How do I know if my tick data is reliable?
I developed an EA that performs very well backtesting the last year, but when I run it on 2018 the result is a huge loss. 2019 also performs badly.
I know that results in a certain year doesn't garantee results in other years. But I would like to be sure if this results are correct.
How do I know if my tick history is reliable? Does the "History quality" indicator on Metatrader 5 shows it correcly?
For me it shows it has a 100% history quality. (Which sounds too good to be true) I'm running on the default EURUSD chart.
Below I attached the earnings curve from each year for better understanding.
Before questioning the quality of Tick data, first consider questioning the strategy and the logic of your EA. Questions like:
- How does the strategy hold up for different conditions?
- Under what conditions does it work best and under what conditions does it work least.
- What were the conditions in 2018 or 2019 that would cause is to do badly?
- What conditions in 2020 caused it to do well?
- Is the logic in my EA working correctly and does it reflect the strategy requirements?
- etc.
Tick data is overestimated in my humble opinion, if you think about it, ticks can be any number the next candle, because nobody knows the future of the markets,
if the ticks comes with 800 or 3000 makes no difference for your EA really,
as long as there are not any data missing and gaps in the data, like if the data is broken or corrupt,
but if it is, Mt4 will tell you about it, in that case just delete it and download it again, the standard data you can download from within the platform, is good enough.
you don't need a monthly subscriptions of expensive historical data, don't let anyone tell you differently, they are just saying that to sell you the data.
the markets always change, so don't count on anything to work forever, also run it on many different pairs instead of curve fitting, that will give you a more honest result.
if a strategy is good it should work on all pairs, if not, its probably curve fitted and only work with good results on past data, I think that's your problem,
you have been optimizing to much, try avoid as the plague but however, sideways markets tends to make most EA's perform as shit .cheers
Tick data is overestimated in my humble opinion, if you think about it, ticks can be any number the next candle, because nobody knows the future of the markets,
if the ticks comes with 800 or 3000 makes no difference for your EA really,
as long as there are not any data missing and gaps in the data, like if the data is broken or corrupt,
but if it is, Mt4 will tell you about it, in that case just delete it and download it again, the standard data you can download from within the platform, is good enough.
you don't need a monthly subscriptions of expensive historical data, don't let anyone tell you differently, they are just saying that to sell you the data.
the markets always change, so don't count on anything to work forever, also run it on many different pairs instead of curve fitting, that will give you a more honest result.
if a strategy is good it should work on all pairs, if not, its probably curve fitted and only work with good results on past data, I think that's your problem,
you have been optimizing to much, try avoid as the plague but however, sideways markets tends to make most EA's perform as shit .cheers
My expectation is that tick data SHOULD matters a lot, because it can make the difference between stopping out or not. A tick can deviate a lot from the median of any bar especially in volatile markets. Any thoughts? Or are stops not implemented like that in MT4/5?
Tick data is overestimated in my humble opinion, if you think about it, ticks can be any number the next candle, because nobody knows the future of the markets,
if the ticks comes with 800 or 3000 makes no difference for your EA really,
as long as there are not any data missing and gaps in the data, like if the data is broken or corrupt,
but if it is, Mt4 will tell you about it, in that case just delete it and download it again, the standard data you can download from within the platform, is good enough.
you don't need a monthly subscriptions of expensive historical data, don't let anyone tell you differently, they are just saying that to sell you the data.
the markets always change, so don't count on anything to work forever, also run it on many different pairs instead of curve fitting, that will give you a more honest result.
if a strategy is good it should work on all pairs, if not, its probably curve fitted and only work with good results on past data, I think that's your problem,
you have been optimizing to much, try avoid as the plague but however, sideways markets tends to make most EA's perform as shit .cheers
Tick data is both reliable and unreliable. The more hard coded numbers you have in your EA, the more unreliable your EA becomes. The more filters you have in your EA, the more likely you have more hard coded numbers in your EA.
It's over-optimization. When you tweak your hard coded numbers, There's chances that some numbers just let your trade narrowly escape from being stopped out and win the trade(you may not be as lucky as this time when a similar setup happens again), more than other numbers. But when you test it with another broker's data or History Center's data, you will find out the result becomes really bad indeed and you may just ignore it until you lose some money in the live market. Tweaking the numbers inside the EA would be a bad practise to polish your EA. So it's better to focus on the strategy and logic other than tweaking the hard coded numbers.
If you use 30 min or higher timeframe, macro condition also needs to be considered, trade war, interest rate policies, QE.
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
You agree to website policy and terms of use
I developed an EA that performs very well backtesting the last year, but when I run it on 2018 the result is a huge loss. 2019 also performs badly.
I know that results in a certain year doesn't garantee results in other years. But I would like to be sure if this results are correct.
How do I know if my tick history is reliable? Does the "History quality" indicator on Metatrader 5 shows it correcly?
For me it shows it has a 100% history quality. (Which sounds too good to be true) I'm running on the default EURUSD chart.
Below I attached the earnings curve from each year for better understanding.