molanis.com/forum - Tools for MetaTrader : The place to ask about the best expert advisor builder, expert advisor downloads, and expert advisor programming

The more ticks you use in backtesting the better. It is not only about trading at the open, it is also about tp and sl.
I do not see a way to backtest a scalping ea if your historical info is made of open, high, low and close. You are missing all the ticks in the middle that may trigger the tp or sl. You will get better result in backtesting if your time frame is 4 hours or more.

That's precisely what I was asking the mod's to confirm above. It seems to me that, unless you're trading within the current bar (shift 0) or have some tick based caculations/indi's built into your EA, tick data is irrelevant to your backtesting--and irrelevant to your live trading for that matter.

Of course, for best results you would have to build the EA to trade on open prices. Otherwise, you face the prospect of a market gap where an open price is a good distance away from the close immediately preceding it. Hence the warning, "explicitly control for bar opening."

I was trading offline tick volume charts when I had my tick data set up, but each tick volume bar contained 1597 ticks. This paralleled the H1 TF pretty well minus the "flat" time, but basic indi's like pivots are, by definition, time based so I abandoned tick data collection. I've moved away from backtesting too--largely because many leading indi's are not backtestable.



Euwe... MTF. You will definitely have to do some work to accomplish this. Search for cfx's thread where he or I mentioned Birt's EA Review. I used the data links there, but never the MTF plug-in. Cfx has. I guess you could post there and hope he's subscribed to the thread. The Member List was recently disabled in this Molanis forum, so private messaging is impossible

Hi Guys

Wow,

If one volume of tick bar has 1597 ticks yes it makes sense to create a scalping EA. The tp and the sl sits very comfortably in there and multiple times too, hense I agree scalping EA developers need the 90% modelling quality historical data. They need every tick they can find and open, close, high and low are not their concern. While I sweat it out making a 'Heavy Long Range Single-Fire Sniper Assault Rifle' I do see a point if my buddys toil it out for 'Short Range 300 Rounds Per Minute Light Machine Gun'. We have different targets. But it all adds up.



When I started with Molanis SB MTF EAs are only bactestable on MT5. Halfway through I got the info that the latest version of MT4 is MTF testable. And hence the creation of my MTF EAs.



The emphasis of the The Smaller Cycles being interdependent on The Large Cycle is documented very well in Walter Bressert's book called the The Power of Oscillators/Cycles combined. That book cost me USD$145. This was 21 years ago. (It's still available, used, cheaper).



If your M30 EA say, employs a TA where, MacD Main Shift0 must be higher than 0.00, and you get 100 Total No Trades with 75% Profit Trades, you can improve this by adding another Macd TA exactly the same, BUT H1. This method is for MacD only, have not tried other indicators.



Your Total Trades will REDUCE, your Profit Trades will INCREASE, and your Profit Factor will IMPROVE.



There you go, you have just turned your EA into a valid Multi-Time Frame EA. I call this TA a 'Filter'.



(and ur one step closer to the secret on how I build my EAs as seen on youtube. Look top left screen, see the word Built By Molanis? I can, means you can)



Goole me - Kitz3281

Thanks for clearing that up. So to test any EA that isn't specifically written to enter AND exit using open prices (I suspect there are very few), you need accurate tick data. This means you're faced with:

1. Locating and purchasing quality tick data that is accurate for your specific broker (assuming it's better than the free stuff), and

2. Unless you want keep buying data for future backtests... babysitting a data logging EA, etc. (you better have a king kong server connection).

Keep in mind that 4 and 5 digit ticks are not interchangeable. The minimum 4 digit tick is a pip, while the minimum 5 digit tick is a point. These are the same issues I faced before finally giving up on assembling accurate tick volume bars. I'm staying with forward testing.