Spread Difference with Corrective Multipliers + Bias

User avatar
dataheck
Posts: 9
Joined: 19 Nov 2019
Location: Toronto, Ontario
Contact:

Spread Difference with Corrective Multipliers + Bias

Postby dataheck » 04 Dec 2019

I took the default spread difference indicator and ran with it, creating the attached. It applies a multiplier to the data series and a bias term. For example, you could use this to trade a linear regression. It supports error terms, so you could plug in your standard error constants for your regression coefficient and bias and see the bars around your regression.

I've attached an example picture from a (bad) regression of robusta and arabica coffee. It infers the price of robusta from the price of arabica. I am not recommending that as a trade, just showing it as an example of the indicator.

There's a bit of a trick with this when it comes to scaling. I recommend plotting this on your data series 1 and setting the scale to the same as the instrument, otherwise you'll see some weird behaviour. You can see a quick example of that by turning the error bars on and off.

Also, data series two is subtracted from data series one. You may need to make your multipliers negative depending on what you're plugging in.

e:

Adjusted to use a fixed confidence interval instead. I was using the wrong approach to consider error. The standard error of a coefficient is not useful at all for determining the expected range of the predicted value. You have to look at the distribution of the residuals instead.
Attachments
Spread_Diff_Corrective.pln
(1.53 KiB) Downloaded 8 times
Annotation 2019-12-04 103348.png
An example use case of this indicator. The middle line is the regression of Arabica coffee, used to infer what the price of this series "should" be. The dashed bars are the low and high standard errors expected from the quality of the regression (this is a bad regression)
(21.94 KiB) Not downloaded yet

Return to “User Contributed Studies”