When using MTF, Multiple Time Frame, support the lowest TF used must be the one on your chart.

All greater TFs used must be multiple of the lowest one.

If you are using 4h + 1h + 5min, then 5 min (or lower, such as 1 minute) is the TF you have to use on your chart. You cannot replace 1h with 3h, as 4 is not a multiple of 3, in such a case you should also replace 4h with 6h, which is a multiple of 3.

samsampop wrote:

Thank you, I had hoped I could run the stop loss on a lower timeframe to be more responsive.  Maybe I will suggest this for the next version of PRT!

… which will never be made as how you imply it should work. What you imply (thus without saying) is that you want the best of both worlds : high resolution (on the response) and a long history at the same time. And if that is not the case then

samsampop wrote:

Strategy is 15 min, stop loss would be 5 or 1 min, so should work but they error. Any help greatly appreciated

then of course this can work by putting all upside down. You put the chart in 1 min, same as the SL, and the remainder (entries etc.) works in 5 or 15 minutes.
And the penalty ? less history (way less). 😉
You can’t have your cake and eat it too …

samsampop wrote:

I just need to get used to the correlation in PRT of the chart TF being the lowest common denominator.

Hi again @samsampop,

Yeah, well, mind your “just”;

So indeed all is related to history and the amount of it, and possibly platforms exist with infinite amounts of free (!) history (tick data) ? In that case it is understandable what you may see lacking in PRT. But the “just” will never count for me and others; envision the 1second timeframe.

So for our fun : my major problem is not being able to go back more than one month. And so the past weeks I am working om extrapolating my 1s TF to 2s and beyond, while all keeps on working in the past (backtest) data. Live will remain working at 1s, but backtesting in 2s and longer should be a kind of representative.
Not an easy task, which foremost comes down to avoiding functions which work at the bar level. For example :

If Close < Close[1] and Close[1] < Close[2] then
  ...
endif

must be re”coded” into something that does the same but is independent of the bar size (like 1s vs 2s).
Already the tests to compare whether the results stay in-tact requires a system in itself !

Maybe interesting to know or see through :
Currently we are in the most lucky situation that until November 2021 we had a clear “Market A”, while after that (on-going) we have a clear “Market B”. These markets are so different, that at least I myself am not really able to trade with it (all is scary, the majority fails). The AutoTrading systems too, don’t work any more. Now, it is very simple to let it work again for Market B, but what if Market A returns (and mind you, in my perception March 2020 (Covid-start) was also still a normal Market A, but maybe A^2. And so … If I today would be able to use a TF of 2s, I would already miss Market A from November. 3s though will allow me to be in there too. And once I can be in both, my strategies will work in both (I will make them do that).

Thus, having sufficient history data is crucial (too all of us). But when trading with the lower time frames, this comes down to sheer sufficient “time period”.
I said “PRT is not going to do that”. But possibly if they realise that this is about time and not about more data points as such, they will allow a year of history with 1s etc. data after all (they will surely preserve the data once it’s in (history data costs $ too from the data provider)). It is only that my backtesting takes 12x longer.

And so we struggle on(wards !) (knowing that the choice for 1s is my own).
🙂