How these numbers are made
Every date and ratio on this site is read from a filing and links to it. The returns are the exception — they are arithmetic over closing prices, and this page explains exactly how that arithmetic works and where it is weakest.
Day zero is the distribution date. The anchor often is not.
A separation has a distribution date, stated in a filing. It does not follow that the shares traded normally that day. Most distributions take effect after the market closes, so the distribution date itself is still a when-issued session — a thin, provisional market in shares that do not exist yet. Returns here are measured from the first regular-way close instead.
The difference is not academic. Comcast's Versant closed at 46.65 on its distribution date on 206,000 shares, and at 40.57 the next session on 41.4 million. Anchoring on the earlier price would overstate the return by about fifteen points. Of the 68 completed separations tracked here, 19 state their first regular-way session in a filing, and that stated date is what is used; for the rest the anchor is identified from the collapse in trading volume that marks the when-issued market ending.
One case shows why a filing beats an inference. When a spin-off inherits its parent's ticker, the sessions before the distribution are the parent trading normally, and no volume test can see the handover. MSG Entertainment closed at 60.56 on 20 April 2023 and 30.99 on the 21st. That is not a crash; they are different companies.
Two benchmarks, because one is misleading
Spin-offs are usually small companies. Measuring them against the S&P 500 imports the gap between large and small companies into every figure — and across this period that gap was wide. Measured deal by deal over each separation's own window, roughly a third of the shortfall against the S&P is company size rather than anything about the separation. Both benchmarks are therefore shown, and no claim here rests on the S&P alone.
Neither benchmark is industry-matched, which is the more rigorous comparison and the one the academic work uses. That is a known limitation, not an oversight.
Every leg is measured over the same window
The spin-off defines the period; the parent and both benchmarks are measured across exactly the same dates. Where a parent stopped trading part-way through — acquired or taken private — its return is left blank rather than shown over a shorter span, because a number covering a different period in a column headed “over the same window” is worse than a gap.
Companies that were acquired stay in the table
Several spin-offs here were later bought or taken private. Their returns run to the day trading stopped, and they remain in the sample. Removing them would flatter every figure on the site, because being acquired is frequently the best outcome a spin-off has — and the research most often quoted in favour of spin-offs found its positive result concentrated in exactly those companies.
What a blank means
A blank is never an estimate that has been left out and never a guess that has been suppressed. It means no filing states that fact, or that the price data does not cover the period needed. Each blank return names its own reason.
Returns are arithmetic over end-of-day closing prices and are the only figures on this site without a filing behind them. Further detail, including the sources behind the research summarised on the research page, is kept with the project's own documentation.