MT4 - Time values in backtesting / detect end of bar

Hi, I've been coding in MT4 for a while now but for a current project I need to detect the end of the bar, or point close to the end of the bar. There is no option to use the open of the new bar unfortunately.

I've tried two bits of code so far, firstly:
Minute() > 58

and a bit of code I found on here:
TimeCurrent() >= Time[0] + (Period() * 60 - 1)

The problem I am having is not with the code but in the backtesting, sometimes the conditions aren't met, as in the backtesting never properly simulates a tick for the final minutes of the bar.. Sometimes it does seem to work fine though.

I've looked through the data and there does seem to be minute data for that time period. I can't work out why it doesn't show the tick.

PS I am aware that when forward testing it if no ticks come through then the event may be missed.

Thanks for your help!

This seems like about the 20th time I have replied to this kind of posting. My answer is always the same. In MT4 it is impossible to detect the last tick in a bar until the new one opens. It actually doesn't matter. What is the difference between performing an action after the so called last tick or before the new tick? None. The action happens in the same state of data values. If you think about it you will realize I am correct.

Your 2 code snippets will never work in a million years. Any more time spent pursuing that approach is a complete waste of time.

Perhaps if I explain this way, you will understand. Ticks are generated to react to market conditions. When the market is slow moving, fewer ticks are generated than when activity is high. The "tick generator" pays no attention to the clock. Divisions of time such as we use make it easier to visualize. If you have ever watched a tick chart, you will get the idea. Because there is no sense of timing, the ticks end up the appropriate time slot they belong in and nobody knows whether the last tick happens 5 seconds, or 5 milliseconds before the end of the bar until the new bar actually opens.

Your notion that some sort of "end of bar" indicator exists has no more validity than a "Holy Grail" EA. In fact it much more likely for the latter to exist than the former. Most people are capable of accepting and correcting their misconception, adjusting their mindset and moving on. I would suggest you do the same or abandon your current project soonest so as not to waste any more time.

Thanks, I understand precisely what a tick is and that its place relative to the candle is almost arbitrary.

The requirement for the end of the bar indicator hovers around the closure of the market on Fridays and the weekend change in price.

I think it's valid to try and find the end of the bar and obviously others do too!

The bolded code shown that you want to know when tick time equals to one second before the new bar open. Please note that the smallest time in MT4 (datetime) is one second. It doesn't recognize fraction of second.

If there was a tick that come exactly at one second before new bar open, then you'll detect with that code. But if the last tick came 2 second or more before the new bar open, your code will miss this tick. Thats why sometimes it works, sometimes not.

you can make experiment with this simple EA code:

int start()
{
if(TimeCurrent()>=Time[0]+Period()*60-1) Print("FOUND");
}

run it on strategy tester M15, check the jurnals tab.
you'll see there are some occasions EA print "FOUND".

hope this help

Thanks sangmane, I re-downloaded/converted the data up from the M1 chart for one of the pairs and now it seems to be working.

I'm not sure why it wasn't working before, maybe there was a chart error I hadn't noticed.

What's the point of writing code that works most of the time and is certain to fail when conditions don't suit the logic?

But I guess that would probably be no different than 90% of the code that is found on the internet.

Well if you want to be that pedantic you could say that the open of a bar isn't really the open either because it may not necessarily occur at the precise time point.

It has just occured to me that it could be possible to catch a tick that is incoming (for example) 10 minutes before the end of the bar, calculate the time remaining in the bar and suspend the EA for a designated time period, say 9 minutes, before refreshing the rates and checking the trade logic. Hopefully with 1 minute remaining

The 10 minutes would be enough time for a tick to come in and trigger the EA while sleep/wait in the EA would work around the dependency that the EA start function can only run when triggered by an incoming tick rather than triggered by time.

Of course this wouldn't work in backtesting.

Won't work in forward testing either.

Care to teach me with your superior knowledge then?

Allow me to put it in the most blunt terms possible.

You're an idiot.

Ticks come whenever they please. Ticks do not come on any set schedule. Ticks come as little as once a minute to as much as 40 a minute. We will never know what the 'last tick' will be. The backtester is quite accurate in simulating tick frequency, and simply because you can't get that last tick doesn't mean it exists.

If you don't understand, I can't explain.

I never said I wanted the final tick of the bar. I made that clear in both my original post and in the code examples I gave.

Ultimately I just wanted to be confident that an order could be triggered in a timeframe in which I consider to be the end of the bar, but it would seem to be a design anomaly of Metatrader that I cannot trigger the EA start() by a time event.

If I could trigger an order based on time it would be the 59th minute of an hour or day bar and I would be happy with that.

Even if you timed the order so it is placed at 23:59:59.999 (end of day), you still have to consider the broker's response, and that usually takes one tick.

I'm not silly enough to leave that small margin for error! Like I have said before 23:59.00 would be fine. There needs to be time for at least one retry if the trade fails. Not scalp trading here either, just H1 and larger timeframes for small-med durations.

I've actually been forward testing for the last 24 hrs on a demo account on EURUSD H1 with 4 EAs using Minute() >= 58, average of two events per bar and it has been quite reliable. I am logging any missed bars but so far the log is empty. But it's still a very small sample so far, I know.

One day you will look at what you have posted here and wondered how could I have been so unreceptive to somebody else's ideas. Its just a matter of time.

You haven't given any ideas to be unreceptive to, unless you're talking about sangmane's post. You've just been patronising, dismissive and rude.

We've been running this code for a couple of weeks now on daily and weekly systems and it's working just fine. Never missed a bar