The drop chance formula is influenced by several factors, notably the Increased Rarity Drop Rate. However, not all currencies are affected equally by this modifier. Currencies such as the Exalted Orb, Chaos Orb, and Divine Orb tend to be more significantly influenced by the Increased Rarity modifier than others, despite having lower individual drop rates compared to some of the rarer items.

We have also observed a diminishing returns effect once the total rarity reaches around 120. This encompasses all sources, including gear, maps, and modifiers. This phenomenon helps explain why farming high-tier maps might not yield expected results compared to farming low-tier maps, especially with a setup focused on maximizing rarity.

In order to create an accurate representation of these dynamics, I combined my own mathematical knowledge with the power of machine learning. Using a Python-based machine learning model, I was able to identify patterns, analyze data, and convert these insights into actionable numbers. The training of the model is similar to a form of mathematical engineering reverse applied to data. Instead of dismantling a physical object, the model "reverses" the hidden processes behind the drop system by learning from existing examples and inferring the rules that govern the behavior of the system. This method allows us to uncover complex relationships that are not explicitly documented but are observable through patterns in the data.

In addition to our own analysis, I have collected a wide range of data from Reddit and firsthand player reports, which include testing various rarity setups, including gear, maps, and tier adjustments. This data helped refine our understanding of the system and improve the accuracy of the model.

The calculations presented here are designed to provide a reliable estimate of drop chances, but it is important to keep in mind that there may occasionally be exceptions or adjustments to the system that we have not yet accounted for.