Hi Gubben.
For the version “Vectorial-DAX-M5-_V6_MOD-1” ==> there is a TMF problem. You launch it in 1 minute or 5 minutes?
Can not run in backtest 5 minutes. The results in a minute in backtest seem bad.
Thanks 🙂
Just closed the short losing position, switched to LONG.
You have the same thing Gubben ?
@maymeric What would be the problem? Wouldn’t the equivalent version without “control” have the same backtest results under that period which you are limited to backtest in 1 minute. I haven’t compared…
@BobFlynn yes it’s going long now. Total gain since november 4th is -195 euro, 28% win and 0.1 G/L. Quite terrible so far but heck I’ll give it more time for now. I always try to be prepared to see a typical drawdown first but when it exceeds backtest drawdowns, it’s tougher to swallow.
Good morning, Can anyone who has the premium version put a backtest of version 6.1? I have the normal version and I can only see 16 months of history and I would like to see more, in case the current Drowdown is exceeded, thank you very much.
The discussion below may be useful and thought provoking on … how accurate is the drawdown figure quoted on the performance results screen after a backtest anyway??
Any further comments specific to drawdown generally (as opposed to specific to Vectorial Systems) then please add to the Topic below.
https://www.prorealcode.com/topic/drawdown-useful-or-misleading/
I’’m trading in real version “_Vectorial DAX (M5) _V6”. Now it has a long in gain position even if from 29/10 it doesn’t go very well. But I’m confident.
I followed this strategy from the beginning and I have all clear. The latest published versions are these:
“Vectorial-DAX-M5-_V6_MOD”;
“Vectorial-DAX-M5-_V6_MOD-1”;
“Vectorial-DAX-v2-WFE-MTF”;
“Vectorial-DAX-v2-WFE-MTF-1”;
“ Vectorial-DAX-M5-_v6.2_Agressive_Reinvest”
As soon as the position is closed I would like to put the updated version.
To avoid making mistakes, tell me which one to put. If it is not among that listed, can you post it? Thanks
PS: I have IG premium so if I have to do some back tests I can do it with 200,000 candles.
All test are good results en 100000 and in 200000.
The problem is the sobreoptimization.
And the results un too cases is similar to the first vectorial of Balmora.
We wanted one good optimización of first vectorial, but all vectorials win similar money.
I do not think it is over optimized, if the results in € 200,000 candles are good I think it is a period long enough to prove that it is a robust strategy. Different would be a 3 month or a year backtest, 200,000 candles are more than 2 and a half years old. The strategy in the coming months will earn a little more or a little less, but the numbers must be similar.
I do not think it is over optimized, if the results in € 200,000 candles are good I think it is a period long enough to prove that it is a robust strategy. Different would be a 3 month or a year backtest, 200,000 candles are more than 2 and a half years old. The strategy in the coming months will earn a little more or a little less, but the numbers must be similar.
If the strategy was originally optimised on 100k candles and then works equally well on 200k candles then that is a good sign. However two and a half years of back test data is an absolutely tiny amount of data to put any trust in at all let alone risk real money on in my humble opinion.
two and a half years of back test data is an absolutely tiny amount of data
Don’t forget though that this is 200k bars over 2.5 years as the Vectorials use 5 min TF.
Vonasi, you normally use Daily TF which for 2.5 years would be < 1k bars worth of data.
Don’t forget though that this is 200k bars over 2.5 years as the Vectorials use 5 min TF.
2.5 years is 2.5 years! Just because there is 200,000 candles in it does not make it any more worthwhile. Three years ago Brexit and Trump happened and the markets changed and so anything that has been optimised for post Brexit and post Trump will not work if markets revert to the norm at some point or if they decide to head off in a different style and path. The world changes fast and markets change fast and that means you can lose money fast when your short history based strategy quickly becomes yesterdays news.
Yes I like daily and weekly strategies because if I can create a non curve fitted robust strategy that has worked consistently since 1995 then it has weathered many different market types and styles and I can be pretty certain how it will perform going forward. If I had 5 minute data going back to 1995 then yes I would be trading 5 minute charts but as I don’t it is currently just guess work based on very limited data as to whether a 5 minute strategy can actually work well in various market types and styles.
The converse can also work … optimise for the current scenario (e.g. Brexit and Trump) and then re-optimise when performance starts to drop off indicating that the current scenario is being overtaken / superceded by a new scenario or trading style etc?
re-optimise when performance starts to drop off
Drop off = you lost more money than you are comfortable with = failed strategy.
So you re-optimise and it shows you the settings that you should have been using to make money instead of lose money so you start using those settings just as the market changes again and you start losing money again. You are always playing catch up – and always winning less than you think you should be and always losing more than you think you should be. Personally I don’t like playing catch up and living in hope that things don’t change with market structure for a while so that I can win back what I just lost.
I agree with you … but it is the beauty of life, everyone thinks they are doing the right thing, but then only 10/15% of those who do this work do it.
And anyway, before being a programmer you have to be a trader and transform what is profitable into the trading system.
