Opinions on how far to backtest | Forex Factory

Options

Permalinks
Print Thread

Similar Threads

How far back should I backtest? 1 reply

Visual backtest result different from non-visual backtest 0 replies

Infomercials --- any opinions 25 replies

Opinions of USD. 10 replies

My 8 new books for $60 (EBAY), please give opinions. 1 reply

Opinions on how far to backtest

  • #1
  • First Post: Nov 10, 2013 3:57am Nov 10, 2013 3:57am
I have been working on a few new technical strategies that work well within the past 3 years. Before 2010 doesn't give many good signals. Can I get some opinions from other experienced traders on this topic? Thanks!
A picture is worth 1000 words.. A chart is worth 1000 pictures
  • #2
  • Nov 10, 2013 4:08am Nov 10, 2013 4:08am
Quoting adamnsim03
I have been working on a few new technical strategies that work well within the past 3 years. Before 2010 doesn't give many good signals. Can I get some opinions from other experienced traders on this topic? Thanks!
If you are in needed to see if a strategy is profitable, only forward test in real time on a live account is valid ... That's the only way to have the real trading conditions.
MT4/MT5 EAs/Indicators/Alerts coder
  • #3
  • Nov 10, 2013 4:33am Nov 10, 2013 4:33am
Quoting adamnsim03
I have been working on a few new technical strategies that work well within the past 3 years. Before 2010 doesn't give many good signals. Can I get some opinions from other experienced traders on this topic? Thanks!
I think back-testing works in so far as to prove to yourself that your edge works. But if as you said it only works for the last 3 years then you might want to be a bit concerned about its future potential given that it has failed before in the past.

The key thing I think is to understand WHY it works, because then you will naturally have the confidence to take the trade in a live account, and to get out when you are proven wrong.

It's a brutal process for the individual trader. Good luck.
Build good relationships with others.
  • #4
  • Nov 10, 2013 9:42am Nov 10, 2013 9:42am
The market uses a very simple strategy: buy low sell high. I am not convinced the strategies you have there will defeat the market's strategy. So testing doesn't really do any good. To defeat the market's strategy, it will take significantly more cleverness, significantly more courage, significantly more experience, and significantly more work from you.
  • #5
  • Nov 10, 2013 9:49am Nov 10, 2013 9:49am
With Forward testing to, if it works keep using it, until it stops working then stop using it and hope you've made more than it loses when / if it stops.
Nothing to it, but to do it!!! Stick to the plan FOOL!!!!
  • #6
  • Nov 11, 2013 11:47am Nov 11, 2013 11:47am
The markets change, so this might be the reason why you lose your edge >3 years ago.

But to use a strategy, and to keep up with changing markets, you will have to re-optimise/re-adapt it. So, if 2010 gives you bad backtests, try to optise your parameters on 2009, then make a backtest for 2010 again and see if that helped. (1 year optimisation window and 1 year testing window are subject to change and I just used them to make numbers easier)

If this changes something, do a full walk forward analysis.
If not, I would not trade this strategy but I am sure there are other opinions on that

-Darwin
  • #7
  • Nov 11, 2013 6:12pm Nov 11, 2013 6:12pm
Quoting Darwin-FX
But to use a strategy, and to keep up with changing markets, you will have to re-optimise/re-adapt it.
Optimize a strategy based on past datas is worthless in my opinion.
MT4/MT5 EAs/Indicators/Alerts coder
  • #8
  • Nov 11, 2013 6:18pm Nov 11, 2013 6:18pm
Quoting FerruFx
{quote} Optimize a strategy based on past datas is worthless in my opinion.
I'm a re optomise every weekend kinda guy well at times, the market changes, so why not have what's worked best for the previous week, good odds it'll be pretty close for the up coming week, perfect ofcourse not, nothing ever will be.
Nothing to it, but to do it!!! Stick to the plan FOOL!!!!
  • #9
  • Nov 11, 2013 6:36pm Nov 11, 2013 6:36pm
Quoting Turveyd
... perfect ofcourse not, nothing ever will be.
Sure nothing is perfect.

Quoting Turveyd
... so why not have what's worked best for the previous week, good odds it'll be pretty close for the up coming week
Previous results won't guarantee futures results. Even the results of the last candle won't give you what will be the next one ...

The trick is to look at the big picture / context ... based on a period of time fitting our trading style.
MT4/MT5 EAs/Indicators/Alerts coder
  • #10
  • Nov 12, 2013 2:44am Nov 12, 2013 2:44am
Quoting FerruFx
{quote} Optimize a strategy based on past datas is worthless in my opinion.
But you are aware that you ALWAYS optimise your strategies on the past data? does not matter if you trade manual or with EAs. also it does not matter if you use fundamental or technical analysis.
The past is only source of knowledge we have, the only way to learn, the only data to analyse, as the future is always unknown and the presend is ...... now already the past

What makes the difference is HOW you learn from the past and if it is in a way that will hold in the future or just a way that produces overfitting (which is what many ppl misunderstand with "optimisation" and why lot people think it does not work).
  • #11
  • Nov 12, 2013 2:50am Nov 12, 2013 2:50am
Quoting Darwin-FX
{quote} But you are aware that you ALWAYS optimise your strategies on the past data? does not matter if you trade manual or with EAs. also it does not matter if you use fundamental or technical analysis. The past is only source of knowledge we have, the only way to learn, the only data to analyse, as the future is always unknown and the presend is ...... now already the past What makes the difference is HOW you learn from the past and if it is in a way that will hold in the future or just a way that produces overfitting (which is what many ppl...
What is important is if that works for you ... so that's perfectly fine.
MT4/MT5 EAs/Indicators/Alerts coder
  • #12
  • Nov 12, 2013 3:40am Nov 12, 2013 3:40am
Right
But you can't argue the fact that you are also learning from the past, can you? If so, I am curious

-Darwin
  • #13
  • Last Post: Nov 12, 2013 3:54am Nov 12, 2013 3:54am
Quoting Darwin-FX
Right But you can't argue the fact that you are also learning from the past, can you? If so, I am curious -Darwin
Of course!
MT4/MT5 EAs/Indicators/Alerts coder
0 traders viewing now