[PYenner]

Sundowner
Thesis is an interesting thread and my impression is that the critical observations have obvious truth to them, macd and rsi do not have a compatible basis in their forex forms for combining as Thesis did. This is likely to be a translation defect not a defect in the original thesis. Also agree that the ranges of the two are incompatible and may only be compatible when used on a pair where they happen to be almost compatible. So the bias favouring long does not surprise me. These problems are probably fixable.
What also worries me is the dogs breakfast in the formula (Stochastic*(MACD+RSI)/Momentum). If this really was an MBA thesis then the maths behind that should be sound... but from memory (only) Macd may be one of those "big swingers" that makes no attempt to calibrate itself to the pair in use, ADX also makes this most basic mistake, no attempt made to ensure they work in one pip units on any pair.
So along the lines that others have said, before macd and rsi can be combined, they must both calibrate themselves to 1 pip steps on any pair (non standard versions of both may be needed), they both have to be in forms that swing about a zero line (non-standard for at least one of them) and they have to have compatible swing ranges.... otherwise it is all total cr*p and a waste of time. The short way of saying the same thing in my own way of saying it... all four variables may need to be normalized to a +1 to -1 range before combining them is mathematically sound, so I expect MBA used them in a normalized form in his original version, mathematical types think in normalized terms, its an essential basic to their thinking, the alternative is just not an option.
The formula looks less suspicious to me if it is rearranged as (MACD+RSI)*(Stochastic/Momentum) but everything then depends on the "periods" that are used particularly in Stochastic/Momentum which is the seriously interesting concept because it may be attempting to get an early warning signal out of an ordinary stochastic signal. Really need to hunt down an online version of the guys thesis which most likely does not exist online. Without that you are shooting in the dark.
The observation that it works on 1hr charts but not on 1min charts bothers me, it may be because of the "big swing" problem that comes from the instruments not being calibrated to work in pip units.
So there is not just one issue here but many, dogs breakfast, not a pretty sight. That Stoch/Momentum step is interesting and I would like to understand that part. But there are so many things that would need to be fixed first, its a big project, unfortunately the lack of a common systematic scientific approach to the design of "standard" instruments like adx macd and rsi turns it into a witches brew not sound chemistry.

Thanks for the interesting thread but the instruments need to be redesigned on a common systematic basis before they have a chance of being used in the Plug'nPlay way being attempted. That is the first step to turning indicators into science but I won't be making that step, I must get on with more pressing things unfortunately.
Regards
Bruce