Can you be profitable without backtesting?

[Spandy]

Hi folks.

I'm trying to get my head around the lengthy process of backtesting strategies. I know that a lot of people find this is a very important part of their trading, and are keen to know that their strategy will be profitable before they engage with it.

I'm keen to learn how people view this.

Back testing takes an enormous amount of resources in terms of time and if someone is going to be managing their trades anyway and using price action, observation of technicals and to justify when to enter and where to exit, is it necessary to define some rule which, in terms of supply and demand is arbitrary, is weighted towards the past and doesn't add to the recognition of what you are seeing?

Aren't defined trading strategies (enter when market trending, at test 50 EMA retrace, with a trailing stop of 3 ATR) simply an effective way of keeping the trader disciplined to principles, rather than a rigid holy grail rule which one can plug into an algo, set and forget and not have to manage?

It feels that it is more important to know why a strategy works or doesn't work, to know what the history R ratio has been.

If you accept the idea that the average retail trader has to manage their trade anyway, that breakouts occur, pullbacks occur, markets trend, and markets range, risk needs to be quantified and managed and targets need to be set per principles,  then what difference does entering lots of data on a spreadsheet of 1 way of decoding the decision pathways above, do to a retail traders bottom line? What does it do if they do this repeatedly? Does it depend on personality? Can you be a profitable trader without it and are you likely to blow up an account if you don't?

[Caseynotes]

5 minutes ago, Spandy said:

Hi folks.

I'm trying to get my head around the lengthy process of backtesting strategies. I know that a lot of people find this is a very important part of their trading, and are keen to know that their strategy will be profitable before they engage with it.

I'm keen to learn how people view this.

Back testing takes an enormous amount of resources in terms of time and if someone is going to be managing their trades anyway and using price action, observation of technicals and to justify when to enter and where to exit, is it necessary to define some rule which, in terms of supply and demand is arbitrary, is weighted towards the past and doesn't add to the recognition of what you are seeing?

Aren't defined trading strategies (enter when market trending, at test 50 EMA retrace, with a trailing stop of 3 ATR) simply an effective way of keeping the trader disciplined to principles, rather than a rigid holy grail rule which one can plug into an algo, set and forget and not have to manage?

It feels that it is more important to know why a strategy works or doesn't work, to know what the history R ratio has been.

If you accept the idea that the average retail trader has to manage their trade anyway, that breakouts occur, pullbacks occur, markets trend, and markets range, risk needs to be quantified and managed and targets need to be set per principles,  then what difference does entering lots of data on a spreadsheet of 1 way of decoding the decision pathways above, do to a retail traders bottom line? What does it do if they do this repeatedly? Does it depend on personality? Can you be a profitable trader without it and are you likely to blow up an account if you don't?

Hi, the average retail trader is destined to lose precisely because they have no repeatable process that has been tested to prove positive expectancy. The cumulative R ratio vs the cumulative win rate plotted on the profitability chart is indisputable as to whether a strategy has a chance of being profitable or not trading live.

Backtesting is not actually such a chore as a simple eyeball backtest of 20 trades is usually enough to rule out most no-chance strategies before moving on to forward testing on demo.

See more in this thread.

 

[Spandy]

23 minutes ago, Caseynotes said:

Hi, the average retail trader is destined to lose precisely because they have no repeatable process that has been tested to prove positive expectancy. The cumulative R ratio vs the cumulative win rate plotted on the profitability chart is indisputable as to whether a strategy has a chance of being profitable or not trading live.

Backtesting is not actually such a chore as a simple eyeball backtest of 20 trades is usually enough to rule out most no-chance strategies before moving on to forward testing on demo.

See more in this thread.

 

Thanks for the reply and the link. A simple eyeball of how a strategy has performed about 20 times or so in similar conditions seems about all I can manage. 

[THT]

I've always back tested over a min of 100 trades - manually on a chart not via a comp, which invariable covers numerous years if done on daily charts

I also make sure I test it through a bull, bear and sideways market too to see if any affect the method - again you'll have to define how you classify those types of market.

Its all about having a positive expectancy which then gives you the confidence to trade it too

 

[Caseynotes]

1 minute ago, THT said:

I've always back tested over a min of 100 trades - manually on a chart not via a comp, which invariable covers numerous years if done on daily charts

I also make sure I test it through a bull, bear and sideways market too to see if any affect the method - again you'll have to define how you classify those types of market.

Its all about having a positive expectancy which then gives you the confidence to trade it too

 

yes, 100 is a good sample size but as the OP says can become very tedious. My system is a little different with 20 trades backtested, then 20 trades forward tested on demo, then 20 trades forward tested on a live account at minimum position size so giving 60 trades in total. 

Forward testing can be a problem when working on higher time frame charts as one trade can take days or longer to set up and complete so many would prefer to continue with the backtesting instead.

[dmedin]

Actually with the Trading View platform you can draw long/short positions on the chart to backtest, so you don't even have to learn programming.  (Although that would obviously be very useful too)

 

[Caseynotes]

Just now, dmedin said:

Actually with the Trading View platform you can draw long/short positions on the chart to backtest, so you don't even have to learn programming.  (Although that would obviously be very useful too)

 

I see that all the time but is that on the free version or only with subscription?

[dmedin]

5 minutes ago, Caseynotes said:

I see that all the time but is that on the free version or only with subscription?

Can do it on the free one

It shows ads and limits you to three indicators though.

[THT]

48 minutes ago, Caseynotes said:

yes, 100 is a good sample size but as the OP says can become very tedious. My system is a little different with 20 trades backtested, then 20 trades forward tested on demo, then 20 trades forward tested on a live account at minimum position size so giving 60 trades in total. 

Forward testing can be a problem when working on higher time frame charts as one trade can take days or longer to set up and complete so many would prefer to continue with the backtesting instead.

 I just like to test anything I'm considering to absolute destruction, to pick outs its flaws or identify the market types it struggles in etc - just my personality though

[Spandy]

1 hour ago, dmedin said:

Actually with the Trading View platform you can draw long/short positions on the chart to backtest, so you don't even have to learn programming.  (Although that would obviously be very useful too)

1 hour ago, THT said:

I've always back tested over a min of 100 trades - manually on a chart not via a comp, which invariable covers numerous years if done on daily charts

I also make sure I test it through a bull, bear and sideways market too to see if any affect the method - again you'll have to define how you classify those types of market.

Its all about having a positive expectancy which then gives you the confidence to trade it too

 

Fair play to you. think the only way I could do 100 would be to run some kind of script, which I don’t yet have the skill to do (but I’m learning, check out  this video...

It would kill me to sit there manually going through and writing all everything that many times. There must be a systematic approach which can automate the data entry. I’m interested also, why you would test a strategy in a bear, bull and sideways market? When you conceive of the strategy, wouldn’t it be in a certain context anyway?

 

[Caseynotes]

7 minutes ago, Spandy said:

 

Free backtesting app here, have not tried it but it looks good (I use a simulator instead 🙂)

https://www.youtube.com/channel/UCgY_eHY4NCTcRnU6CCZXWng

[dmedin]

The Pine code sounds pretty neat.  https://www.tradingview.com/pine-script-docs/en/v4/language/Structure_of_the_script.html

[THT]

3 hours ago, Spandy said:

 

I don't trust computers to do it properly, they'll apply perfect trades, probably ignore gaps and whatnot - I've not back tested anything for about 10 years but when i did I devoted a day or 2 to manually do it - also I'm useless at computer code, coding or anything mildly technical, so doing it manually would have been much quick than me learning to code etc