not sure delisted always means bad for a backtest

Been stuck on my first proper backtest for two weeks, mostly because I only had the current ticker list and kept getting suspiciously nice equity curves. Then I read that piece about all the names that quietly disappeared over a few months, and yeah, it rattled me a bit. But the one thing I'd push back on: it reads like every vanished ticker is a hole in your data where a loss should be. Plenty of them go away because somebody bought them, or they merged, or they moved listing. That's not the same as going to zero. Maybe I'm being naive here, the skew in what they were worth before vanishing does suggest a lot of them were junk. I just don't think "absent from the last five sessions" tells you which was which without the delisting reason attached, and I'm not confident I can get that reliably.
MLcurious
19 posts · 0 followers
+ Follow

Related reading

1 reply

TraderWeTraderWe Team· Sep 2026 ago
You're right that absence alone doesn't tell you the reason, which is why that research cross-checked against vendor delisting records rather than stopping at "gone". Those records do carry a security type and a date, and mergers/acquisitions are a real chunk of them. What we can't give you cleanly is the *outcome* - a buyout at a premium and a wind-down both show up as a delisting date. So your instinct is fair, but the direction of the bias still tends to favour the survivors, especially at the low-price end. If you want to be conservative, run it both ways and see how far apart the two curves sit.
Sign in to reply →
← All q&a