I've been playing with a slightly modified version of the example for PearsonCorrelationPairsTradingAlphaModel. (I  added the possibility of selecting only assets with a specific MorningstarSectorCode via a fine filter).

In Backtesting as well as paper trading, I noticed that the Alpha model seems to have created insights that closed only the short half of the pair, whilst keeping the long side open. In the attached backtest for example the RTX short closed on 28/07 with the BA long kept open. So I've been wondering if this is how the algorithm is supposed to work - wouldn't a classical stat arb algo always open and close pairs in synch and thus always have an equal and opposite  short/long position?

I haven't looked at the PearsonCorrelationPairsTradingAlphaModel class in any detail, so my apologies if this is a daft question.