Silly Questions #2

New comment
 

Just what exactly is the "counter points"? (Meta Trader Strategy Tester)

Can an EA be programmed to trade based on them?

If not, what in the world is it in the tester for?

 

You mean Control points ? look in the help . . .


  • Control points (based on the nearest less timeframe with fractal interpolation of 12 control points )
    The control points modeling method is intended for a crude estimate of experts efficiency that trade within the bar. The history data of the nearest less timeframe must be available to apply this method. In the most cases, the available data of the less timeframe do not completely cover the time range of the timeframe under test. If the data of the less timeframe are missing, the further bar development will be generated on close prices of 12 preceding bars. It means that the changes within the bar are the same as those of price within the last 12 periods. It is fractal interpolation.

    As soon as history data of the less timeframe appear, fractal interpolation will be applied to these new data. But there will be used not 12, but just 6 preceding bars. It means that really existing Open, High, Low, Close prices are reproduced, and two generated prices more. Values and locations of these two generated prices depend on that on 6 preceding bars.

 

so what you are saying, is if I understood all that, I could in fact, write my ea to trade using "control points". ?

I'm not saying I understood all that, but it might behoove me to learn it.

 
mrchuckw:

Just what exactly is the "counter points"? (Meta Trader Strategy Tester)

Can an EA be programmed to trade based on them?

If not, what in the world is it in the tester for?

These are NOT silly questions. These are very, very good qns.

For many of us, It can help to ascertain whether our EA is good to go, in operational sense - able to execute trades, no errors, etc. Like a trade simulator. At the same time, our strategies.

Whether an EA can be programmed to trade on the real market but based on those "Control Points" - now, THAT you have to ask the founders himself.

 

Here's an Idea I had a while back but never implemented. Maybe someone who knows more about tester vs live can critique.

Since most people use 1-minute-data for testing instead of tick-data. Perhaps allot of strategies would work similar to how it worked within the back tester if someone limited the execution of the ea to once_per_minute_bars?

So why does it matter.... Well my thinking is that mt4 creates Simulated Ticks in the path of the Bar. Therefore if you have a Bullish Bar, it'll start at Open>then to Low>then High>then Close and Reverse for a Bearish Bar. But Honestly, I don't even think it creates Simulated ticks. I think it just goes OLHC. If the truth about the bar is that it went Up first (case of Bullish Bar) then bounce down below the open and then Up again; someone could be fooled into thinking their Long Stoploss was hit first instead of the Take_Profit etc.

One thing off the top of my head which wouldn't work with my approach like the back-tester is Broker_Side Sl and Tp. And this is a Big problem because allot of people use Sl and Tp orders. If your system uses Sl or Tp orders instead of having the EA close the order, then it really wouldn't matter if you're waiting for the next M1 bar. The order would be closed at the last Order_Modify Sl or Tp. Good example of something which would be allot different from tester to real is Tick Based Trailing-Stops.

Anyway, since I'm here, I might as well take a shot at the original question. Control Point is a real bad method of testing Imo. All the tests I performed with it in my early days were drastically off compared to Tick or Open Method. Open Method testing is a much better compromise on Speed_vs_Accuracy. Open-Method testing showed Identical results to Tick testing within my earlier days Systems. I was almost convenience this was the way to go. But the reason it worked Identically was because most of my earlier systems was based on Indicator/Oscillator Cross Over upon Bar[1] and Bar[2]. Example: Buy when Ma1 cross Ma2; and Close when the opposite happens. As you start using current Bar or Bar[0] information within your EA, Open-Method testing loses its appeal.

The above example of Ma1 vs Ma2 is what I believe to be a good example of an EA to use Once_per_Minute EA executions upon. Other possible problem maker for this approach would include Scalping, Slippage, Connection Issues, Increased Spreads....blah blah blah... pretty much everything else which limits the back-tester to real-life.

New comment
Reason: