Opinion about backtesting.

[OntradingX]

Hi friends, I opened this thread to get your opinions about backtest results vs real market.

Developing EA's I could see that many times there are very good results from backtests, but when the EA is put into real account we really know if it works or not.
Based on this I started looking to develop strategies where the entries really happen (equal) in the backtest and in the real account.
For this I always run a backtest of the day before to find out if the result that happened in the real account happens similar in the backtest.
When I see that there is similarity (entries and exits, with close prices) I understand that I can trust the results of that backtest.
But here's the question, should I believe in a Backtest done in the format (OHLC or EVERY TICK), I've heard from many developers, one DEPENDS.
Many say the following:
If the Target is too short, always use (EVERY TICK)
If Target is too long, always use(OHLC)
Today I always try to use both modes, but I prefer the results (EVERY TICK), because I realize that in fact it is the closest to the real market result regardless of whether the targets are long or short.

What is your opinion on this?

[jstap]

@ontradingx my opinion on this is if I want to see what and how the EA is working I use OHLC, once I see it working as wanted I switch to ticks.

[OntradingX]

@jstap Nice observation, I've also been doing this, the OHLC being faster makes it possible to verify this.

[TipsyWisdom]

I would never ever use Every Tick. Only 1M OHLC until I can visually see the EA is coded to how its supposed to trade visually, and then changed to real ticks.

I guess you must you all ticks if you are using a custom symbol though...

Its all about candle control.

[Uriel 0]

@ontradingx well, there are Many, many and a lot of variables here...

Starting with MT4/5 history data. Metaquotes provide no trustworthy data (causing so many results that are not reliables...). After that, on Strategy tester, swaps, slippage, etc... Are not considered. To you can check if EA is robust, make the backtest om the 3 options of ticks ( Open prices, tick and control points), the stats must be similars, the graph curve too..

And more:)

[Uriel 0]

@tipsywisdom sometimes the EA got sobreoptimized... You can know it by its stats and perfect curve, to avoid that there are some filters

[OntradingX]

@uriel-0 You are right.
I am following an EA of a Brazilian that he operates in the Brazilian market and is performing excellently, I had access to perform backtests and I realized that the results in (each tick and each real tick) are identical, and the EA actually replicates the same entries and exits from the real market in the backtest, but for the programmer to be able to make this almost miracle happen, he created his own historical database.
I will try to follow this same line of reasoning with my EA's.

[jerkha]

As @TipsyWisdom said I'd rather use Real ticks than Every ticks.
Chose a broker that provide a good history (shouldn't be below 99%) and backtest on a real account rather than on a demo one.
I first use real ticks to make sure that visually it does what I want then I optimize and backtest with real ticks as well.
My fear with 1m OHLC is the spread hunting that could be "unseen" by the backtest.