What premarket volume does and doesn't tell you

Same conclusion, took me longer and cost more to get there.

"Volume says there is a story, not that it ends well" — that's the cleanest phrasing of the distinction I've seen.

Filter not signal is the right frame. The bit I'd add: heavy premarket volume tends to mean wider spreads and messier fills at the open, so if a name passes your filter, size it smaller than a quiet one, not bigger. People do the opposite because it feels exciting.

Crypto version of this is 24/7 so there's no premarket, but funding rate does the same job for me — tells me a crowd exists, tells me nothing about which way they're about to get flushed. Learned that at 10x, which is a story for another day.

The way I think of it: premarket volume shifts your prior that the name is tradeable at all, but it barely moves your prior on direction. Two different questions, and conflating them is how people end up with a high hit rate on "something happened" and a coin flip on "I was right." Only the second one pays.

Two things I found when I tried to test this properly: 1) any absolute volume threshold I picked was garbage across time — had to normalize against the name's own average premarket volume, otherwise I was just building a mega-cap detector. 2) once normalized, the filter did separate "tradeable" from "dead" reasonably, and did nothing at all for direction, which matches what you're saying. Warning attached: I burned a couple of weekends tuning the exact multiple and the improvement vanished out of sample. Pick a round number and move on.

So the indicator that everyone stares at for 90 minutes before the bell is useful for deciding which names to ignore. Sounds about right. I'd go one step further than filter. I don't look at it until the open prints and I can see whether the range actually holds, because a gap on volume that immediately gets faded tells me more than the volume ever did. Rita's point about smaller size is the only thing in this thread I'd tattoo on anyone.