Libraries: BestInterval - page 8

New comment
 
Maxim Dmitrievsky:

Let's add an OOC

This is a new run I did, the old one got lost. Test on opening prices.

2018.10.20 03:41:42.414 2018.10.18 23:59:59   SUMMARY: 00:00:00 - 23:59:59 : Profit = 591.80 (100.00%), Total = 130 (93.85%), PF = 28.15, Mean = 4.55
2018.10.20 03:41:42.414 2018.10.18 23:59:59   
2018.10.20 03:41:42.414 2018.10.18 23:59:59   final balance - InitBalance (10000.00) + Profit (593.97) with BestInterval.

Is the highlighted part OOS? I'm worried that the bibla isn't working properly.

It's a good thing you posted the third chart. We need to bring it down to earth.

 
fxsaber:

Is the highlighted part OOS? I'm worried that the bibla isn't working properly.

Good thing the third chart was posted. It needs to be brought down to earth.

No, it's not OOS.

 
Maxim Dmitrievsky:

No, it's without O.C.D.

Then I need to figure out why there's a discrepancy.

 
fxsaber:

But it would be self-defeating. The picture will be prettier, but it won't make much sense.

Yes, I understand that, I added my post above.

Here I understand how it all works (or doesn't work at all - I'll just get to know your example):

- using your example we choose the best time for the TS by prohibiting trading outside the BestInterval

- Having obtained the BestInterval on the history it will still not be clear whether the TS has seen any regularities or has been adjusted and optimised.

That's what I propose to study the behaviour of the TS outside the BestInterval interval by reversing trades ... what will be seen there ... I don't know, it's necessary to study

 
fxsaber:

Then I need to look into why there is a discrepancy.

Here's another run in the tester, with true - same time interval.

2018.10.20 04:06:13.375 2018.10.19 23:59:59   BestInterval Action(true - single pass & MT4-style is required) = true
2018.10.20 04:06:13.375 2018.10.19 23:59:59   Calculation time activated intervals is 2018.10.20 04:05:22 - Fuzzy_logic_for_fuzzy_algotraders (common folder) 00:00:51 ago.
2018.10.20 04:06:13.375 2018.10.19 23:59:59   
2018.10.20 04:06:13.375 2018.10.19 23:59:59   Amount of Delete Intervals = 4
2018.10.20 04:06:13.375 2018.10.19 23:59:59   00:30:01 - 03:44:59 : Profit = 186.10 (18.45%), Total = 55 (94.55%), PF = 187.10, Mean = 3.38, DD = 14.40, RF = 12.92
2018.10.20 04:06:13.375 2018.10.19 23:59:59   03:45:01 - 04:44:59 : Profit = 144.30 (14.31%), Total = 24 (100.00%), PF = Max, Mean = 6.01, DD = 3.80, RF = 37.97
2018.10.20 04:06:13.375 2018.10.19 23:59:59   16:45:01 - 19:14:59 : Profit = 421.78 (41.82%), Total = 78 (67.95%), PF = 14.65, Mean = 5.41, DD = 65.48, RF = 6.44
2018.10.20 04:06:13.375 2018.10.19 23:59:59   19:30:01 - 21:29:59 : Profit = 256.40 (25.42%), Total = 45 (100.00%), PF = Max, Mean = 5.70, DD = 33.10, RF = 7.75
2018.10.20 04:06:13.375 2018.10.19 23:59:59   SUMMARY: 00:00:00 - 23:59:59 : Profit = 1008.58 (100.00%), Total = 202 (86.14%), PF = 32.62, Mean = 4.99
2018.10.20 04:06:13.375 2018.10.19 23:59:59   
2018.10.20 04:06:13.375 2018.10.19 23:59:59   final balance - InitBalance (10000.00) + Profit (1051.91) with BestInterval.
2018.10.20 04:06:13.375 2018.10.19 23:59:59   OnTester - Virtual InitBalance (10000.00) + Profit (207.94) without BestInterval. Profit is calculated with TickValue=1 and w/o Commission+Swap.
2018.10.20 04:06:13.375 final balance 11051.91 USD
2018.10.20 04:06:13.375 OnTester result 10207.94

period - 7 months, 15 minutes, opening prices

I ran it on a shorter time interval before - the numbers matched.

 
Igor Makanu:

I propose to study the behaviour of the TS outside the BestInterval interval by reversing the trades... what will be seen there ... I don't know, you need to study it.

Rollover is available in Virtual-.

So far, I can't imagine the practical use of flipping trades outside the BestInterval. Correct reversal in the false mode will not work. It is possible only in the true mode. But it is intended for a single pass. It is of little use. In general, try what you have. I think, disappointment will come quickly, because almost all beautiful pictures will fall apart on OOS.

Even though everything connects half-way, it's still good that it's not a one-click operation. Otherwise a lot of pictures would be here already. And so also MT5 scares me away. But biblical works fully in MT4. But for that you need to read it in the description. So the tool turned out to be self-filtering for users.

 
Maxim Dmitrievsky:

Here's another, another run in the tester, with true - same time interval.

period - 7 months, 15 minutes, opening prices

Before that I ran it on a shorter time interval - the figures coincided.

Now I see. Test by bars will not match if there are SL/TP or pending orders. Only by ticks can coincide now. I use this mode.

 
fxsaber:

Now I see. A bar test will not match if there are SL/TP or pending orders. Only on ticks it is possible to match now. This is the only mode I use.

The orders are only market orders, without stops and takeouts... in principle it is not critical that 2 trades were missed, but still I wonder why. I'll run it again tomorrow, as it's already late, and I'll report back what I notice

 
Maxim Dmitrievsky:

Orders only market orders, without stops and takeouts. in principle it is not critical that 2 trades were missed, but still I wonder why. I'll run it again tomorrow, since it's late, and I'll report what I notice.

Yes, it would be good to figure it out. Although I don't test by bars.

 
Igor Makanu:

In a week, half the market will be in new products.

Yes, almost any TC has a beautiful picture in one or two clicks. The more beautiful it is, the more it compromises itself. So it's good for the bottom line.

But in reality, almost no one will do anything.

123456789101112131415...29
New comment