Noticed a slight error in the calculation of accuracy. The last part was calculating accuracy of all my predictive models, not the best performing ones.
The results look a lot more promising, given that I haven’t applied the recession forecast model into the price adjustment yet.
The spread accuracy is currently at 85%, well above my target of 80%.
Forecast error has also been reduced to 21%, well below my target of 25%.
Model is still imperfect since I’m using 1.25std as the spread vs. the 1std that I want to be at. I’ll be reviewing the performance at a sector, subsector and stock level to see if there are any patterns that I can implement to improve performance without overfitting. Stay tuned for more updates.