How long should I test a strategy for?

Trading Systems
#1

Greetings all, I have a strategy and have back tested it from 2019 to current day on 3 different currency pairs. It had an average win rate of 76% sounds pretty neat right?

Upon using it on a demo account it really hasn’t been as successful as the win rate suggests. I iterated it a bit and have developed it into a swing strategy as I feel more comfortable and it is more accurate over these time frames.

Back to my opening question, how long should I use it in my demo? And if it does fail, how long of failing does everyone think I should take before making a new system?

how-long-test-trading-strategy
how-long-test-trading-strategy1200×628 94.5 KB

#2

You should practice on demo till you get confidence to take risks on a live account,

3 Likes
#3

How long is a piece of string? There is no simple answer - it all depends on your lifestyle, what works for you and what doesn’t.

IMO, the only way to test a strategy is ongoing - backtesting is history and not current market conditions, albeit some chart patterns are replicated.

I would carry on for at least a couple of months testing, experimenting, and tweaking on your demo account until these three fundamentals are clear:

  1. Biden is elected President, without Trump clogging up the process.
  2. Brexit negotiations are resolved
  3. Covid vaccine is being rolled out globally.

Why? Because currently, markets are high octane running on uncertainties - so even the best strategies are subject to daily economic fundamentals taking precedence over technical indicators.

3 Likes
#4

@steve369

While I agree to a point markets are always worrying about something or other.

Right after all this current uncertainty is cleared up here’s a load more to worry about

will the Fed raise or lower rates
will congress okay a new stimulus bill
will deutsche bank implode
will the italians have a referendum to leave the EU
will Tesla valuation lead to a full market meltdown

Im sure there are plenty more

If we are trading we have to live with uncertainty constantly.

4 Likes
#5

Yes, very true. We live in an uncertain world. However, there is a difference between short and longer term economic effect. My three indicators are major uncertainties that have been affecting the markets for some time now and will continue through to mid 2021.

Add on your contributions to uncertainty and we’ll have a roller-coaster ride to oblivion…

2 Likes
#6

Haha I love this. How long is a piece of string!

1 Like
#7

I suggest try to test it in your demo account till you are 100% sure of the strategy, It is simply not worth losing your money in actual market when you could have already identified and avoided your losses. Hence dont be in a hurry to go for live trade, Backtest and papertrade your strategy a lot.

2 Likes
#8

I will certainly keep my trust in the strategy even if it doesn’t perform exactly as well as I want it to, a constant profit is way much better then consistent losing with one big win here and there

#9

@steve369

I can see you focus on the fundamentals on the markets quite a lot, don’t get me wrong it should be a daily habit to make sure you’re not entering into a market about to flip, however I do rely more on the technical side but I will definitely keep this in mind, thanks for the input

#10

Would you say this is more based on performance than on 100% certainty? How would you define “100% sure”?

#11

Less than 2 years… definitelly too short.
I started algo trading in 2015 and I backtested on 4 years.
Now I am backtesting on 6-8 years.
I know traders that are happy only with 10 years.
Also the number of trades matters, it should be at least 300.

2 Likes
#12

Here is my 2 cents worth of advice. Demo trade the strategy until you have a minimum sample of 100 fully executed setups and evaluate the performance. If you could trade more than 1000 setups, that’s even better. The reason why I say fully executed setups is that, you might be using a multiple entries on one trade, and could be confused as multiple trades. So to avoid such, count the setups instead.

2 Likes
#13

Hi, go back to your code and find out why the testing is not matching the demo. There is an error in your coding and testing process, it should match exactly to the demo testing. Use Mt5 for coding and testing. This will give you the confidence to move forward with the system you have developed. Your time frames are too short, test over 5 years, and do a minimum of 500 trades per currency pair. When you trade on the demo you will know with the first 100 trades if your coding and testing process is working properly. 76% win ratio is not high enough for consistency of system return and phycological confidence in the system you have developed.
Regards

1 Like
#14

WHAT??? And here I’ve been trolling around all this time waiting for one of you yardbirds to let slip where I can get that magic pill that makes me take only winning trades… :stuck_out_tongue_closed_eyes:

I’ve kinda been wondering about that: how much did Covid change the baseline? I mean, with all this talk about “the new normal” is backtesting prior to March 2020 still valid?

#15

I trade manually

#16

There is nothing “normal”… :rofl:
Or everything is normal if you prefer.
I want my stragies to work pre covid, covid and post covid, just like pre brexit and post brexit, pre Draghi and post Draghi, pre Trump and post Trump …
A robust strategy should be able to survive and prosper trough many market conditions.

3 Likes
#17

:face_with_raised_eyebrow:

Do you mean backtest was also MANUAL?

#18

Yes indeed

1 Like
#19

Hi,
I suggest you add information like trading and back testing manually when you post a question. Manually back testing is inefficient and inaccurate. Move into the 21st Century and use the power of algorithms to assist your testing & trading. Its like driving a horse and cart on the road while all your competitors are driving cars.
Regards

2 Likes
#20

@Jckbt

So you suggest I take the time to learn how to algo trade which I am already interested in. Thanks for the advice, would you have any links or places where I could start to learn?