Feature Request: Minimum Pick Rate Threshold to Filter Low-Sample Skew

[Sicarignus]

Overview
I have been using DraftGap frequently and seeing the draft come together is fantastic. However, I have identified a specific issue regarding how the tool handles "off-meta" picks that persists regardless of current settings.

Currently, the tool tends to recommend niche, off-meta picks—such as Singed Mid or Ivern Top. While these champions possess high win rates in those roles, that data is heavily inflated by OTPs.

Why Current Solutions Aren't Working

1. "Ignore Individual Champion Winrates": Toggling this creates a new problem. It treats a 53% WR meta champion the same as a 46% WR weak champion. This "flattens" the data spectrum, penalizing high-performing meta picks while artificially boosting underperforming ones.
2. "Risk Level" Settings: Even when setting the tool to "Low Risk," these OTP-skewed picks often still appear. While Risk Level seems to weigh confidence intervals/variance, it does not act as a hard floor. Consequently, a champion with an exceptionally high win rate (even with a tiny sample) often bypasses the Low Risk filter, leading to recommendations that are statistically "strong" but practically useless for a standard draft.

Proposed Solution: Minimum Pick Rate Toggle
I request the addition of a "Minimum Pick Rate Threshold" (e.g., a slider or toggle to exclude champions with <0.5% or <1% pick rate in a specific role).

Addressing Potential Concerns (Data Cost & Integrity)
I anticipate a concern regarding data costs or sample size integrity, specifically regarding the Lolalytics API limits. However, this feature should not negatively impact those areas for the following reasons:

• No Additional API Cost: unlike requesting specific Rank filtering (e.g., "Diamond+ only"), which would require fetching entirely new datasets and doubling the cost imposed on Lolalytics, a Pick Rate filter is a post-processing operation. It operates on the dataset you have already fetched. It simply hides existing rows based on the pick_rate column rather than requesting new data.
• Negligible Impact on Sample Size: Filtering out low pick-rate champions will not create data scarcity issues. By definition, removing champions with a <1% pick rate only removes a tiny fraction of the total matches analyzed. The integrity of the data for the remaining 99% of the roster remains untouched.

Summary of Benefits
By adding a hard filter for pick rates, we can eliminate specific low % counter-pick noise and OTP inflation without sacrificing the accuracy of high-win-rate meta champions. This will bridge the gap between raw math and practical drafting, preventing the tool from suggesting Singed Mid simply because three specialists have a 60% win rate on it.