Market phenomena - page 68

 
alexeymosc:
Where does the 5 pips spread on the yew come from? In the kitchens? )) Where does the slippage come from? maybe the requotes, that's right...

5 pips is a bit too much. Spread can be floating. Spread may widen quite widely at night and on the news. Well, let it be requotes, not slippage. If the goal is necessarily to open a position, it is likely to take several attempts to do so and the price may move significantly during this time (in a fast market). Everything is cool and perfect in the tester. But the real conditions can spoil the system that is beautiful in the tester. That is why it is better to test with inferior conditions. But everybody is free to do it as he thinks fit. )) The higher the TF the less the system is subject to these "trifles". I do not go below one-hour timeframe. I use small TFs to test TS in real time and perform debugging.
 
tol64:

5 points is an extra margin. The spread may be floating. The spread may spread wide enough at night and on the news. Let it be requotes, not slippage. If the goal is necessarily to open a position, it is likely to take several attempts to do so and the price may move significantly during this time (in a fast market). Everything is cool and perfect in the tester. But the real conditions can spoil the system that is beautiful in the tester. That is why it is better to test with inferior conditions. But everybody is free to do it as he thinks fit. )) The higher the TF the less the system is subject to these "trifles". I do not go below one-hour timeframe. I use small TFs to test TS in real time and perform debugging.

I understand everything. It's just the way it turned out: MO = 0.00023. In general I agree, we should simulate under the worst conditions. But it is not a trivial task to create a nice balance graph on the whole history and in the worst conditions.

Ok, let's look at it from the other side. There have to be options.

PS: I am attracted, as in the joke, by the process itself. I'll unload 10 years' worth of pentaminoes and get an array of a million records. It's an open field of imagination ).

 
alexeymosc:

PS: like the anecdote, it's the process that appeals to me. If I unload 10 years' worth of five-minute notes, I'll get an array of a million records. Now that's room for imagination )

It's great when you enjoy the research process. :) Can Excel handle a million entries? Or will you use another program? Test it in the MT tester, then. :)
 
tol64:
It's great to enjoy the research process. :) Can Excel handle a million records? Or will you use another program? Test it in the MT tester, then. :)
Excel can barely hold a million. When I get to a clear idea, I'll test in MT.
 
alexeymosc: Excel can barely pull in a million.
It depends on what stone and on what tasks... But in general, yes, there is no smell of parallelization there.
 
Mathemat:
It depends on which stone and on which tasks...


Alexei, didn't mean to interfere ...
 
tara: Alexei, didn't mean to intrude ...
You say 'A', you say 'B'.
 

2 Mathemat. Subjectively, it is not comfortable to work: drawing formulas, filtering, graph output, etc. takes some time. Optimizing with Solution Finder takes minutes. And if you create dozens of formula columns, it consumes 3GB of memory.

 
Clearly, it's not all that fast. It is best to do the same thing in MT. But there will be problems finding a solution.
 
Mathemat:
It's clear that everything is not so fast. The best way is to do the same in MT. But there will be problems finding a solution.

I am used to the machine (Excel), but am fully aware that MT is a more adequate choice.

And another point, the cornerstone of modelling, is the intra-bar price behaviour. I can't simulate TP and SL in excel, but you can't do it on OHLC prices in MT either (the result won't be correct). I can simulate setting pending order inside a bar, it may be possible to do it in MT too, but I'm not sure.

And about search for solution, GA-optimizer in MT does almost the same (I'm using GA-search in excel 2010, there are other search methods too). All in all, you can do the same thing in MT and even better. But as long as I don't need detailed statistics on price movements within a bar, I will use excel.