Testing Strategies

U

ukdaytrader

Well-known member
Messages
273
Likes
5
How do you guys go about testing a strategy. For example I have a few new strategies that I want to test out. Is it a matter of trailing through the historic data, looking at the graphs to see what price you would have bought in / sold out etc.

This is really really painful to test out...its taking days...even more....sometimes I can't be bothered to test it using historic data and just trade small amounts to test it, which is not always the best thing to do....Do any of the brokers you use allow excel downloads of the historic data say 30mins, I also need indicators with this including rate of change and standard deviation (outside and inside market hours) ............ otherwise I'm back onto the f****** Bloomberg again.

This trading takes up quite a massive amount of time to do all the prep work.......It may only take a few seconds to place a trade.
 
you gotta be willing to put the work and effort in order to make the dough...
you know money don't just jumps on you from the markets...

a bloomberg station costs what? 1200$ a month? you can get reuters for less, but I really prefer bloomberg for historic daily data...

the other ways - esignal and tradestation can backtest. or you can pay a 1 time $3K for trading blox and back test whatever strategy you want.

G/L
 
When i have a strategy i want to test i do the following.
1. Look a my faviourite markets and skim through the history to see how it would have performed as a rough idea.
2. Go through a historical backtest, one bar at a time to look for entries and exits and see how it does.
3. Make a backtesting program to get more systematic results
4. Demo for 100 trades
5. Live with 0.1% of account with an increase of 0.10% every week succesful week.
6. Live trading 1% risk forever ^^

I find this process is long and tiring and full of effort. Every day it seems to get no where and time is drifting by... But its all worth it in the end when you can finally name your system.

'The mighty- PowerSword action combo-killer deluxe ultimate robot fighting, massive punch, sex machine, money pulling - awesome EMA slope detecter, sexy stochastic rebound detected, pullback dinosaur crunching super droid'

... I find when i try make a new system its massive process. But you just have to do it :)
 
If you really want to develop a skill and a proper methodology you have to move away from the mechanical concept of 'systems' and 'strategies'. You have to develop the kind of understanding which allows you to act correctly to situations as you encounter them. The market does not lend itself to automated back testing. This is why people who test 'systems' in such a way say idiotic things like "The market has changed" when their system fails to deliver profits the way it did in back testing...as if every trader in the world conspired against them and their system :LOL:
 
new_trader said:
If you really want to develop a skill and a proper methodology you have to move away from the mechanical concept of 'systems' and 'strategies'. You have to develop the kind of understanding which allows you to act correctly to situations as you encounter them. The market does not lend itself to automated back testing. This is why people who test 'systems' in such a way say idiotic things like "The market has changed" when their system fails to deliver profits the way it did in back testing...as if every trader in the world conspired against them and their system :LOL:
Click to expand...

I don't disagree with all in your post and there are some relevant points, but you are stipulating that no 'system' mechanical or otherwise can make a consistent profit. And by definition you must be saying that the only way to make constant profits is through discretionary trading. As usual you are assuming that there is only one way to do things and that is the way you believe in. I beg to differ.



This is not a dig at you, but just an observation of your posts.
 
please advise - i have a list of 5000 equities from global markets and i want EOD data historically for these equities - is there a cheap way of getting this? Thanks
 
andinio said:
please advise - i have a list of 5000 equities from global markets and i want EOD data historically for these equities - is there a cheap way of getting this? Thanks
Click to expand...

You can normally purchase historical data from data providers at a fairly reasonable price, especially if you purchase a data feed from them at the same time. If you're serious about trading and backtesting, then good quality data is vital and will have to be purchased.

Avoid free data at all costs such as from sites like Yahoo. Their prices are not always accurate and do not allow for Stock splits etc so will differ from more reliable sources.
 
Rossini said:
I don't disagree with all in your post and there are some relevant points, but you are stipulating that no 'system' mechanical or otherwise can make a consistent profit. And by definition you must be saying that the only way to make constant profits is through discretionary trading. As usual you are assuming that there is only one way to do things and that is the way you believe in. I beg to differ.



This is not a dig at you, but just an observation of your posts.
Click to expand...


It's the "either or" fallacy. I know at least 5 systematic traders who make consistent profits every year for the last 10 years or more. Purely mechanical. Money management is the key after you have a good system. I know dozens of discretionary traders who keep llosing and keep fighting it because they think they will "learn" the secret. There is no secret. There is nothing to learn by watching screens. Maybe I should say there are billions of things you can learn by watching screens and you never know which ones to use at a given time.

Every time I lost is because I tried to outsmart my system.
 
intradaybill said:
It's the "either or" fallacy. I know at least 5 systematic traders who make consistent profits every year for the last 10 years or more. Purely mechanical. Money management is the key after you have a good system. I know dozens of discretionary traders who keep llosing and keep fighting it because they think they will "learn" the secret. There is no secret. There is nothing to learn by watching screens. Maybe I should say there are billions of things you can learn by watching screens and you never know which ones to use at a given time.

Every time I lost is because I tried to outsmart my system.
Click to expand...

I'm inclined to agree.

IMO once you save a sound trading method based on set-up that you find are consistantly profitable then all that needs to be learned then is how to manage profitable trades to keep them running for as long as possible and how to add to and scale out of your poistion to maximise profitablility.
 
intradaybill said:
It's the "either or" fallacy. I know at least 5 systematic traders who make consistent profits every year for the last 10 years or more. Purely mechanical. Money management is the key after you have a good system. I know dozens of discretionary traders who keep llosing and keep fighting it because they think they will "learn" the secret. There is no secret. There is nothing to learn by watching screens. Maybe I should say there are billions of things you can learn by watching screens and you never know which ones to use at a given time.

Every time I lost is because I tried to outsmart my system.
Click to expand...

What an absolute load of rubbish. You can't comment unless you have actually put in the time and effort to find out for yourself. I only ever hear anecdotal evidence of mechanical system traders making consistent profits from people who then post utter rubbish like this! I only endorse what I put into practice, no anecdotes, just reality.
 
ukdaytrader said:
How do you guys go about testing a strategy. For example I have a few new strategies that I want to test out. Is it a matter of trailing through the historic data, looking at the graphs to see what price you would have bought in / sold out etc.

This is really really painful to test out...its taking days...even more....sometimes I can't be bothered to test it using historic data and just trade small amounts to test it, which is not always the best thing to do....Do any of the brokers you use allow excel downloads of the historic data say 30mins, I also need indicators with this including rate of change and standard deviation (outside and inside market hours) ............ otherwise I'm back onto the f****** Bloomberg again.

This trading takes up quite a massive amount of time to do all the prep work.......It may only take a few seconds to place a trade.
Click to expand...

You have to put in the hours of number crunching, no way around it, I did and when I developed my trading strategy, I back tested again and then forward tested. I have continued to record statistically trading data from the markets I trade and have developed a mental and physical picture of how they behave. I can tweak my strategy accordingly to match current conditions whilst not moving away from the fundamentals of the set up.

I did all this by hours, weeks and months of testing, its the only way you can get the data/inforamtion your wanting, you only get out what you put in imho
 
ukdaytrader said:
How do you guys go about testing a strategy. For example I have a few new strategies that I want to test out. Is it a matter of trailing through the historic data, looking at the graphs to see what price you would have bought in / sold out etc.

This is really really painful to test out...its taking days...even more....sometimes I can't be bothered to test it using historic data and just trade small amounts to test it, which is not always the best thing to do....Do any of the brokers you use allow excel downloads of the historic data say 30mins, I also need indicators with this including rate of change and standard deviation (outside and inside market hours) ............ otherwise I'm back onto the f****** Bloomberg again.

This trading takes up quite a massive amount of time to do all the prep work.......It may only take a few seconds to place a trade.
Click to expand...

mt4 can perform simple back testing; view - strategy tester....
 
Back testing is mostly a waste of time other than to show you what probably wont work. Mechanical systems are designed by curve fitting to past data and only the inexperienced believe that the market will behave this month like it did last month. People think that the further back you test the better but you are simply getting further away from today's conditions.

Isn't it obvious the market is different after the credit crunch compared to before?

I actually sell mechanical systems but stress to my clients that this only provides a framework and that discretion is required. Every couple of months I get somebody who designs an EA for a system and finds that it makes a loss in back testing over a period when we've made a few thousand pips profit. EAs simply don't work.

Many times we close trades early or tighten stops ot even suspend trading altogether. My systems are slightly different now than they were a few months ago.

It's a common illusion that newbies chase (often for years) that there exists a mechanical system where they don't have to use their brains or have any responsibility. Don't you think we'd all know about this system by now and have proof that it works mechanically?

As for the guy who 'knows at least 5 people' making money from mechanical systems, why doesn't he get hold of one of them and do the same? Trading is full of BS...
 
£10kLoser said:
Back testing is mostly a waste of time other than to show you what probably wont work. Mechanical systems are designed by curve fitting to past data and only the inexperienced believe that the market will behave this month like it did last month. People think that the further back you test the better but you are simply getting further away from today's conditions.

Isn't it obvious the market is different after the credit crunch compared to before?

I actually sell mechanical systems but stress to my clients that this only provides a framework and that discretion is required. Every couple of months I get somebody who designs an EA for a system and finds that it makes a loss in back testing over a period when we've made a few thousand pips profit. EAs simply don't work.

Many times we close trades early or tighten stops ot even suspend trading altogether. My systems are slightly different now than they were a few months ago.

It's a common illusion that newbies chase (often for years) that there exists a mechanical system where they don't have to use their brains or have any responsibility. Don't you think we'd all know about this system by now and have proof that it works mechanically?

As for the guy who 'knows at least 5 people' making money from mechanical systems, why doesn't he get hold of one of them and do the same? Trading is full of BS...
Click to expand...

Completely disagree, I've being trading full time as my only source of income now in excess of 10 years, my trading stategy is mechanical, and hours/weeks/months of number crunching gave me a system that is very simple and easy to work with, no system is 100%. Trading is very simple, whichever method you use, its only the trader who makes it difficult, there was one stonkingly good short on the ftse today from 7:50, not brian science to work it out, even the most simpliest of MACD set ups would have got you into that. Wouldn't say all are BS are you say, its easy to spot them though, just my thoughts
 
£10kLoser said:
Back testing is mostly a waste of time other than to show you what probably wont work. Mechanical systems are designed by curve fitting to past data and only the inexperienced believe that the market will behave this month like it did last month. People think that the further back you test the better but you are simply getting further away from today's conditions.

Isn't it obvious the market is different after the credit crunch compared to before?

I actually sell mechanical systems but stress to my clients that this only provides a framework and that discretion is required. Every couple of months I get somebody who designs an EA for a system and finds that it makes a loss in back testing over a period when we've made a few thousand pips profit. EAs simply don't work.

Many times we close trades early or tighten stops ot even suspend trading altogether. My systems are slightly different now than they were a few months ago.

It's a common illusion that newbies chase (often for years) that there exists a mechanical system where they don't have to use their brains or have any responsibility. Don't you think we'd all know about this system by now and have proof that it works mechanically?

As for the guy who 'knows at least 5 people' making money from mechanical systems, why doesn't he get hold of one of them and do the same? Trading is full of BS...
Click to expand...

It depends why you are back testing. Development of trading skills starting with the basics of lower lows, higher highs, s/r, entries, exits, trade management, spotting different market conditions, etc etc. I find this helpful, because I feel better prepared for live trading conditions, when other skills such as correlation come into play.
 
Interestiong debate... I often giggle to myslef when traders are arguing about what works and what does not..

The beauty of this game is that nobody is wrong, and nobody is right... strikes me as funny that people who have many years of experience in either camp cannot understand this ??
 
You must log in or register to reply here.

Similar threads

T
Trading Strategy - Simulated testing
Replies
3
Views
4K
jasper36
J
K
Is backtesting necessary before Walk Forward Analysis?
Replies
2
Views
3K
malaguti
M
P
Strategy Testing
Replies
3
Views
4K
Tuffty
T
A
Back testing with technical and fundamental data
Replies
0
Views
3K
AdrianR
A
M
Trading Strategy Research
Replies
6
Views
3K
MoneyWeaver
M
Top