Menu

Options

Bookmark Thread

First Page First Unread Last Page Last Post

Print Thread

Similar Threads

Why your broker is your friend and your enemy 4 replies

EA Key - Any example how to validate a key 3 replies

How to validate a strategy 23 replies

How to validate a DAILY trading strategy? 2 replies

How do you validate your trading idea?

  • Post #1
  • Quote
  • First Post: Jan 8, 2015 6:50am Jan 8, 2015 6:50am
Hello All,

I asked this question within another thread on FF but as it was a bit off topic for that thread replies were very limited so I thought I would create a new thread in order to hopefully get a good sample size of replies.

I would like to know how others validate their own trading ideas. Most of here would be trading real money so hopefully you must have gone through this exercise at some point.

At this stage I'm focusing on those who base their strategies on charts using 1 hour bars or higher or even better daily bars.

Hopefully the feedback can include the methods you employ, whether you use dedicated software for the task (MT4, Excel, etc) and if so what for, and how long you normally allocate to it.

Basically once you come up with a new idea, what testing/validation steps do you take before being confident in trading the system with real money?






Any feedback welcome....

Cheers,

Chorlton
Any new ideas I may have will tend to be rule-based but still subject to discretionary elements and gut feel, and therefore not amenable to automation.

I forward test all new ideas by trading in live market conditions with real money.

Since my ideas are not automation-friendly, I never, ever, back test.

In fact, in any case whatsoever, I will never back test.
I'm not trying to convince anyone. I'm not in the "convincing" business.
Quoting skenobi
Any new ideas I may have will tend to be rule-based but still subject to discretionary elements and gut feel, and therefore not amenable to automation. I forward test all new ideas by trading in live market conditions with real money. Since my ideas are not automation-friendly, I never, ever, back test. In fact, in any case whatsoever, I will never back test.
same here. except that I support trading demo for months.
I'll add my thoughts here. I have always validated my trading ideas with real money, whether my own capital or my banks. I know a lot of traders here validate their ideas/systems with paper trading and/or demo accounts. The only warning I would give is that you also need to validate your emotions as well as your system. Want I mean by that is if you can guarantee that you will trade your demo account and live account without any change to your emotions and/or trading psychology then you may well have a winner.
Quoting Chorlton
Hello All, I asked this question within another thread on FF but as it was a bit off topic for that thread replies were very limited
//-----

hey chorlton..... metastock pro and metatrader are the only 2 programs i currently use..... there have been others in the past like chartsmart, tc2000, esignal and such.....

skenobi brings up a very good point in , Since my ideas are not automation-friendly, I never, ever, back test. ......

i'm the exact opposite..... have mentioned many times in the past , manual trading is just not one of my abilities..... my ideas must be automation-friendly ...... and i always backtest...... and my backtests will be as accurate as any forward test because i take great care in writing the ea's ......

that is the key..... the ea , along with the trading idea, must be backtest friendly...... otherwise, the almost unanimous opinion that backtests are useless is true.....

bottom line, i always backtest to validate.....

as far as how much time is allocated, that would be probably less than a day but rarely up to a week..... there will be no demo ..... it will go straight in a live account.....

testing in a live mini account was my preference, but with ibfx's demise, a standard account fxcm account must be used..... that can be expensive as my december results show..... but to second jacob's point above, the idea must emotionally tested with real money as well.....

still manage to have my share of losses though...... go figure.......h
to trade and code, keep both simple... no call to impress....h
Quoting skenobi
Any new ideas I may have will tend to be rule-based but still subject to discretionary elements and gut feel, and therefore not amenable to automation.I forward test all new ideas by trading in live market conditions with real money.Since my ideas are not automation-friendly, I never, ever, back test.In fact, in any case whatsoever, I will never back test.
Quoting JacobfordFX
I'll add my thoughts here. I have always validated my trading ideas with real money, whether my own capital or my banks. I know a lot of traders here validate their ideas/systems with paper trading and/or demo accounts. The only warning I would give is that you also need to validate your emotions as well as your system. Want I mean by that is if you can guarantee that you will trade your demo account and live account without any change to your emotions and/or trading psychology then you may well have a winner
Thanks for the comments so far....
Quoting hayseed
skenobi brings up a very good point in , Since my ideas are not automation-friendly, I never, ever, back test. ...... i'm the exact opposite..... have mentioned many times in the past , manual trading is just not one of my abilities..... my ideas must be automation-friendly ...... and i always backtest...... and my backtests will be as accurate as any forward test because i take great care in writing the ea's ...... that is the key..... the ea , along with the trading idea, must be backtest friendly...... otherwise, the almost unanimous opinion...
Hi Hayseed,

Thanks for the detailed response.

Quoting hayseed
...... and my backtests will be as accurate as any forward test because i take great care in writing the ea's
Would you be willing to expand on this? What kind of steps do you take when writing the EA to ensure this?

Also, as you rely on backtesting what software do you use for this task? Strategy Tester in MT4?? I ask, as I've read many times on FF that strategy tester is not very accurate and should only really be used for checking code and ensuring the buys/sells occur when they are meant to as opposed to providing any actual accurate statistics on the profitability of a particular strategy.

As I'm also keen on backtesting my ideas before trading real money, I'd be really interested in any feedback.

Cheers,

Chorlton
For any strategy done on daily charts, I manually backtest on paper from MT4.
For any strategy using <Daily, I have a freelancer code an EA for me, then I backtest with strategy tester.

After I'm convinced the strategy is solid, I take it straight to a live account. Demo results can be misleading. I only trade on demo accounts when I'm testing an EA for bugs. To me, using a demo account is like trying to learn poker by playing with fake money. There's no tangible stake, so I'm less inclined to iron out fine details and improve performance.
Quoting Chorlton
How do you validate your trading idea? Cheers, Chorlton
Trade it in a very small LIVE account. I can assure you though, you don't and will not make the same trading decisions or think the same way, as you would, trading a demo account. So it's fine and dandy saying it works on demo, but trading is more psychological than anything else and trading live to demo differs greatly.

Take what I said for what you willl, it's from my own experience and nothing more.

G.
Quoting Chorlton
Also, as you rely on backtesting what software do you use for this task?.
//-----

hey chorlton..... metatrader and metastock pro......

really not sure where to start on the rest..... other than to say,

to fully understanding code and the strategy tester takes some time..... it helps if you run the code heavily yourself..... then as the unforeseen problems arise you look close to see why...... one by one you solve the problems and prevent them from happening again.... you slowly learn how to avoid these next time......

if it looks good on the initial backtest and you feel comfortable, run it live..... then occasionally backtest across the same live period and see if the trades and prices match up......

if they don't find out why...... be sure you prepare for latency, requotes, off quotes, missing data and such....... usually those things are not positives......h



//-----
//-----
Our technical support team has identified the solution to restore your missing chart data, but they will have to wait until after trading closes this Friday at 5pm to implement this fix

Frequent "off quotes" or "invalid prices" errors could be due to latency

in my screen all the data between 15/12/2014 - 01/01/2015 seems to compressed into one candle.
to trade and code, keep both simple... no call to impress....h
Thanks for all comments so far...

One of the main points being suggested is running the strategy on a live account. As this would be in the testing phase, keeping lot sizes and account size as low as possible would be paramount.

Consequently, what lot/account sizes do you use whilst in the testing/validation phase? Can anyone recommend a good Broker who offers small account, lot sizes etc?
Quoting Chorlton
As I'm also keen on backtesting my ideas before trading real money, I'd be really interested in any feedback.
//-----

hey chorlton..... let me impose once more......

as for proof that backtests can be accurate you need look no further than webjeff , someone i've mentioned many times before...... he was once up over 300%..... he has ran the exact same ea for close to 4 years now without, to my knowledge, ever changing a single thing...... until a few months ago at least......

i don't know him , never communicated with him and have never seen his ea...... but by looking at his trade list it only took a few minutes to see what his ea was and a few minutes to code a replica ...... my backtest results of that replica matched his trades exactly +/- a fraction of a pip..... in reality live trades exits will sway a fraction of a pip where the strategy tester exits are exact.....

his ea is a perfect example of one whose backtest will be almost identical to a forward test...... until a few months ago at least......

we both used ibfx and were transferred to fxcm...... then his ea began having problems......

Quoting webjeff
So.... My EA keeps showing "Invalid Price" and fails to trade. This is not making me happy. This never happened so frequently at IBFX.
the strategy tester, backtest, would still be placing the trades correctly so at this point the live results, missing trades, and backtest would not match.... the ea needed to be edited to handle "Invalid Price" .....

my ea's were gong thru the same thing and were edited..... jeff chose a different route.....

assuming they should, if my backtest results do not match my live results something is fixin to get corrected.......h
to trade and code, keep both simple... no call to impress....h
Quoting Chorlton
Hello All, I asked this question within another thread on FF but as it was a bit off topic for that thread replies were very limited so I thought I would create a new thread in order to hopefully get a good sample size of replies. I would like to know how others validate their own trading ideas. Most of here would be trading real money so hopefully you must have gone through this exercise at some point. At this stage I'm focusing on those who base their strategies on charts using 1 hour bars or higher or even better daily bars. Hopefully the feedback...
I look at the charts and apply the idea as i have defined it, candle by candle from left to right.
My signature is: "Classified".
do your homework, and look before you leap .,essential practice that keeps you ahead.
Here are some ideas that can help you out based on my experience.

1) be as specific as possible to what your entry specifications are.

2) be as specific as possible as to your trader steps are ( what you do and look at for confirmation of your trade entry)

3) write down what must happen to invalidate a trade in progress.

4) write down what your expected trade should look when happening in real time

5) after all those are done, manually back test for 1 year.

6 ) after the year take note of how effective your stop loss and take profit are statistically, then fine tune.

7) retest using different data.

8) test live or demo using fresh data.
Edit: the reason for specificity is so you don't have to be subjective during crunchtime, but objective during the entire trade. Objectivity gives solid data results, the more unknown variables you add the more uncertain your results are for analysis
Hi All,

Following on from my previous question, if I have a strategy which I would like to demo test on a live account what is the best approach for doing this?

I'd like to test it over a few months (say 3-6mths) and the system is designed to look for opportunities 24/7. As I'm not keen to have my PC turned on for the entire testing duration is my only other option to host it on a VPS? If so, what would be the process to do this, costs involved etc??

Also I've seen many on here refer to Trade Explorer but have never looked into it. Is this something I should also be including?

As a newbie to this any help and advice is most welcome.

Chorlton
Hi,

Just thought I'd bump this up in case anyone who may be able to offer advice missed it.

Thanks.
Quoting skenobi
Any new ideas I may have will tend to be rule-based but still subject to discretionary elements and gut feel, and therefore not amenable to automation. I forward test all new ideas by trading in live market conditions with real money. Since my ideas are not automation-friendly, I never, ever, back test. In fact, in any case whatsoever, I will never back test.
skenobi, any specific reason as to why you would " Never Back Test" ? Do you consider back testing to be a waste ?

If a specific trading system is completely subjective and cannot be automated, then that is a different issue, but if a system can be coded 100 % in an objective manner, even then, would you not consider any back testing ?

Thanks
For anyone who wish to succeed in this business, your trading plan/strategy/methodology must have the following criteria to be able to do the followings:-

1. Identify both Intraday and Intermediate Trend with specific and definitive price levels.
2. Identify both Topping and Basing movements and price actions.
3. Having identified either topping/basing signal should also followed by whether it is just a pullback, retracement or an Intraday Trend reversal. All of these scenarios have its own specific Breached and Hold price levels which are dynamical in nature.
4. Identify between a Trending or a Ranging market price conditions and also a specific set of method on how to trade it.

Price targets are where an Topping/Basing is identified and generated. Nobody can foretell where price is finally going, therefore it is essential and imperative that your trading plan have all the above criteria for one to be able to follow the path of the price movement wherever it takes us in order to succeed in this very challenging business.

No validation is necessary if all the conditions are met.

Hope it answered your query.
Thanks for the reply Genghistar. Good points made.

I'm actually looking for advice as to how others go about "confirming" the reliability of their strategy via a demo account. ie. do they host it on a vps, use trade explorer, etc, etc and if so, how do they do this?

Although I'm familiar with designing equity trading models, how to actually test a forex one is new to me given the limitations of the testing software available. Consequently, others have suggested that to overcome this actually trading on a demo account or a live one with a very small lot size is the best way to go and for this reason, I'm looking for further guidance.
0 traders viewing now