In case anyone hasn’t already made up their mind on the system presented on the first post of this thread (5 high/low EMA scalping,) I just wanted to say I scripted it and ran it over a year’s worth of data across a few pairs and a few timeframes… and the result proves there’s no edge to be found.
Issues are as followed:
[li]Gunning for 5 pips with a stop of 8 makes the only effective timeframes 15min (general) and 30min (only with some pairs.)
[/li][li]Like some of you have noticed, losers are often the result of a candle opening beyond the 5EMA high/low but moving further away and causing the 5EMA line to swallow up the entry price… This isn’t a false signal, since it technically signaled correctly, it’s just a product of the EMA line being drawn to include the current bar’s price info, and thus, a long bar will drastically alter the value of EMA (it’s a front weighted moving average after all.)
[/li][li]Assuming a perfectly behaving pair, with default values for the stop and TP, the expectancy of the system ends up being a loss of the spread. This means there’s no edge, and the outcome is random, so each time you’re just playing the spread. (In the example below, I used a spread of 1.5 pips on EUR/USD, and the expectancy was -1.6 pips.)
[/li][li]Adjusting the values of TP, SL, and even adding in my own ideas to filter out trades never improved the expectancy to a better value. (I tried filtering out the “near line” signals by saying the open price must also be a given pip amount away from the line. While this filtered out a lot of trades, it didn’t improve the expectancy.)
Most results end up looking like this:
So I think this puts the issue to rest… the idea looks awesome on paper, since when you visually go back and look at charts you’re seeing the ‘post’ price movement of EMA, and this cuts out all the ‘near line’ bad trades. But in forward testing, the system does not give you any edge over randomness.