I've been experimenting with TICKS using EUR/USD, Stop, Trail, & enterLong over 1 year. For small BarPeriods (like 15), TICKS / no TICKS results seem pretty close. However, as BarPeriod increases the results seem to diverge. TICKS is always more pessimistic (opposite to what the manual says, if this simple test is "normal" ). The no TICKS tests can sometimes change from loss at low BarPeriods to profitable at higher BarPeriods, even though the supposedly more accurate TICKS run remains a (usually substantial) loss.

Is this the expected behavior of TICKS / no TICKS in such a simple strategy, or might there be a problem somewhere?

Yes, this is the normal behavior. TICKS can be sometimes more pessimistic, dependent on the strategy. For small stop distances that are likely to close a trade within the same bar when it was opened, I would expect TICKS to be pessimistic.

Thanks jcl.

But why does the length of BarPeriod matter? This setting still just creates bars. TICKS / No TICKS are still dealing with bars with the same HLOC... Why should the results of a very simple strategy diverge substantially just because the bars were created from a broader dataset?

Thanks.

I do not understand the second part of your question. As to why the bar period matters, I think the longer a bar period is, the more likely it is that the price curve undergoes great fluctuations inside the bar. At least that would by my spontaneous guess.

OK. Maybe greater fluctuations does account for what I'm seeing... Does the following make sense?

Bars have HLOC and fluctuations inside. TICKS uses the fluctuations. No TICKS uses an (unknown to us) approximation/simulation/whatever - presumably simple since there's not much point in making it complex.

My simple test strategy is using only Stop & Trail for exiting.

The manual says No TICKS does stops before trails. So, for any given bar if the stop is within the bar No TICKS does it 1st, presumably at the stop price. TICKS would follow the price curve, and possibly hit the stop at the same price, but possibly trail up 1st. So, for every bar, for stop processing results TICKS >= No TICKS. And as the fluctuations (i.e., BarPeriod) increase, the likelihood of trailing increases, hence greater divergence.

If the stop is not within the given bar, and H does not allow trail, then neither TICKS or No TICKS trails in that bar.

If the H allows for trailing, No TICKS will trail, and sell at the new trailed stop if the C is below it. TICKS will follow the price curve and trail & sell per the curve, possibly at a lower point than No TICKS depending on the fluctuations. So for trail processing, TICKS <= No TICKS. Again, as BarPeriod/fluctuations increases, greater divergence - but in the opposite direction.

If the stop has been set tight, the 1st effect would overwhelm and No TICKS should be more pessimistic, and increasingly so as BarPeriod increases.

If the stop has been set distant, the 1st effect would more or less vanish and TICKS should be more pessimistic, and again increasingly so as BarPeriod increases.

In between, the net positive/negative as well as greater/lesser divergence is a function of the asset's test sample volatility and/or trending.

Is all this correct?

Thanks!

That's correct but for one point. If TICKS is not set, trailing and stops don't happen in the opening bar, only in the subsequent bars. I have here a list with changes planned for the next update and that's one of the points about to change - I suppose it will then make tests without TICKS more realistic.

Thanks jcl.

But maybe I don't understand so well after all...

I ran tests using Stop & Trail of 1, 10, 100, & 1,000 pips. (With a BarPeriod on the high side to see the effects.)

With 1 pip, TICKS was much more pessimistic - opposite of what I theorized above.

With 10 pips, TICKS was more pessimistic - possibly consistent with the theory above.

With 100 pips, the results were almost the same - possibly consistent with the theory above.

With 1,000 pips, the results were identical - because nothing closes until the end of the test.

Would the change you mentioned above account for the opposite result at 1 pip?

I'm a bit suspicious that 100 pips didn't show any divergence...

Thanks.

Hmm, maybe then _I_ did not understand what you explained about your understanding.

Anyway the results you got are just what I would expect, assuming that the trades with 1 and 10 pips are quickly stopped out. I don't know if the mentioned change will give the opposite result, but it can give a more precise result without TICKS in those cases.

Thanks jcl.

I'll play some more while waiting for the new version.

Feels like we're standing in the line, waiting for the new version.