| |
|
Last Poster |
Posts |
Pages |
Last Post |
| |
|
|
|
|
|
|
| Author |
Message |
     stevo
Member
Username: stevo
Post Number: 99
Registered: 01-2003Rating: N/A
Votes: 0
|
| | Sunday, December 28, 2003 - 10:40 am: |  
|
"One of the great advantages of Technical Market Indicators is that they can be tested against actual market history"
Robert Colby, The Encyclopedia of Technical Market Indicators.
Backtesting can be a real minefield for many. So many dead-end paths, and so many issues about testing methods.
However if we don't try our methods on past data to see if they work, what faith would we have in them working going forward into the future? It easy easy to follow general TA wisdom, lose money, and then say that technical analysis doesn't work. This is where backtesting comes in.
A thoroughly tested strategy using objective indicators is the best chance that I have to know that a particular strategy may or may not work.
I am able to build a precise trading strategy that enables me to go forward with confidence. I can compare results from the developed strategy with backtested results over the same period.
I can't imagine that anyone would disagree with the value of thorough and well thought out backtesting. Yet so few seem to put the time, money and effort into backtesting their theories on the market.
regards
Steve
|
shaguar
Member
Username: shaguar
Post Number: 210
Registered: 10-2002Rating: N/A
Votes: 0
|
| | Sunday, December 28, 2003 - 01:14 pm: |
|
Hi Steve,
"I can't imagine that anyone would disagree with the value of thorough and well thought out backtesting. Yet so few seem to put the time, money and effort into backtesting their theories on the market."
I guess why there are so few people actually backtesting their system is because most of people dont know where to start. Unless, you are programmer like me and midge, who can build their own system for trading and backtesting. Otherwise, people would need easy to use backtesting software. They will also need to know their system well (logic and practical usage) and understand those statistical infomations generated by backtesting.
Believe me! After everything is done, and you got your first ever backtest result. You would wonder how much those technical indicators fails your expectations.
So, can you show us the starting point to backtesting?
I used a free software to charting and backtesting, but it need programming skill to make it work. I will post the link here only if colin dont mind. And I think it would be a good idea if there are some backtesting tools addon to increditable chart package. 
(Message edited by Shaguar on December 28, 2003)
Cheers,
Shaguar.
|
midge
Member
Username: midge
Post Number: 222
Registered: 10-2002Rating: N/A
Votes: 0
|
| | Sunday, December 28, 2003 - 02:30 pm: |
|
For backtesting I would recommend TradeSim Enterprise Edition. I used it for literally hundreds of hours.
It improved my understanding of trading immensely.
Cheers
MidGe
|
perler59
Member
Username: perler59
Post Number: 30
Registered: 09-2003Rating: N/A
Votes: 0
|
| | Sunday, December 28, 2003 - 02:32 pm: |
|
Hi stevo and shaguar,
I'd love to see back testing added to incredible charts. All that excellent historical data is sitting there on the server just waiting to be put to good use... I think it would have to run on each client’s computer rather than the server, to avoid CPU overload. One problem with back testing is that you can only mechanically test a system that is mechanical. The other problem is that you soon discover that almost nothing "works"! I'd like to be able to construct a sufficiently complex automated trading system, so that it adjusts to market conditions as they change. I'm a bit of an amateur programmer, so I'd like to have a programming language in which to express my system, (because its far more flexible) rather than pages of forms with tick boxes and numerical fields to fill in... I'm prepared to pay extra for this... Does anyone else have any thoughts on this?
|
stevo
Member
Username: stevo
Post Number: 100
Registered: 01-2003Rating: N/A
Votes: 0
|
| | Sunday, December 28, 2003 - 05:46 pm: |
|
All
I, like Midge, use TradeSim Enterprise edition combined with MetaStock. You can get the versions without Monte Carlo analysis but this feature is very powerful. I have also spent countless hours testing with this software. It improved not just my understanding of trading, but also my profitability! Software like this is a revelation.
There are some other ways to approach the problem. Firstly you could use Roy Larsen's excellent Trade Equity code with Metastock. It is free and it is very powerful. This is where I started.
I have used AmiBroker for backtesting. I prefer TradeSim.
These are just a couple that I know. I have not upgraded to MS 8 so I don't know how good this package is.
"One problem with back testing is that you can only mechanically test a system that is mechanical."
If it can't be expressed mathematically and coded then I suggest that the system is too subjective to be considered worthy of use. There is not much that can't be coded.
I would also suggest that a mechanical trading system doesn't have to be all that complex. Often adding more to a system doesn't really improve results - but there are always exceptions to most rules when it comes to system testing. I like the idea of a trading system that adjusts to market conditions, as well as a stock's character.
There is a lot of satisfaction to be gained from creating and trading your own system. It can be quite difficult to trade someone else's strategy, although it can be done - it would take quite a bit of training. I know - I taught my brother to trade a system.
Steve
|
|
Threads by Last Post Time: First
Previous
483
484
485
486
487
488
489
490
491
492
Next
Last
|
