NexusFi: Find Your Edge


Home Menu

 





Backtesting Vs Live Trades


Discussion in Traders Hideout

Updated
      Top Posters
    1. looks_one Lysakat with 6 posts (5 thanks)
    2. looks_two emini_Holy_Grail with 6 posts (1 thanks)
    3. looks_3 vitrader with 5 posts (0 thanks)
    4. looks_4 treydog999 with 5 posts (4 thanks)
      Best Posters
    1. looks_one DbPhoenix with 2.7 thanks per post
    2. looks_two juanman with 2 thanks per post
    3. looks_3 Lysakat with 0.8 thanks per post
    4. looks_4 treydog999 with 0.8 thanks per post
    1. trending_up 19,718 views
    2. thumb_up 28 thanks given
    3. group 11 followers
    1. forum 40 posts
    2. attach_file 0 attachments




 
Search this Thread

Backtesting Vs Live Trades

  #31 (permalink)
 
treydog999's Avatar
 treydog999 
seoul, Korea
 
Experience: Intermediate
Platform: Multicharts
Broker: CQG, DTN IQfeed
Trading: YM 6E
Posts: 897 since Jul 2012
Thanks Given: 291
Thanks Received: 1,039


vitrader View Post
Wow! And you still had 15% execution errors?! Trying to get into this space as a retail trader is such a ridiculous hurdle. I am trying to look at 1 min, 5 min, 15 min. charts.

First of all getting into trading and being successful is a ridiculous hurdle, for anyone. Being retail just makes it less linear, not less difficult. The only reason I say that is because if you are at an institution they will be focusing on specific products, and strategies. You will be learning about those first hand and whatever internal education or trading they give you will be geared to that. Retailers end up splashing around a lot instead of focusing on one area and getting good at it, lots of jumping. Using the time less efficiently.

Yes 15% execution errors, its part of understanding your tools. I understood that MC/Ninja/TS etc have limitations, so I designed strategies that used their strength and accessibility. Specifically I was using MC.

Are doing faster frequencies possible on these platforms? Maybe but I was not willing to put in the work to make it so. When I know that there are other better platforms (see more expensive) that were specifically geared to that type of activity. Putting me at a larger and larger competitive disadvantage as I worked at higher frequencies.

You have 2 choices, either get the right tools to design, test, and implement your 1 minute strategies, or design strategies that make the most of the tools you got. Both are inside your control. But if you are just getting started, lower frequencies (higher time frames) strategies are easier to work with and make something "viable" . takes probably 3-5x as much time to design something for the 1 minute as it does for the 1 day. IMHO

Reply With Quote

Can you help answer these questions
from other members on NexusFi?
Futures True Range Report
The Elite Circle
Deepmoney LLM
Elite Quantitative GenAI/LLM
New Micros: Ultra 10-Year & Ultra T-Bond -- Live Now
Treasury Notes and Bonds
My NT8 Volume Profile Split by Asian/Euro/Open
NinjaTrader
Online prop firm The Funded Trader (TFT) going under?
Traders Hideout
 
Best Threads (Most Thanked)
in the last 7 days on NexusFi
Get funded firms 2023/2024 - Any recommendations or word …
56 thanks
Funded Trader platforms
44 thanks
NexusFi site changelog and issues/problem reporting
24 thanks
GFIs1 1 DAX trade per day journal
22 thanks
The Program
19 thanks
  #32 (permalink)
 
vitrader's Avatar
 vitrader 
New York, New York, United States
 
Experience: Beginner
Platform: TOS
Trading: Options
Posts: 53 since Jul 2015
Thanks Given: 13
Thanks Received: 7


treydog999 View Post
First of all getting into trading and being successful is a ridiculous hurdle, for anyone. Being retail just makes it less linear, not less difficult. The only reason I say that is because if you are at an institution they will be focusing on specific products, and strategies. You will be learning about those first hand and whatever internal education or trading they give you will be geared to that. Retailers end up splashing around a lot instead of focusing on one area and getting good at it, lots of jumping. Using the time less efficiently.

Yes 15% execution errors, its part of understanding your tools. I understood that MC/Ninja/TS etc have limitations, so I designed strategies that used their strength and accessibility. Specifically I was using MC.

Are doing faster frequencies possible on these platforms? Maybe but I was not willing to put in the work to make it so. When I know that there are other better platforms (see more expensive) that were specifically geared to that type of activity. Putting me at a larger and larger competitive disadvantage as I worked at higher frequencies.

You have 2 choices, either get the right tools to design, test, and implement your 1 minute strategies, or design strategies that make the most of the tools you got. Both are inside your control. But if you are just getting started, lower frequencies (higher time frames) strategies are easier to work with and make something "viable" . takes probably 3-5x as much time to design something for the 1 minute as it does for the 1 day. IMHO

With the data I want to buy, it's only for 1 instrument. I want to test longer term timeframes along with shorter term timeframes. The thing with getting the right tools is that I have searched the Internet high and low and cannot find much of anything. Any idea?

Reply With Quote
  #33 (permalink)
Bollinez
Oxford, UK
 
Posts: 5 since Dec 2015
Thanks Given: 3
Thanks Received: 1


Well, I hope it is okay continuing this quite old discussion, as I haven't found any newer one that fits my question. I am backtesting very actively for two years now and of course noticed real-life differences. My algorithm itself performs the way it has to, but what makes me desperate real-life is primarily the rule on trading hours and days. I know it is not uncommon to exclude volatile hours like noon or to exclude very unprofitable weekdays, but I am wondering a lot whether it is a quite lazy method to exclude time periods that superficial. I tried a volatility-based exclusion, which improved my statistics, but I am asking you...

...whether you exclude time periods in backtests at all and

...did you also ask yourself whether real-life trading would require a different way to handle such periods?

Thanks,
Bollinez

Reply With Quote
  #34 (permalink)
 
rleplae's Avatar
 rleplae 
Gits (Hooglede) Belgium
Legendary Market Wizard
 
Experience: Master
Platform: NinjaTrader, Proprietary,
Broker: Ninjabrokerage/IQfeed + Synthetic datafeed
Trading: 6A, 6B, 6C, 6E, 6J, 6S, ES, NQ, YM, AEX, CL, NG, ZB, ZN, ZC, ZS, GC
Posts: 3,003 since Sep 2013
Thanks Given: 2,442
Thanks Received: 5,863


Bollinez View Post
Well, I hope it is okay continuing this quite old discussion, as I haven't found any newer one that fits my question. I am backtesting very actively for two years now and of course noticed real-life differences. My algorithm itself performs the way it has to, but what makes me desperate real-life is primarily the rule on trading hours and days. I know it is not uncommon to exclude volatile hours like noon or to exclude very unprofitable weekdays, but I am wondering a lot whether it is a quite lazy method to exclude time periods that superficial. I tried a volatility-based exclusion, which improved my statistics, but I am asking you...

...whether you exclude time periods in backtests at all and

...did you also ask yourself whether real-life trading would require a different way to handle such periods?

Thanks,
Bollinez

It is not uncommon to have a 'controller' above an automated trading bot, that triggers the bot only under certain market conditions... If you can quantify the conditions to which the bot is best performing then that is a valuable approach.

Conditions can be (non exhaustive) :
- volatility
- news release
- ...

Follow me on Twitter Visit my NexusFi Trade Journal Reply With Quote
Thanked by:
  #35 (permalink)
Bollinez
Oxford, UK
 
Posts: 5 since Dec 2015
Thanks Given: 3
Thanks Received: 1


rleplae View Post
It is not uncommon to have a 'controller' above an automated trading bot, that triggers the bot only under certain market conditions... If you can quantify the conditions to which the bot is best performing then that is a valuable approach.

Conditions can be (non exhaustive) :
- volatility
- news release
- ...

Okay and thanks for your early reply. Unfortunately I have no possibility to identify news releases in NinjaTrader and I guess this needs the connection to a news feed tool in NinjaTrader right?

So do you stuck to your time-related conditions 100% in real-life? Or do you handle it on your own when there is e.g. a press conference with Draghi? I never entered or closed a position that did not follow my algorithm conditions, but I have doubts regarding time rules.

For example: For Cocoa I excluded 14:30-15:00 (GMT+1), but should I exclude it when there are no news on that day? On average it was beneficial for my backtest results, but would it be better if I could distinguish between days with and without news (on my own and not based on quantifiable conditions)?

Thank you

Reply With Quote
  #36 (permalink)
 
rleplae's Avatar
 rleplae 
Gits (Hooglede) Belgium
Legendary Market Wizard
 
Experience: Master
Platform: NinjaTrader, Proprietary,
Broker: Ninjabrokerage/IQfeed + Synthetic datafeed
Trading: 6A, 6B, 6C, 6E, 6J, 6S, ES, NQ, YM, AEX, CL, NG, ZB, ZN, ZC, ZS, GC
Posts: 3,003 since Sep 2013
Thanks Given: 2,442
Thanks Received: 5,863


Bollinez View Post
Okay and thanks for your early reply. Unfortunately I have no possibility to identify news releases in NinjaTrader and I guess this needs the connection to a news feed tool in NinjaTrader right?

So do you stuck to your time-related conditions 100% in real-life? Or do you handle it on your own when there is e.g. a press conference with Draghi? I never entered or closed a position that did not follow my algorithm conditions, but I have doubts regarding time rules.

For example: For Cocoa I excluded 14:30-15:00 (GMT+1), but should I exclude it when there are no news on that day? On average it was beneficial for my backtest results, but would it be better if I could distinguish between days with and without news (on my own and not based on quantifiable conditions)?

Thank you

There is an indicator that is doing that for you.
You can take a look at it and even tweak the code a bit...

it's something like jteconnews ...

it will take information from public website, agenda ...
those news releases are published way ahead

Follow me on Twitter Visit my NexusFi Trade Journal Reply With Quote
Thanked by:
  #37 (permalink)
Bollinez
Oxford, UK
 
Posts: 5 since Dec 2015
Thanks Given: 3
Thanks Received: 1


rleplae View Post
There is an indicator that is doing that for you.
You can take a look at it and even tweak the code a bit...

it's something like jteconnews ...

it will take information from public website, agenda ...
those news releases are published way ahead

Cool, didn't know that. I'll have a look at it. Thanks!

Reply With Quote
  #38 (permalink)
 GFIs1 
who cares
Legendary Market Wizard
 
Experience: None
Platform: nobody interested
Broker: none
Trading: forget about it
Posts: 6,921 since Feb 2012
Thanks Given: 6,181
Thanks Received: 15,564

Hi @Bollinez

Backtesting is very helpful to see occurences (at least how many) came into favor with your system.
The real work begins after the backtests.

First check for the bigger drawdowns and the "why" these happened.
Then best is to program your system to do not interfere with a "hot" finger :-)
Then take out the bad days (as discussed above).
Then omit longer phases of uncertainty in direction. I use Ichimoku cloud for this within appropriate period.
Define your system with rules and stop loss settings. SL need to be optimized!

Having all done try your system under real conditions.
Forward tests are not really saying much.

Hope this helps
GFIs1

Follow me on Twitter Visit my NexusFi Trade Journal Reply With Quote
  #39 (permalink)
 kevinkdog   is a Vendor
 
Posts: 3,647 since Jul 2012
Thanks Given: 1,890
Thanks Received: 7,338


Bollinez View Post
Well, I hope it is okay continuing this quite old discussion, as I haven't found any newer one that fits my question. I am backtesting very actively for two years now and of course noticed real-life differences. My algorithm itself performs the way it has to, but what makes me desperate real-life is primarily the rule on trading hours and days. I know it is not uncommon to exclude volatile hours like noon or to exclude very unprofitable weekdays, but I am wondering a lot whether it is a quite lazy method to exclude time periods that superficial. I tried a volatility-based exclusion, which improved my statistics, but I am asking you...

...whether you exclude time periods in backtests at all and

...did you also ask yourself whether real-life trading would require a different way to handle such periods?

Thanks,
Bollinez

Trade exactly what you test. Excluding days/time periods can be a valid approach. Volatility based exclusions can be valid too. Either can work or not work.

The worst thing you can do - in my opinion - is test an approach, and then trade live differently. In that case, why even bother testing?

Follow me on Twitter Reply With Quote
Thanked by:
  #40 (permalink)
Bollinez
Oxford, UK
 
Posts: 5 since Dec 2015
Thanks Given: 3
Thanks Received: 1


Hello everybody,

sorry for the delay, but I had exams and also some problems with my NinjaTrader. I noticed an unusual change in performance of our algorithm and searched for the error and finally found it. We have some problems with the data feed:

We wrote down some extraordinary candles for multiple years and compared these candles with our NT data files and there was an offset of one hour for 2012 and two hours for 2006 and so on. We finally noticed that our data provider sent us London Time Data. I exported the data and imported it with another time zone but the data update includes some issues concerning GMT/UTC/BST.

I don't want to ask my data provider for solutions, reasons, updates because I would need to search for other errors and I am pretty sure the error occurred due to BST because the time offsets appear after this British Summer Time changes. I just want to understand it and how to fix such errors by my own.

Do you maybe know how to fix it? I saw that NT 7 exports it UTC, but does this export consider BST? Can I import it in order to adjust it for German GMT + 1 or is it not that easy? Because the offset keeps coming back. Do you manage this somehow or do you never noticed this BST problem?

Well, we also decided to start anew and begin with focusing on market trends and not with entries/exits. And I would like to avoid any mistakes concerning time zones and feel these live-differences in my depot :sos:

Thank you very much for your help.

Reply With Quote




Last Updated on May 5, 2016


© 2024 NexusFi™, s.a., All Rights Reserved.
Av Ricardo J. Alfaro, Century Tower, Panama City, Panama, Ph: +507 833-9432 (Panama and Intl), +1 888-312-3001 (USA and Canada)
All information is for educational use only and is not investment advice. There is a substantial risk of loss in trading commodity futures, stocks, options and foreign exchange products. Past performance is not indicative of future results.
About Us - Contact Us - Site Rules, Acceptable Use, and Terms and Conditions - Privacy Policy - Downloads - Top
no new posts