WorldQuant BRAIN

1. Part 1Trying to ship one alpha, I built an alpha factory

   첫 알파를 올려보려다, 어쩌다 알파를 자동으로 찍어내는 파이프라인부터 짓게 된 하루의 기록.

2. Part 2The 0.95 wall, and orthogonality as the honest way through

   1편에서 멈춘 0.95의 벽을, 상관 없는 직교 신호를 섞어 넘어보려 한 하루의 기록.

3. Part 3My first submittable alpha: cash-flow reversion

   단기 가격도 교과서 펀더멘털도 다 막혔는데, '현금흐름 대비 주가의 평균회귀' 한 줄이 마침내 제출 기준을 넘었다.

4. Part 4Multiplying signals, and the real rules of the scoreboard

   Multiplying the proven pcf signal by analyst revision lifted Sharpe to 1.91, and I learned the real rules of the daily scoreboard.

5. Part 5Outside the family: competitors, options, and their intersection

   The intersection of competitor spillover and option call-put implied-volatility spread became the third alpha, with a self-correlation of just 0.14 to the existing ones.

6. Part 6Building a factory, hitting a wall, then changing the method

   After eight variants of the same combination stalled at the same Sharpe ceiling, scaling the competitor-return signal by the number of linked competitors produced the fourth alpha.

7. Part 7Crossing silver, stopped 0.0063 short

   The fourth alpha pushed the score into Silver, but the fifth candidate missed the self-correlation gate by 0.0063 and was abandoned after more than twenty tweaks.

8. Part 8Do not multiply, add

   Multiplying three signals collapsed from concentration, but adding their z-scored ranks produced a well-balanced fifth alpha that cleared every gate.

9. Part 9The sixth alpha I could not find, and what remains

   The fifth alpha brought the score within 571 points of Gold, but two new directions, network centrality and conditional trading gating, failed to yield a sixth alpha.

10. Part 10Past Gold, then across the last 0.05

    The sixth alpha lifted the score to Gold at 11,394 points, and the seventh crossed a 0.05 sub-universe gap through diagnosis and weight adjustment.