Added Relative Conviction — a trade's size versus the member's own estimated wealth.

• A 7th ex-ante factor, Relative Conviction, now contributes to the score (default weight 15, alongside the existing six). It measures how large a trade is as a share of the member's own estimated net worth: a $50k trade is a rounding error for a nine-figure fortune but a 10% bet for a member worth $500k.
• Net worth is estimated from the annual Financial Disclosures, which report assets and liabilities only in ranges — so it is a range, and the score uses the member's wealth as known AT the time of the trade (never a later filing), keeping the factor fully ex-ante.
• A data gap never moves a ranking: when we have no wealth estimate for a member, Relative Conviction is left out entirely and the score is exactly the original six-factor score — no neutral fudge, no penalty.
• When the factor DOES apply, the score is renormalized over the new total weight (100 + 15 = 115), so the final number stays on the same 0–100 scale rather than inflating — the six original factors keep their relative proportions, Relative Conviction takes its 15/115 share, and the result is still percentile-normalized across all disclosures.

First public version of the Interest Score.

• Launched the ex-ante Interest Score: a 0–100 ranking built only from information knowable at filing time, with no look-ahead to how the stock later moved.
• Six weighted components: recency (20), size (20), consensus/clustering (18), unusualness for the member (15), the member's trading history (15), and committee jurisdiction over the traded sector (12).
• Post-trade price movement is measured and shown separately as an outcome — deliberately kept OUT of the ranking score, so the score can be evaluated honestly against returns it never saw.