
Free 2x Tickets to Odyssey IMAX at 13:40 Today
I can’t make it so don’t want it to go to waste. Let me know if you make it.

I can’t make it so don’t want it to go to waste. Let me know if you make it.
What are some good pubs/places to watch tonight in the city center?
- Cafe Faber
- Tierney’s
- Shamrock??
White paper documenting a multi-factor equity ranking system for small-cap U.S. stocks. The system blends 51 factors via coordinate descent within a 19-fold expanding-window walk-forward design covering 1 January 2000 to present. The factor list is the union of two pre-existing frameworks: a publicly disclosed Portfolio123 (P123) composite contributing 33 factors, and a separate ElasticNet study contributing 22 factors selected via out-of-sample walk-forward from a candidate pool of 299.
A possible deployed configuration is 15 stocks, weekly rebalanced, in a liquid P123 universe. The paper covers methodology, scalability across universe sizes and position counts, slippage sensitivity, sub-period stability across five non-overlapping windows, and a Limitations section that names framework selection bias, capacity constraints, and behavioral risk as the most important concerns.
Posting here primarily for record-keeping. The paper treats headline backtest figures as upper-bound reference points and specifies a 25-50% haircut as a planning convention. A follow-up paper reporting actual realized results against these benchmarks will be published in May 2028, regardless of whether those results are good, bad, or mixed.
https://drive.google.com/file/d/1Ne_GdrlI5hPDu-Bu9sbFV1-T11Cic0eY/view?usp=sharing
RemindMe! 2 years
https://drive.google.com/file/d/1Ne_GdrlI5hPDu-Bu9sbFV1-T11Cic0eY/view?usp=drive_link
This is a white paper for a multi-factor stock ranking system, most suitable for small-caps. The ranking system blends 51 factors from value, growth, quality, and momentum primarily. I used two frameworks (a community-sourced ranking system with OOS success & an internally built ElasticNet ML model) and coordinate descent with walk-forward validation of expanding windows to come to these factors. I tried to minimize overfitting the best I could.
I'm posting here for commentary but also record-keeping. The headline backtest numbers are probably unrealistic as time goes on, but even with a 30-40% shave, I'd be thrilled. Anyways, I want to trade this for two years then come back here for a follow-up.