Deals

How the take-private model works

The model learns from 144 buyouts by private-equity firms, sponsor-backed platforms, holdcos and insider groups across 11,725 company-years. It weighs size, goodwill from past acquisitions, activist 13D stakes and what each 10-K says about the business. In backtests on 2020–2024, companies in its top tenth were about 3.8× as likely as average to be taken private.

Which signals showed up before a deal

Every company with at least $500M of public float, as of April 1 each year from 2019 to 2025, scored only on what it had filed by then, including companies since bought and delisted. 541 of 11,725 company-years (4.6%) had a merger proxy, tender offer or going-private filing within the next 12 months. Lift is how much more often that happened when the signal had fired.

* Read from the 10-K text, which comes from a sample of 3,244 company-years: every company that later had a deal, plus five random others for each, weighted back up to the full panel. Treat these two as estimates.

Not in the backtest, because their history isn't kept: TSR below peers (pay-versus-performance tables start with fiscal 2022); High CEO pay despite TSR (pay-versus-performance tables start with fiscal 2022); Shareholder dissent (vote results are kept for two years); Net insider selling (form 4s are kept for two years).