A wall of aged manila file folders stacked floor to ceiling in warm directional light, one folder with a red tab pulled slightly forward.

MakeStockMoney · Evidence brief · 19 September 2026

What 599 videos already told you

Every number here comes out of library.sqlite, not out of a best-practices article. Two of them contradict advice I gave you this morning.

01 · The shape of the business

It is a lottery, and the ticket is one video

599 YouTube posts carry a view count. Six of them are half the channel.

A ten by ten grid of grey squares with a single red square breaking the pattern.

One square in a hundred. That is the real distribution of your catalogue.

49.2%
of all views
from the top 6 videos
91.8%
from the top 10%
59 videos of 599
4,994
median video
the typical upload
13.07M
best single video
2,617x the median

So optimising the median is close to pointless. Lifting a typical video from 4,994 to 7,000 moves total views by a rounding error. What matters is how often you buy a ticket in a category that can produce a 13 million view outlier.

02 · The trap

Short videos do not win

Averages say sub-30-second videos beat 45-to-60-second ones by 5.1x. That finding is an artefact of five videos. The medians are flat.

Mean views Median views two panels, two scales, never one axis
Mean views by duration scale to 200,000 · outlier-driven Under 30s: mean 198,982 views, n=144 198,982 <30s 30 to 45s: mean 73,708 views, n=184 73,708 30-45s 45 to 60s: mean 39,038 views, n=200 39,038 45-60s Over 60s: mean 69,205 views, n=80 69,205 60s+ Median views by duration scale to 6,000 · the typical video Under 30s: median 4,027 views, n=144 4,027 <30s 30 to 45s: median 5,664 views, n=184 5,664 30-45s 45 to 60s: median 4,089 views, n=200 4,089 45-60s Over 60s: median 3,782 views, n=80 3,782 60s+ Same 608 videos in both panels. Only the statistic changes.
Remove the five biggest videos from each band and the sub-30s mean collapses from 198,982 to 21,359, while 45-to-60s lands at 13,713. The medians barely move: 3,859 against 4,044.

Correction to this morning. The V-Spec specifies 38 to 52 seconds on council advice. Your own data supports no duration rule in either direction. Build to whatever length the argument needs, and keep the 60-second ceiling only because a Short over 60s carrying a Content ID claim is blocked outright.

03 · The actual lever

Theme predicts. Nothing else comes close.

Grouped by the taxonomy already sitting in library.sqlite. One category runs 16x the channel average.

Average views per videolinear scale, sample size on each row
Politics, policy and markets: 1,597,719 average views across 14 videos Politics, policy & markets 1,597,719 n = 14 · avg 35s Company, stock and ticker analysis: 103,611 average views across 12 videos Company / stock / ticker analysis 103,611 n = 12 Bank and hedge fund insider culture: 100,141 average views across 19 videos Bank / hedge-fund insider culture 100,141 n = 19 Famous investor anecdote: 97,150 average views across 67 videos Famous-investor anecdote 97,150 n = 67 Trading psychology and mindset: 64,990 average views across 107 videos Trading psychology & mindset 64,990 n = 107 · most produced Crypto and bitcoin: 16,123 average views across 9 videos Crypto & bitcoin 16,123 n = 9 Billionaire wealth and lifestyle: 14,453 average views across 4 videos Billionaire wealth & lifestyle 14,453 n = 4 Pop culture money crossover: 14,174 average views across 4 videos Pop-culture money crossover 14,174 n = 4 AI and markets: 7,470 average views across 9 videos, the weakest named theme AI & markets 7,470 n = 9 · weakest
The top bar is 15.4x the second place. You produce trading psychology 7.6x more often than politics and policy, and it earns 4 percent of the views per video. An unlabelled bucket of 19 videos averaging 196,692 is excluded, because content with no theme tag cannot guide a decision.
Table view
ThemeVideosAvg viewsAvg duration
Politics, policy & markets141,597,71935s
Company / stock / ticker analysis12103,61140s
Bank / hedge-fund insider culture19100,14166s
Famous-investor anecdote6797,15052s
Trading psychology & mindset10764,99047s
Crypto & bitcoin916,123n/a
Billionaire wealth & lifestyle414,453n/a
Pop-culture money crossover414,174n/a
AI & markets97,470n/a

04 · The source, actually looked at

140 hours exist. 468 pixels wide.

This is a real frame pulled from the AIG derivatives session, at its true size. The dashed outline is a 1080x1920 Short at the same scale.

A frame of FCIC vice-chairman Bill Thomas speaking at the hearing table, shown at its native 468 by 263 pixel size.

Both at 25%. Source fills 5.9% of the canvas area.
June 30 2010, session 2. Public domain.

468×263
every session
mp4 and mov masters both
56
sessions
about 140 hours

Which settles the format argument. Upscaled 2.3x to fill a Short, this is mush. The sworn audio is the asset and the picture has to be authored. That is not a style preference, it is what the file is.

Screenshot of the Stanford Law FCIC archive page for the derivatives hearing, listing witnesses Greenberger, Kohlhagen, Kyle and Masters, each with a Testimony PDF and Video link.

The archive itself. Every witness carries a transcript PDF beside the video, which is why the pipeline indexes text first.

05 · The formula

Four variables that carry weight

Ranked by how much the data supports them, not by how good they sound.

Abstract line illustration: a house, a stack of tiles, a cylinder, a sphere and a cracked sphere connected left to right by red lines.

The securitization chain, the master map. Video 01 draws this and every later video points back to it.

The one sentence version. Pick the winning theme, produce far more inside it, and let the length follow the argument. Everything else in the catalogue is noise around those three moves.

Variable 1 · strongest

Pick the theme, not the format

Politics, policy and markets averages 1,597,719 across 14 videos. Trading psychology, your most-produced category at 107 videos, averages 64,990. You have been producing the weaker category 7.6x more often.

Variable 2

Volume inside the winner

Because the top 1 percent is half your views, the correct move is more attempts in the category that can produce an outlier. Fourteen videos is too thin to be your main line.

Variable 3

Reclassify, do not abandon

AI and markets sits last at 7,470. The doomer AI-jobs angle is not a new topic, it moves AI into the policy bucket. Name the ticker and the line item and it stays finance.

Variable 4 · unbuilt

TikTok is not a channel yet

29 TikTok posts carry a view count, averaging 632 against YouTube's 97,437. Best TikTok post to date: 5,267. The rails exist and are barely loaded.

Variables that did not survive contact with the data. Duration, in any band. Posting cadence, which the ledger cannot separate from theme. Caption style, which is not recorded per post and therefore cannot be claimed either way.

06 · The fit

Which is why the hearings work

Sworn testimony is not a new theme. It is a supply of material inside the two themes that already perform best.

1st
Politics & policy
FCIC is federal policy, on the record
3rd
Bank insider culture
executives under oath, named
140h
supply
56 sessions, public domain
$0
licence cost
NARA release, no claim

The honest caveat. Politics, policy and markets rests on 14 videos, and the theme labels are your own taxonomy rather than an external classifier. Some of the gap could be labelling. The direction is strong enough to act on. The exact multiple is not.

The production plan that follows sits in specs/fcic-shorts-engine.html, with one amendment: drop the duration rule.

Source · rooms/operations/msm-scheduler/library.sqlite, 599 YouTube postings and 941 clips, read 19 September 2026.
Frame · FCIC hearing, 30 June 2010 session 2, public domain via NARA, hosted by Stanford Law.
Charts · palette #9B2226 and #00688F, validated against the six-check dataviz validator on this surface.
Photography and illustration · generated, decorative, carrying no data.