Download MetaTrader 5

How to make someone believe the backtest ?

To add comments, please log in or register
Sabil Yudifera Daeng Pattah
17372
Sabil Yudifera Daeng Pattah  
I am a developer. Testing robots on live is very long and market conditions are always changing every month and year.
How i solved this problem ?
Alain Verleyen
Moderator
33334
Alain Verleyen  
Are you asking how to cheat with backtests ?
Sabil Yudifera Daeng Pattah
17372
Sabil Yudifera Daeng Pattah  
Why do you think so, it's about statistics.
I do not cheat.
Alain Verleyen
Moderator
33334
Alain Verleyen  
Sabil Yudifera Daeng Pattah:
Why do you think so, it's about statistics.
I do not cheat.

Ok. The title of your topic is weird "believe" the backtest.

A back test will never be the same as a living trading account results. What are you asking for ?

Sabil Yudifera Daeng Pattah
17372
Sabil Yudifera Daeng Pattah  
Testing robots on live is very long and market conditions are always changing every month and year.

Icham Aidibe
10315
Icham Aidibe  

Hi ! 

Reals ticks reports aren't a guarantee. For old datas, the every tick mode is used. 

In 2015 : 

2017.07.29 16:07:45.185 EURUSD : real ticks begin from 2013.05.22 00:00:00
2017.07.29 16:07:45.185 EURUSD : 2015.01.01 00:00 - 2015.02.01 00:00  real ticks absent for 53 minutes of 30198 total minute bars, every tick generation used
2017.07.29 16:07:45.185 EURUSD : 2015.01.01 00:00 - 2015.02.01 00:00  real ticks discarded for 15746 minutes of 30198 total minute bars, every tick generation used
2017.07.29 16:07:45.185 EURUSD : 2015.01.01 00:00 - 2015.02.01 00:00  tick prices of 42117 ticks not matched for 15746 minute bars

In 2016 : 

2017.07.29 16:09:28.847 EURUSD : 2016.01.26 23:59 - 8604 tick prices mismatch for 1417 minute bars
2017.07.29 16:09:28.951 EURUSD : 2016.01.27 23:59 - real ticks discarded for 1432 minutes out of 1439 total minute bars within a day
2017.07.29 16:09:28.951 EURUSD : 2016.01.27 23:59 - 9326 tick prices mismatch for 1432 minute bars
2017.07.29 16:09:29.078 EURUSD : 2016.01.28 23:59 - real ticks discarded for 1425 minutes out of 1439 total minute bars within a day
2017.07.29 16:09:29.078 EURUSD : 2016.01.28 23:59 - 8903 tick prices mismatch for 1425 minute bars
2017.07.29 16:09:29.101 EURUSD : 2016.01.29 23:59 - real ticks discarded for 1364 minutes out of 1378 total minute bars within a day
2017.07.29 16:09:29.101 EURUSD : 2016.01.29 23:59 - 8769 tick prices mismatch for 1364 minute bars
2017.07.29 16:09:29.147 EURUSD : 2016.01.31 23:59 - real ticks discarded for 54 minutes out of 60 total minute bars within a day
2017.07.29 16:09:29.147 EURUSD : 2016.01.31 23:59 - 542 tick prices mismatch for 54 minute bars
2017.07.29 16:09:29.147 EURUSD : real ticks begin from 2013.05.22 00:00:00
2017.07.29 16:09:29.147 EURUSD : 2016.01.01 00:00 - 2016.02.01 00:00  real ticks absent for 7 minutes of 28820 total minute bars, every tick generation used
2017.07.29 16:09:29.147 EURUSD : 2016.01.01 00:00 - 2016.02.01 00:00  real ticks discarded for 28467 minutes of 28820 total minute bars, every tick generation used
2017.07.29 16:09:29.147 EURUSD : 2016.01.01 00:00 - 2016.02.01 00:00  tick prices of 173301 ticks not matched for 28467 minute bars

Half 2016, still every tick mode :

2017.07.29 16:14:19.978 EURUSD : 2016.11.01 23:59 - real ticks absent for 1 minutes out of 1440 total minute bars within a day
2017.07.29 16:14:19.978 EURUSD : 2016.11.01 23:59 - real ticks discarded for 5 minutes out of 1440 total minute bars within a day
2017.07.29 16:14:19.978 EURUSD : 2016.11.01 23:59 - 14 tick prices mismatch for 5 minute bars
2017.07.29 16:14:21.936 EURUSD : real ticks begin from 2013.05.22 00:00:00
2017.07.29 16:14:21.936 EURUSD : 2016.07.01 00:00 - 2017.02.01 00:00  real ticks absent for 1 minutes of 218165 total minute bars, every tick generation used
2017.07.29 16:14:21.936 EURUSD : 2016.07.01 00:00 - 2017.02.01 00:00  real ticks discarded for 5 minutes of 218165 total minute bars, every tick generation used
2017.07.29 16:14:21.936 EURUSD : 2016.07.01 00:00 - 2017.02.01 00:00  tick prices of 14 ticks not matched for 5 minute bars

November 2016 : 

2017.07.29 16:18:12.156 EURUSD : 2016.11.01 23:59 - real ticks absent for 1 minutes out of 1440 total minute bars within a day
2017.07.29 16:18:12.156 EURUSD : 2016.11.01 23:59 - real ticks discarded for 5 minutes out of 1440 total minute bars within a day
2017.07.29 16:18:12.156 EURUSD : 2016.11.01 23:59 - 14 tick prices mismatch for 5 minute bars
2017.07.29 16:18:13.392 EURUSD : real ticks begin from 2013.05.22 00:00:00
2017.07.29 16:18:13.392 EURUSD : 2016.11.01 00:00 - 2017.02.01 00:00  real ticks absent for 1 minutes of 93033 total minute bars, every tick generation used
2017.07.29 16:18:13.392 EURUSD : 2016.11.01 00:00 - 2017.02.01 00:00  real ticks discarded for 5 minutes of 93033 total minute bars, every tick generation used
2017.07.29 16:18:13.392 EURUSD : 2016.11.01 00:00 - 2017.02.01 00:00  tick prices of 14 ticks not matched for 5 minute bars

December 2016 is clean. 

2017.07.29 16:19:30.168 EURUSD : real ticks begin from 2013.05.22 00:00:00
2017.07.29 16:19:30.698 2016.12.01 00:00:20   market buy 0.01 EURUSD sl: 1.05886 (1.05954 / 1.05959)
2017.07.29 16:19:30.698 2016.12.01 00:00:20   deal #2 buy 0.01 EURUSD at 1.05959 done (based on order #2)

Reality is that there's only few months (less than 1 year in real ticks). We all know that 8 months (December 2016 to July 2017) isn't enough to pretend it will forward in live trading.

The solution would be to import the ticks using the custom symbol feature, these datas files are heavy enough to not be kept by brokers more than 1 year, TickStory (lite) doesn't do it neither. 

What would you rely to ? Sky color ? 

Ashish Kulkarni
510
Ashish Kulkarni  

Best way to make people believe in your backtest - open a live signal and have an affordable initial price / rent

Stanislav Korotky
21549
Stanislav Korotky  
Sabil Yudifera Daeng Pattah:
I am a developer. Testing robots on live is very long and market conditions are always changing every month and year.
How i solved this problem ?

Use walk-forward optimization and testing.

Terry Barnes
7
Terry Barnes  
Sabil Yudifera Daeng Pattah:
I am a developer. Testing robots on live is very long and market conditions are always changing every month and year.
How i solved this problem ?
You can solve by building a time traveling robot. Travel to the future, back test, and return back in time. 

My guess is that your robot skeptics understand that building a system (robot) that can optimize a strategy based on known past market conditions does not mean the strategy will work in the future.  It is easier to reverse engineer a solution to a problem where the answer is known. How about when the answer is a mystery?

Paper trade or actually trade to prove your robot holds up under live trading.   However, the past inefficiency your robot found in the market likely no longer exists or will not exist for long - Arbitriage!
Sabil Yudifera Daeng Pattah
17372
Sabil Yudifera Daeng Pattah  

Ok thanks I just want to be given many opinions.

Mohammad Soubra
30906
Mohammad Soubra  
Sabil Yudifera Daeng Pattah:

Ok thanks I just want to be given many opinions.

 You can also try real time your expert for (let's say) a week then back test it with same parameters in same period
To add comments, please log in or register