All updates

Recommendation Quality Update

Clearer signals, stronger Planner picks, and fewer high-profit traps.

I pushed a recommendation quality update focused on one thing: helping you find flips that are worth your time, not just flips that look good in a single column.

What changed

  • Planner Auto-Fill is more selective. It weighs profit, recent buy/sell support, estimated close time, GP/hr, and your budget before filling slots.
  • Planner can leave slots or GP unused when the market does not have enough supported opportunities. I would rather show fewer strong ideas than pad the plan with weak ones.
  • Recommendation labels are clearer. Tags like Profit + Volume and Fast Value are meant to explain why an item is being surfaced.
  • Score language is now Timing. That number was always closer to a timing signal than a total item grade, so the UI now says that more directly.
  • Planner cards now show estimated GP/hr and estimated close time, with better tooltips explaining what those estimates are based on.
  • High-profit items with weak execution support can show an execution-risk warning on the free table.

Why I made the change

A tiny 8K profit flip may be safe but not worth babysitting. A giant profit number may look exciting but fall apart if the item barely trades, the sell side is stale, or the graph is mostly one-sided volume.

The goal is to get closer to the question real users are asking: “Given my capital, what should I actually try right now?”

A note on estimates

GP/hr and close time are comparison tools, not promises. They use displayed profit plus a conservative share of recent and baseline buy/sell volume. Markets move, other players compete, and OSRS liquidity is messy, so the UI now tries to be clearer about what is estimated.

What comes next

I also added privacy-conscious telemetry around broad recommendation categories and value bands. No item-level trade details. The point is to learn whether people are keeping Planner picks, avoiding risky profit-sort traps, and actually using the better-supported recommendations.

This update is a step toward fewer flashy traps, more realistic opportunities, and a recommendation system that keeps getting sharper over time.

All platform updates