The first part of this series measured what it takes to get on each list. This one measures how long
anybody stays.
Both lists have ten slots. That is the only thing they have in common: the top gainers list
replaces itself, and the dollar-volume list barely moves.
1. What survives to the next session
For each pair of consecutive sessions we counted how many of today's names were still there
tomorrow.
| List | Names kept, median | 10th percentile of sessions | 90th percentile |
| Percent, top 10 | 10.0% | 0.0% | 20.0% |
| Percent, top 30 | 6.7% | 3.3% | 16.7% |
| Percent, top 50 | 10.0% | 4.0% | 18.0% |
| Dollar volume, top 10 | 80.0% | 70.0% | 100.0% |
| Dollar volume, top 30 | 83.3% | 76.7% | 90.0% |
| Dollar volume, top 50 | 82.0% | 76.0% | 88.0% |
The percentage list keeps one name in ten. On a tenth of sessions it keeps none at all, which is
what the zero at the 10th percentile means. The dollar-volume list keeps eight, and on a tenth of
sessions it keeps every single one.
Both are far above what chance would give. Drawing ten names at random from a qualified universe of
6,193 would repeat 0.16% of the time. The point is not that either list is random. It is that one of
them is eight times stickier than the other.
2. How many names it takes to fill the list
Across 181 sessions the top ten by percent has 1,810 slots to fill. It filled them with 957
different companies.
| List | Distinct names, top 10 | Appeared exactly once | Distinct names per month |
| Percent | 957 | 58.6% | 155 |
| Dollar volume | 40 | 20.0% | 20 |
Nearly six in ten of the names that reached the percentage top ten did it once and were never seen
there again. To follow that list for a month you would have met 155 different companies. To follow
the dollar-volume list you would have met 20, and you would have met most of them again the
following month.
The ratio between the two is 23.9 to one at the top ten. That is the size of the difference in what
these two lists are asking you to keep track of.
3. Regulars
We counted names appearing on at least half of the 181 sessions.
| List | Regulars in the top 10 | Top 30 | Top 50 | Slots taken by the ten most frequent names |
| Percent | 0 | 0 | 0 | 5.1% |
| Dollar volume | 8 | 25 | 39 | 78.0% |
The percentage list has no regulars at any depth. Not one company in nine months reached its top
fifty on half the sessions. Its ten most frequent names account for 5.1% of all slots, against 78.0%
for the ten most frequent on the other list.
The dollar-volume list is the opposite object. Eight of its ten slots are held by names that show up
on most days, and the ten most frequent take 78.0% of every slot in the sample. Three of them, SPY
(181), QQQ (181) and NVDA (181), were present on every session we measured.
On the percentage side the most frequent name managed JLHL (12) appearances out of 181.
4. What this means in practice
A watchlist built from the percentage list is not a watchlist. It is a queue. Whatever you learned
about the names on it yesterday applies to one of the ten today, and the tooling that matters is the
tooling that handles strangers quickly rather than the tooling that remembers.
A watchlist built from dollar volume is a watchlist, and a short one. Twenty names a month covers it,
and the same names return, so knowledge about them compounds.
Neither is better. They are different jobs, and choosing between them decides how much of your
preparation can be reused tomorrow.
5. What this does not measure
We measured membership, not behaviour. Whether a name that stays on a list does anything different
from one that leaves is a separate question, and we did not touch returns here.
Daily bars decide membership at the close. A list built intraday from a live scanner will not hold
the same names, and it will turn over faster still, which we have measured separately in
https://traderwe.com/research/half-the-stocks-on-your-day-trading-watchlist-at-09-35-are-gone-by-10
The percentage list here uses the tradable filter established in part one. Without it the turnover
would look even higher, because the raw list reaches further into names that trade once and vanish.
6. Sample accounting
The sample is 181 sessions between 2025-12-02 and 2026-08-21, spanning 9 calendar months. The
qualified universe used for the percentage ranking had a median of 6,193 symbols per session after
requiring a $1 close and $1,000,000 of turnover.
Exchange test symbols are excluded from both lists. Part one describes why: one of them entered the
qualified top ten on a third of the sessions in this sample.
Retention is measured between consecutive sessions in the sample, so a gap in the file series is
treated as consecutive. Every figure is the median across sessions unless a percentile is named.
