nyasekai

Graveyard / Equity events / US stocks

Buying new spin-off stocks

Do newly spun-off companies outperform after they list?

Family
Equity events
Verdict
Dead
Cause
No edge
Buried
10 October 2026

The idea

When a parent company spins off a subsidiary, shareholders who never chose it (index funds, mandate-bound funds) sell without caring about price. The new stock should then be cheap and recover over one to three years.

How we tested it

DataSpin-offs identified from Form 10-12B filings with a NYSE or Nasdaq ticker, plus 15 delisted spin-offs mapped by hand, with daily prices. Close at least $5 and dollar volume at least $5M on listing day. Buy at the first close, hold 250 sessions, short an equal dollar amount of IWM, 30 slots.
PeriodJuly 2019 to October 2026, split at 25 April 2025.
Costs10 bp per side plus 1 bp on the hedge; stress at 1.5x.
HoldoutFull-period Sharpe, in-sample Sharpe and out-of-sample profit factor all had to pass.
ControlNo random-entry control was run. An event study of average returns by horizon is reported.

Result

Trades51
Sharpe, full period0.64
Sharpe in-sample0.31
Out-of-sample Sharpe / profit factor1.96 / 1.11
Profit factor3.77
Max drawdown−15%
Mean vs median trade+43% vs +3.4%
Largest winner+1,635% vs IWM
Event study, 20 / 60 days−0.7% / −1.4%
Sharpe in bull / bear / chop markets1.66 / −1.15 / 0.43

Why it failed

It failed the Sharpe and out-of-sample bars. The mean of +43% is one lottery ticket: a single memory-chip spin-off in 2025 returned +1,635% and carries the average, while the median trade made +3.4%. The early weeks after listing are negative, which fits index-holder selling. Spin-offs that were acquired or failed may be missing from the sample.

What would make it worth retesting

A variant that waits 20 sessions after listing looked better (+14.9% over 125 sessions, t 3.0, 54 events), but it was picked after seeing the data. It needs a pre-registered test on independent data (2005-2018 spin-offs with delisted names) plus new listings from late 2026.

Past results of a backtest say nothing certain about the future.