out spreads and slippages by myself. There are statistics readily available on different broker’s pages…

Thing is, that I’ve tried to make some EAs (actually even one would be nice) using the EURUSD 1H video tutorial, but for other pairs (mainly GU and UC) but I’m having a really hard time producing anything…

So I am wondering, how I can ” fine tune” the settings in the tutorial, so that the strategies that I produce are still of high quality, but there are actually some getting made…

Like, earlier this week, I used the exact settings as on that tutorial, to build 3500 4H strategies from the GU pair…

But later on, when I used the Retest process, it was obviously too harsch on the strategies, and left me with nothing… what things are OK to leave out? what things should be taken more lightly or heavily depending on what time frame I use for the strategies?

Or is it just that I should just build another 3,5K strategies, and gope for the best? is there some “magic number” of strategies, that almost always yields at least one good one?

I’ve made a clumsy cheat sheet of the retest methods used in the tutorial, and I’ll attach it here. maybe somebody will have a use for it… But I’ll also post it here, so that maybe, somebody could instruct me on some “convenient practices”… Stuff to have in mind, or change or tweak, when working with it… especially when it comes to different timeframes on FOREX.

I am also interested in indicines and stocks. but one thing at a time I guess…

Second OOS test
Settings > DATA > Precission – change from “Selected time frame” to “1 minute data tick simulation”.
Set dates to “from 2003.5.5” (from the beginning) “to 2010.12.31”
Turn off “OOS”
Cross Checks > turn off “Higher backtest precision”
Ranking > only leave “Profit factor” (1.1) and “Ret/DD Ratio”(2.4) (0.3/year. 8 x 0.3 =2.4)

Slippage test
in “DATA”, change dates to 2003.05.05 – 2016.12.31
set slippage to 3

In Ranking, change “Ret/DD Ratio” to 4.2
Testing on another market
Put the slippage back to 1 pip
Change from EURUSD to GBPUSD

in Ranking, only leave Profit factor > 1.1.

Test on different time frames
Switch market back to EURUSD
Change timeframe to M30
test
Switch timeframe to H4
Change profit factor to 1
Test

MC trades manipulation
Set timeframe back to 1H
Cross Chacks > MC trades manipulation
# simulations = 200, “Ranodomize trades order: on”, Method: Exact.
Filtering > Ret/DD ratio (MC trades manipulation conf. lvl 95% > 50% of Ret/DD ratio (Main data))

Run test

MC Randomize trades order using resampling
Same MC test as before
Switch from “Exact” to “Resampling”

Filtration > Set that we require only 25% instead of 50%.
Test

MC Randomly skip trades
Same MC test as before
Switch from “Randomize trades order” to “Randomly skip trades” (10% of trades)
Change back to 50% in Filtration.

Test

MC Randomize strategy parameters

Turn off “MC trades manipulation”

Turn on “MC retest methods”

Turn off all except “randomize strategy parameters”

Set probability to 20%. Max change to 30%
# of sims to run = 200

Filtering > Ret/DD ratio (MC retest methods conf. lvl 95% > 50% of Ret/DD ratio (Main data))

test

MC randomize history data

In “DATA”, change precission to “Selected timefram only”
Cross Checks > MC retest methods > Switch to “Randomize history data”. Everything else off

Probability = 30%
Max change = 10%

test

MC randomize slippage and spread

In “DATA”, switch to “1 min data tick simulation”
In Cross Checks, MC retest method, check “Randomize slippage” value. EurUSD 2 pips spread = 0-5 pips slippage

Set # of sims to 10
test

Switch to “Randomize spread” (same spread as slippage)
Test
Overall MC test

Still in MC retest methods – Set slippage to 3
Set spread to 1-4

Set # of sims to 200
in “DATA”, change precission to “Selected timeframe only”

Test

Filtering – filter out strategies that look very similar
Last OOS test – in “DATA” – reset the dates and add OOS from 01.01.2017 and forward. Save the ones that have Profit factor >1