What's to this trading malarky after all ?

[qwertyuiop1]

Ok people - go easy on me here as I am a newbie.

But - Lets assume a trader has backtested their system extensively and it is successful in getting a positive expectancy.

Also - lets assume that that trader is disciplined and risks at most 1%-2% of their equity per trade.

Like - is that it at that stage ?

Is that pretty much job done and it's just a case of sit back and watch the greenbacks roll in?
And if not then what are the other main difficulties ?

 

[Mathemagician]

If the backtesting is done properly and the system performs well in live trading, then you've laid out a solid description of the conditions that must be present in order for a trader to go live for the first time. After that, you must track costs in your markets to fine-tune your estimates, monitor your current models for signs that they are exhibiting behavior outside of the range of expectations, and continue developing new ideas and trading models.

jj

 

[TheBramble]

I'd stress the importance of forward testing.

The fact that I have no confidence whatsoever in backtesting should not sway you. I'd still emphasis forward testing even if I thought there was any merit at all in backtesting.

I'd also suggest an intermediate stage between demo and 1-2%. If at all possible, use fractions of 1% to start with. Real money is very different to paper trading even at a low level.

 

[black bear]

Holy Smoke two great posts on the opening page and no BS

Wow ............................faith restored

Good posts you two

 

[qwertyuiop1]

I'd stress the importance of forward testing.

The fact that I have no confidence whatsoever in backtesting should not sway you. I'd still emphasis forward testing even if I thought there was any merit at all in backtesting.

I'd also suggest an intermediate stage between demo and 1-2%. If at all possible, use fractions of 1% to start with. Real money is very different to paper trading even at a low level.

Forward testing ??
Whats that ?

ANd why do you not trust backtesting?
Assuming the backtesting is done correctly with no curve fitting, then I'm presuming you are in a minority ?
Or is there generally a split camp on backtesting?

Monitoring your results in real live trading?

 

[TheBramble]

Forward testing ??
Whats that ?

Using the trading model you've developed at the right hand edge of the chart. Going forward in time. Tick by tick. It looks a lot different there.

ANd why do you not trust backtesting?
Assuming the backtesting is done correctly with no curve fitting, then I'm presuming you are in a minority ?

You've answered your own question with your assumptions. Backtesting is by its nature an exercise in optimisation and curve fitting.

Me in a minority? I feckin hope so....

 

[chump]

Using the trading model you've developed at the right hand edge of the chart. Going forward in time. Tick by tick. It looks a lot different there.

You've answered your own question with your assumptions. Backtesting is by its nature an exercise in optimisation and curve fitting.

Me in a minority? I feckin hope so....

"in a minor***?"...my god ,it's Gary Glitter.

 

[Mathemagician]

Backtesting is by its nature an exercise in optimisation and curve fitting.

Not when it's performed properly! When done properly, backtesting is a dispassionate evaluation of the risk and return inherent to a particular trading idea. This does produce risk/return estimates (with confidence intervals) which can be extrapolated into the future. The moment one takes an idea and tries to "make it work", one is no longer performing proper backtesting.

Granted, very few approach it this way, but that's not the fault of the concept. It's the fault of the practitioner.

jj

 

[TheBramble]

Not when it's performed properly! When done properly, backtesting is a dispassionate evaluation of the risk and return inherent to a particular trading idea. This does produce risk/return estimates (with confidence intervals) which can be extrapolated into the future. The moment one takes an idea and tries to "make it work", one is no longer performing proper backtesting.

Granted, very few approach it this way, but that's not the fault of the concept. It's the fault of the practitioner.

jj

jj - you said it far better than my faltering attempt in your earlier comment "monitor your current models for signs that they are exhibiting behavior outside of the range of expectations, and continue developing new ideas and trading models.". For me that translates to "it ain't doing what it did before when it did that, I'd better find something new"...

Even when you do 'do it properly' (and you're right, so few even know how to appoach this) it guarantees you nothing. Sure, you might be right as often as you're wrong, but on that basis, you're just as well off (better off in fact) going forward with the most basic of setups and assumptions rather than relying on 'what worked then'.

All just my VHO and always happy to be comprehensively convinced otherwise.

I don't know too many traders who are still in the game who are using anything other than the most basic setups. The complex stuff tends to be just for a moment in time.

 

[Mathemagician]

Who said anything about using complex setups?

You are dead right on that point. Most of the entry/exit models traded by the successful traders I know can be expressed in 2 lines of code, one for a buy and one for a sell. The rest of the code is for things like position sizing, portfolio construction, output generation, and other such fluff. When things get much more complex than that they tend to defy proper analysis by burning through degrees of freedom far too quickly relative to the amount of relevant data available.

You're also dead on about there being no guarantees. If there were guarantees one would not need to define a range of expectations and monitor for that range being exceeded! The whole backtesting process (again, when done properly) reflects an acknowlegement of this inherent uncertainty and an attempt to deal with it in a rigorous and systematic way.

In essence, my approach revolves around attempting to quantify the inherent randomness associated with very basic setups and using this information to build a structure within which I can make the most effective decisions possible regarding when "it ain't doing what it did before". In this sense, though our tactics may differ, I believe our strategy is very similar.

jj

 

[TheBramble]

Well jj, sounds like you use backtesting with a keen knowledge of the limits and constraints inherent within that process. It also sounds like you're doing a better job of it than I ever did! If ever we get the chance to compare notes I'd be obliged.

 

[TheBramble]

"in a minor***?"...my god ,it's Gary Glitter.

D'ya wanna be in my gang?

 

[Mathemagician]

Well jj, sounds like you use backtesting with a keen knowledge of the limits and constraints inherent within that process. It also sounds like you're doing a better job of it than I ever did! If ever we get the chance to compare notes I'd be obliged.

Thank you for the kind words. Sounds like fun. What did you have in mind?

jj

 

[damianoakley]

Ok people - go easy on me here as I am a newbie.

But - Lets assume a trader has backtested their system extensively and it is successful in getting a positive expectancy.

Also - lets assume that that trader is disciplined and risks at most 1%-2% of their equity per trade.

Like - is that it at that stage ?

Is that pretty much job done and it's just a case of sit back and watch the greenbacks roll in?
And if not then what are the other main difficulties ?

Hi,

I would have to agree with The Bramble and say that forward testing is vital, but you should only be forward testing if your backtesting has proved profitable and you think that you can mentally deal with the backtested drawdowns in real life.

The thing with backtesting, is that it is only a simulation - you never really know what your system will do when you take it to market. The very act of you getting involved in a market for real could have a bearing on the results of your system, even though that effect may be marginal.

Plus the fact, when you forward test and real money is suddenly involved, you will have to deal with all the emotions that come your way trade-by-trade.....this in itself can change your results if you cannot control your emotions and make mistakes or start changing your rules mid-trade.....something that is very common.


Thanks

Damian