Staying In My Lane - A Read-Only Journ(al/ey) | Page 21 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

The Journ(ey)al of a Lazy Trader 0 replies

mt4 read only password question 2 replies

"Read only view" on Metatrader 10 replies

Attachments: Staying In My Lane - A Read-Only Journ(al/ey)
Exit Attachments

Staying In My Lane - A Read-Only Journ(al/ey)

  • #401
  • Apr 27, 2019 12:46pm Apr 27, 2019 12:46pm
And just to say it... I know I am complaining about Sierra a lot in my journey but it is unfair to the engineers as there are far more advanced capabilities in Sierra (ASCIL, Spreadsheet based trading etc) that I have deliberately chosen NOT to get involved with. Use of basic studies in Sierra is not appropriate for what I am trying to accomplish but it is perfect for plug-n-play strategies during design/development/testing phases of systemic strategy. It is actually far superior than any product out there. It's just the clunky and quirky in user interface design that bugs the crap out of me.
Staying in my lane...
  • #402
  • Apr 27, 2019 1:14pm Apr 27, 2019 1:14pm
Quoting VEEFX
{quote} Thanks Nik. I agree and fully understand the process of signal conversion into an edge and the difference between the two. I also agree on the expectency or Profit Factor. I don't even bother to explore an idea further if PF is less than 1. It's a losing strategy. With that said, I want to explore further with you on the process and the rigor I am putting behind it to see if there is a better/faster/easier approach to my madness. Here we go with the sequential nature of this journey (bare with me please) 1: Conversion of an idea into algo/study....
Vee,

Not surprised to see this type of rigor, but lets take a quick step back. Few very basic questions:

1- Do you have an alert/s study that will signal when your conditions are met?
2- One for long, and another for short?
3- Also, what type of bars are you using?


Edit: i also curse sierra on a weekly basis, but know its actually a great platform which in the end saves me a boatload of time.
  • #403
  • Apr 27, 2019 1:27pm Apr 27, 2019 1:27pm
Quoting Nik-Nyc
{quote} Vee, Not surprised to see this type of rigor, but lets take a quick step back. Few very basic questions: 1- Do you have an alert/s study that will signal when your conditions are met? 2- One for long, and another for short? 3- Also, what type of bars are you using? Edit: i also curse sierra on a weekly basis, but know its actually a great platform which in the end saves me a boatload of time.
1- Do you have an alert/s study that will signal when your conditions are met? Yes, Absolutely! each for buyentry, buyexit, sell entry and sell exit (using Trading based on Alert Condition study with conditions coded as alert formulas)
2- One for long, and another for short? Yes....Long/Short/Both are on menu settings and applicable to all strategies.
3- Also, what type of bars are you using? This one is tricky. I try with M1 first and foremost, if M1 is no successful then I try the following in the order before I trash the strategy>>> S30, S5, 3T/6T/9T,144T/512T V3/6/9/144/512, range/4/8/12/18/21 ticks, 10/100 volume delta bars and 3/6/9T renko reversal bar (under chart settings)
Staying in my lane...
  • #404
  • Apr 27, 2019 1:28pm Apr 27, 2019 1:28pm
V,

If you listen to Leda Braga video posted above, few times, you will get answers to your quests.

in the attached screenshot, there is a big difference between systematic and discretionary styles. you have to choose one.

CP for example, trades manually, just like Hilmy. it is hard to automate his trading ideas. i tried and it did not work. why? because besides trading signals, he uses screen time expertise, aka intuition.

if you decide to trade manually, you have to do some screen time first. 10,000 hours to be exact.
Attached Image (click to enlarge)
Click to Enlarge Name: Screenshot_20190427-123618_YouTube.jpg Size: 199 KB
  • #405
  • Apr 27, 2019 1:32pm Apr 27, 2019 1:32pm
Quoting alphadude
V, If you listen to Leda Braga video posted above, few times, you will get answers to your quests. in the attached screenshot, there is a big difference between systematic and discretionary styles. you have to choose one. CP for example, trades manually, just like Hilmy. it is hard to automate his trading ideas. i tried and it did not work. why? because besides trading signals, he uses screen time expertise, aka intuition. if you decide to trade manually, you have to do some screen time first. 10,000 hours to be exact. {image}
Agree.. even if I succeed in finding a suitable algo, I will always trade manual or monitor for atleast a 3month window. The whole idea behind systemic engineering is knowing first, execution next then comes validation and reevaluation/tweaks. It's never a ready/set/go/ travel to bahamas... i know that lol
I have also said this... before I quit, I need 3 full years of income generation as proof and atleast 2 yrs of profits saved prior to going full time. I am not cutout to earn income on a month to month basis that I can rely on. I am too weak bro when it comes to preserving financial security especially when I have worked my ass off to get to where I am in life :-
Staying in my lane...
  • #406
  • Apr 27, 2019 1:40pm Apr 27, 2019 1:40pm
Quoting VEEFX
{quote} 1- Do you have an alert/s study that will signal when your conditions are met? Yes, Absolutely! each for buyentry, buyexit, sell entry and sell exit (using Trading based on Alert Condition study with conditions coded as alert formulas) 2- One for long, and another for short? Yes....Long/Short/Both are on menu settings and applicable to all strategies. 3- Also, what type of bars are you using? This one is tricky. I try with M1 first and foremost, if M1 is no successful then I try the following in the order before I trash the strategy>>> S30,...
I am pretty sure I will go with only one algo which is currently not listed on my screen as a menu item to turn on/off. Visually, it has an edge written all over it but coding is a bit tricky. I now have just one strategy enabled. Disabled everything just to try this particular one out with tick-based backtest (i am done with stupid and inaccurate standard replay)

This one should "reverse" on the orange dots to catch reversals (sort of always-on strategy but running only for 15 mins max with stop trading when 3 wins or 1loss is met) whenever I can spare some time to trade (retarded thinking I know!)

EDIT: I think it has something to do with the reversal setting... my weekend gig to get this resolved :-(
Attached Images
Staying in my lane...
  • #407
  • Apr 27, 2019 1:50pm Apr 27, 2019 1:50pm
It is also worth mentioning that I am ONLY using Market Orders in ALL my algos. I don't believe in Limit or pending order types. Never have and never will. Same for Stop Loss and same for Take Profit. My SL and TP settings in the trade window is set to large number... protective stop really against connectivity drops etc!

I wait for conditions to occur and then trigger a simulated limit order via market order.... frankly, I don't even know how to systematically submit limit orders ROFL
Staying in my lane...
  • #408
  • Apr 27, 2019 1:53pm Apr 27, 2019 1:53pm
Quoting VEEFX
{quote} 1- Do you have an alert/s study that will signal when your conditions are met? Yes, Absolutely! each for buyentry, buyexit, sell entry and sell exit (using Trading based on Alert Condition study with conditions coded as alert formulas) 2- One for long, and another for short? Yes....Long/Short/Both are on menu settings and applicable to all strategies. 3- Also, what type of bars are you using? This one is tricky. I try with M1 first and foremost, if M1 is no successful then I try the following in the order before I trash the strategy>>> S30,...
V,

Excellent. Few more points and questions before getting started.

1- At this stage, forget about any buyexit, sellexit conditions and alerts. Regarding the buyentry, sellentry, how many do you have? Are they coded in their own separate study?
2- For the purpose of simplicity, do you mind if we proceed with range bars for now? In my experience 8t range bars are a good starting point for NQ(the beast)
3-Lets also put aside reversal conditions as this complicated the data.
  • #409
  • Edited 2:32pm Apr 27, 2019 2:19pm | Edited 2:32pm
Quoting Nik-Nyc
{quote} V, Excellent. Few more points and questions before getting started. 1- At this stage, forget about any buyexit, sellexit conditions and alerts. Regarding the buyentry, sellentry, how many do you have? Are they coded in their own separate study? 2- For the purpose of simplicity, do you mind if we proceed with range bars for now? In my experience 8t range bars are a good starting point for NQ(the beast) 3-Lets also put aside reversal conditions as this complicated the data.
I am zero-ing in on just one setup... so only one buy entry and one sell entry alert condition study... everything else is "disabled". took me 30 mins to disable them all lol

range bars is fine but I do have some wick measurements and calculations from mid-point on a variable candle size etc to factor into my trend determination.let me see if I can pull out an older collection that worked well on range bars

EDIT: I take that back.... Range bars will work if I show it as Candlesticks
EDIT: Each buy and sell as it's own study (per Sierra. it has to be one order type per study... so two in all)
Attached Image
Staying in my lane...
  • #410
  • Apr 27, 2019 2:43pm Apr 27, 2019 2:43pm
Quoting VEEFX
{quote} I am zero-ing in on just one setup... so only one buy entry and one sell entry alert condition study... everything else is "disabled". took me 30 mins to disable them all lol range bars is fine but I do have some wick measurements and calculations from mid-point on a variable candle size etc to factor into my trend determination.let me see if I can pull out an older collection that worked well on range bars EDIT: I take that back.... Range bars will work if I show it as Candlesticks {image}
Great, now that we have the signal isolated the testing can begin.

Now, i know u dislike TP, BE, SL and all that conventional stuff, but bear with me. Run a single week backtest with the following parameters:

-single entry
-single contract
-Hard TP @ +16
-Hard SL @ -16
-No BE or other trade management
-Trade only your session NYO+X

The +16/-16 are just a multiple of the range bar size, feel free to increment to 24,32,40 if visually that makes more sense.

Is this doable? I know its basic, but the fun comes later, not now.


Edit: this is not real testing, it is pre-testing so i can learn a few things about your system such as frequency, etc.
  • #411
  • Apr 27, 2019 2:56pm Apr 27, 2019 2:56pm
careful vee, i think nik is using you to backtest his strategy lol!
  • #412
  • Apr 27, 2019 3:23pm Apr 27, 2019 3:23pm
Here's a little perspective. Lewis was one the top dog in the s&p pit back in the day. He spent a brief time as an online trader. Look at his charts..it's laughable to me. I mean this is like 3-4 years ago, so it's pretty recent.

The funny thing is, no quant picked him up to help him adapt to the algo trading world (or even consider partnering with him due to his experience).

I don't think he's trading anymore..he's working in the cyber security field

https://www.linkedin.com/in/lewis-j-borsellino-aa42496

Inserted Video
  • #413
  • Apr 27, 2019 3:24pm Apr 27, 2019 3:24pm
Quoting Nik-Nyc
{quote} Great, now that we have the signal isolated the testing can begin. Now, i know u dislike TP, BE, SL and all that conventional stuff, but bear with me. Run a single week backtest with the following parameters: -single entry -single contract -Hard TP @ +16 -Hard SL @ -16 -No BE or other trade management -Trade only your session NYO+X The +16/-16 are just a multiple of the range bar size, feel free to increment to 24,32,40 if visually that makes more sense. Is this doable? I know its basic, but the fun comes later, not now. Edit: this is not...
will do but first please help me figure this out as my OCD is through the roof and frustration beyond words can describe. I think something is messed up with my Sierra. I have re-installed atleast a dozen times, edit>delete and download data atleast 50 times and it still crashes is "tick-based backtest" mode after about 1 week of test. Sierra freezes even when I am running only in 5 or 10 speed. Its runs like a charm for few days and then forces me to use task manager to terminate session and when I return, I get this shit.... over and over and over again. I just don't get this. I simply cannot continue if my backtest results are fake/unreliable. I am paid royally to be "OCD" in finding/fixing problems at work but when it comes to Sierra, I am just utterly miserable.

Could you please check the filesize on your install? Mine is over 3GB with just 1/1/19 till date range using 8tick range bar for the entire NYSE session. Is this normal? I am using the Contineous Contract Rollover setting.

EDIT: This still happens after I have pretty much disabled ALL entries except two for one algo, ALL TP and SL settings except 4 and almost every study is "hidden" or disabled that are no longer in use. Just the ones I use to display information on my screen are enabled. Still crashes after a while of running.
Attached Image(s) (click to enlarge)
Click to Enlarge Name: Image1.png Size: 62 KB
Click to Enlarge Name: Image3.png Size: 38 KB
Attached Image
Staying in my lane...
  • #414
  • Apr 27, 2019 4:03pm Apr 27, 2019 4:03pm
Quoting VEEFX
{quote} will do but first please help me figure this out as my OCD is through the roof and frustration beyond words can describe. I think something is messed up with my Sierra. I have re-installed atleast a dozen times, edit>delete and download data atleast 50 times and it still crashes is "tick-based backtest" mode after about 1 week of test. Sierra freezes even when I am running only in 5 or 10 speed. Its runs like a charm for few days and then forces me to use task manager to terminate session and when I return, I get this shit.... over and over...

-Your scid file size seems perfectly normal. I assume you are recording tick data, which you should be, and those files are significantly larger. Mine are 2-4gb.

-Regarding the crashes, its difficult to know for certain, but i have a hunch. Firstly, not all studies can be enabled/disabled, including simple alert condition studies. The issue that arises with back testing systems with a very large number of studies (which you have) is the automatic looping of calculations conducted by sierra. Each new tick of data triggers a full recalculation of all your studies per bar, including the history. I am guessing that when you get to about a week of data your hardware simply cannot cope with all these calculations. This is the drawback of alert condition studies and systems. When a system is written in spreadsheets for trading, the user is given the ability to control how far bar (in bars) the calculations and re-calculation must go.

-Remember, hiding a study does not stop the calculation, it only hides the graphical output

-Only suggestion i have is to create a new chartbook with only the essential studies. Do not hide the study, remove the unnecessary studies. Then, under chart settings in days to load select a small number 2-3days, and then in use date range input the first day of your backtest. Loading less days should help the issue, but probably not solve it.
  • #415
  • Apr 27, 2019 4:18pm Apr 27, 2019 4:18pm
Quoting Nik-Nyc
{quote} -Your scid file size seems perfectly normal. I assume you are recording tick data, which you should be, and those files are significantly larger. Mine are 2-4gb. -Regarding the crashes, its difficult to know for certain, but i have a hunch. Firstly, not all studies can be enabled/disabled, including simple alert condition studies. The issue that arises with back testing systems with a very large number of studies (which you have) is the automatic looping of calculations conducted by sierra. Each new tick of data triggers a full recalculation...
Appreciate it Nick. It could be the auto looping. Before I go into removing stuff, I need one more confirmation on this. I am always showing only the bars that I want to test on my chart. Could the freeze be related to the "skip empty period" setting in replay? I didn't think so or else it would always freeze at end of day1. I am going to try the number of days route with market depth setting of 10 with just 15 mins of NYSE Open bars.

EDIT: I must add, I have the latest Intel chip, 500GB SSD, 4GB graphics card ( latest Gaming laptop) and when Sierra runs in backtest (before, during/after freeze) the CPU is never above 15%, Memory is never above 30% and disk I/O is just not there at all because of Sierra running on SSD. I think it is related to some study incorrectly coded or causing conflict within the sequencing of execution of each study. I will get rid of all this shit soon as that was always my plan... develop algo, test, proven edge and strip out the rest to make it light-weight before going live
Attached Image (click to enlarge)
Click to Enlarge Name: Image1.png Size: 39 KB
Staying in my lane...
  • #416
  • Apr 27, 2019 4:41pm Apr 27, 2019 4:41pm
Quoting VEEFX
{quote} Appreciate it Nick. It could be the auto looping. Before I go into removing stuff, I need one more confirmation on this. I am always showing only the bars that I want to test on my chart. Could the freeze be related to the "skip empty period" setting in replay? I didn't think so or else it would always freeze at end of day1. I am going to try the number of days route with market depth setting of 10 with just 15 mins of NYSE Open bars. EDIT: I must add, I have the latest Intel chip, 500GB SSD, 4GB graphics card ( latest Gaming laptop) and...
V,

I am not familiar with the skip empty periods setting, but also have my doubts such a simple setting could be the cause.

New hardware certainly helps, but if the looping issue is the actual cause then you should think of it like a memory leak, which will crash all hardware eventually, no matter the amount of physical ram.
  • #417
  • Apr 27, 2019 4:44pm Apr 27, 2019 4:44pm
Quoting hilmy83
careful vee, i think nik is using you to backtest his strategy lol!
First rule of trading forums, never trust anyone's backtesting results, only your own.
  • #418
  • Edited 7:02pm Apr 27, 2019 4:48pm | Edited 7:02pm
Quoting alphadude
magnumfreak: well to the point. good to see more engineers here the key secret to automated trading is simplicity and the law of large numbers. also; you need to revise your target returns. your goal is not possible. if you want to live to trade another day; aim at no more than 10% DD and no more 100% a year. try alphaomega strategy mentioned in his thread: simple mean reversion strategy. it works well on eurusd. listen to leda braga on this youtube. somewhere in the middle she has answers to your questions: https://youtu.be/TvZG9iQ2e5Q...
Thanks for posting Alpha, some nice tidbits in there . . . what struck me the most is when she said (paraphrasing here): for intraday opportunities we look at volume and the orderbook. Also known as, Orderflow :-)
  • #419
  • Apr 27, 2019 4:57pm Apr 27, 2019 4:57pm
Quoting Nik-Nyc
{quote} V, I am not familiar with the skip empty periods setting, but also have my doubts such a simple setting could be the cause. New hardware certainly helps, but if the looping issue is the actual cause then you should think of it like a memory leak, which will crash all hardware eventually, no matter the amount of physical ram.
deleting all the junk...might take me a while
Staying in my lane...
  • #420
  • Apr 27, 2019 5:09pm Apr 27, 2019 5:09pm
Quoting Nik-Nyc
{quote} V, I am not familiar with the skip empty periods setting, but also have my doubts such a simple setting could be the cause. New hardware certainly helps, but if the looping issue is the actual cause then you should think of it like a memory leak, which will crash all hardware eventually, no matter the amount of physical ram.
I have a dozen+ trailing SL and TP studies disabled. Deleted those also?
Staying in my lane...
1 trader viewing now