Designing an edge

Instead of searching for an edge through data mining, how about designing an edge?

I came up with a theory how the market "should" move and coded two oscillators that show how the market is moving and how it "should" move.
The Primary Momentum is based on actual price movement and the Secondary Momentum is the model. If the market were efficient, the two oscillators should move in unison. When the Primary Momentum moves up or down and the Secondary Momentum does not confirm the move, an inefficiency exists and price will likely reverse to liquidate the difference.

I attached a few charts for the last few hours of the last trading session (PST) showing how the two oscillators indicate likely reversal points.

 

So.. Did you tried to backtest your oscillators on long run? All I see is two indicators calculating something.

 

Welcome!

Something approximating a (fractal) sine wave?

If the market was efficient, shouldn't it be random walk?

How do your generated reversal signals compare to price/momentum divergence signals?

Without more details it's difficult to know if you have something or not.

Are you a neophyte?

 

Interesting idea, but what are the hard-and-fast rules for placing long or short trades ?

 

OP (aka Mr. "1 post") is no doubt laying the groundwork for a sales pitch.

 

How do your generated reversal signals compare to price/momentum divergence signals?


In the case of price/momentum divergence, you compare a function with its first or second derivative, i.e. the time series with its corresponding speed and acceleration. They show devergence because you compare the original function with its transform.

My two oscillators are of the same order but are driven by different input information. Changes in speed or acceleration do not make them diverge. They diverge only when the input information of the Secondary Momentum (the model) has not been factored in price.

 

Eh?! Who says it should move the way you say it should?

 

What you're saying could very well be going over my head. It could be interesting.

Is it intended as an improvement on standard momentum indicators or as a complete system where you'd act on all the buy/sell signals?

 

Is your model of how it should move going to be correct going forward?

 

I decided to show the inefficiencies in the form of divergences between two oscillators (of the same order, frequency, normalization, etc.) because it possibly is the easiest way to see them.
This not a complete trading system yet. This method generates so many signals, that only the "best" could be taken - for example, only with the trend signals or when price is flat. Also, an exit method has to be devised - possibly volatility based or the location of the Primary Momentum, etc.