StashSage for POE2

Does this sound like you?

• You pick up items that could be good, but aren't quite sure
• Your stash fills up with "I'll price this later" items
• Pricing through the in-game trade UI feels tedious, so you avoid it
• Typing mods, comparing rolls, and overthinking prices takes more effort than it's worth
• You don't need a perfect price - you just want a smart starting point
• One hotkey, instant comparable items, list it, and get back to mapping

1. Models are trained on current, unsold listings, so the data can be noisy. Even reasonably priced items may not sell.
2. Models #1 (XGBoost) and #2 (KNN) use different approaches, so their predictions may sometimes disagree. Both provide useful perspectives on an item's potential value.
3. The Model #1 visualization shows how an item's predicted value compares with other items in the same category.
4. XGBoost is a tree-based model that evaluates the value of each modifier and how modifiers interact with one another. It uses patterns from all items in the dataset—their modifiers and listed prices—to estimate how the target item's complete set of stats may affect its value.
5. KNN uses XGBoost's modifier weighting to identify the items in the dataset most similar to the target item. These “nearest” items do not need to match perfectly; they can still provide useful comparisons even when some modifiers differ. The mean and median prices of these comparable items are then used to estimate the target item's value.
6. Careful when using Model #2 (knn) to price items with less than 6-mods, comparing to 6-m items can be tricky. Ask yourself, is a 4-m item more similar to (a) a 4-m item with the same 4-m but very different values, or (b) a 5-m item with 3 of the exact same mods and values, and 2 non-overlapping mods?
7. Models trained to price rare items, not magic or unique.
8. Overlay is optimized for Windowed Fullscreen only.
9. New models trained every few days and posted for download.

1. The model ignores "mark of the abyssal lord", "allocates passive", "on corruption" mods.
2. The model ignores socketables as they are not considered an underlying modifier.
3. Items with quality and socketables have had their modifier values normalized to their base values.
4. Armour, evasion, energy shield mods are represented in the base stat values.

Interested in learning more about other DataWiseGuys projects? Check them out below.

SocialSync - social media orchestration and automation with optional AI for small businesses.

Want to brainstorm developing your own app with analytics or machine learning? Reach out to us.