Orthogonal Spread Analysis

Here we apply an orthogonal regression (also known as total least squares) to both price series. We are interested in some key statistical properties (like β, ...) and in analysing orthogonal residuals:

Cointegration Analysis

In this section, Engle-Granger cointegration test is performed using OLS regression method in both directions. Residuals of the regression are plotted and analyzed further:

RIO / VALE

VALE / RIO

Profit Analysis

Profit analysis is a set of backtests performed using multiple pair trading models over significant portion of parameter space. Backtest results are displayed in a form of scatter plot. In addition, you can see aggregated statistics in the left panel. All backtests are performed using 100% margin (leverage 1:1). This is the profit analysis where backtested strategies are allowed to open both long and short positions:

RIO / VALE

VALE / RIO

This is the profit analysis where backtested strategies are allowed to open long positions only:

RIO / VALE

VALE / RIO

This is the profit analysis where backtested strategies are allowed to open short positions only:

RIO / VALE

VALE / RIO

Profit Analysis (Inverted!)

This profit analysis has been performed using strategies with inverted signals and transformed prices. Mostly useful for trading pairs composed of an inverse ETF in order to stay market-neutral (more info). "even" means that the first leg is both inverted and transformed, "odd" means that the first leg is transformed but the second leg is inverted.

Backtested strategies have been allowed to open both long and short positions:

RIO / VALE (even)

RIO / VALE (odd)

VALE / RIO (even)

VALE / RIO (odd)

This is the profit analysis where backtested strategies are allowed to open long positions only:

RIO / VALE (even)

RIO / VALE (odd)

VALE / RIO (even)

VALE / RIO (odd)

This is the profit analysis where backtested strategies are allowed to open short positions only:

RIO / VALE (even)

RIO / VALE (odd)

VALE / RIO (even)

VALE / RIO (odd)