Every message from the Nov/Dec 2025 subathon placed on a 24-hour clock, in Eastern time. This is the shape a 42-day subathon makes: a long overnight tail and a dead zone at dawn.
Cheers on the same clock, drawn as its own chart rather than a second axis. Bits and messages are different units, and overlaying them would invent a correlation.
Subscriptions, gifts and raids are not recoverable here: Twitch’s VOD chat export omits them. Bits survive because they ride on ordinary messages.
The whole run, day by day.
Who did the most talking across the whole subathon. Bots excluded.
Counted by messages, not by screen. 220 of them carried an awoo, and 164 were the wolf itself drawn in braille. The largest a single message of 453 characters. Each of those counts once here, however much of chat it filled.
How long each category was actually being broadcast, from Twitch’s own reading once a minute. Colours match the bands under the clock on the Real time tab.
Gift bombs are counted once per sub they delivered, not once per announcement, so this adds up to subs rather than to messages.
Every gifted sub, split between the ten most generous contributors and the rest of the room.
The same split for bits cheered.
Every contributor sorted into a band by how many subs they gifted, then the share of all subs each band accounts for. Shaded light to dark as the bands rise.
A resub announces the streak it is on, so this is tenure as Twitch reports it rather than an estimate. Gifted subs carry no streak and are not counted here.
Loading…
The red line is when the subathon ends if nobody adds more time. It only moves when someone subs, cheers or donates. The dashed lines show where the end would be if people keep giving like they did in the last six hours.
REAL EDGERS USE ZCHUM.TV TO AVOID TWITCH LAG
How long the clock has gone without a close call. A close call is a dip under one minute.
How many times the clock fell under each line. A dip counts once the clock had been above that line for at least ten minutes.
How long the clock has spent at each level since the subathon started.
Hours the clock spent in each range of time left. The ranges get wider as the clock gets higher.
People who added time while the clock was under one minute. A close call counts once per person. Donations only count when they came with a media share. ZChum keeps his own board of top edgers at zchum.tv/clock-edge-heroes.
The ten lowest the clock has been. Saved by is the first person to add time after the lowest point. A dash means the time came without a name. Plain tips have no name.
Close calls by the hour they started. Times are Eastern.
Close calls under one minute on each day of the subathon.
Loading…
Clock gained is how much the clock went up during the train.
We have a program that watches the hype train banner on ZChum’s channel. So far no train has changed type partway through. A treasure train is a treasure train from the start, and the same goes for golden kappa trains. We’ve only seen 149 trains, though, and just 10 of them were treasure trains.
Each column is one train. The colour is the type of train, and the height is how high it got. Columns that start higher up skipped the “incoming” countdown.
We looked at 107 things in the 20 minutes before every train, like chat, subs, cheers, viewers and who was on camera, to see if anything gives away a treasure train before it happens. Nothing does, so far. If something did, the gold bars would be bunched up on the left.
Each bar is one train, ranked by how much the 20 minutes before it looked like the time before a treasure train. Random trains rank this well about 1 time in 6, so the treasure trains don’t stand out. We’d need more treasure trains to know for sure.
Each mark is one hype train, from the first day to now. Treasure trains are the gold bars and golden kappa trains the gold outlines; a mark is as wide as its train ran.
Minutes put on the clock for each hour a game was on. Double time counts at the normal rate. The black mark is what other games got on the same days.
Games that were on for less than five hours are left out. Sleep is not counted in the marks.
Hype trains that started during each game, per 10 hours it was on. Treasure and golden kappa trains count too. The black mark is what other games got on the same days.
Games that were on for less than five hours are left out. Sleep is not counted in the marks.
How busy chat was during each game compared with other games on the same days. Commands like !play and !boost are not counted. Bots are not counted either.
Games that were on for less than five hours are left out. Sleep is not counted in the comparison.
Click a column to sort it. Viewers is the average while the game was on. Chat a minute leaves out commands and bots. Close calls are the times the clock fell under one minute.
Each row is one day of the subathon, from midnight to midnight Eastern. The colour shows what was on. Blank stretches are when the stream was down.
How much of each game’s hours each of them was on camera. They are often on camera together. The two sides can add up to more than 100%.
Only games that were on for at least ten hours are shown.
A close call is a time the clock fell under one minute. It counts for the game that was on. Dips less than ten minutes apart count as one.
How many different games had been played by each day. Each dot is a game’s first time on stream. The ten biggest games are named.
How much of the chat, subs and bits happened on each day of the week, in Eastern time.
Each dial is a 24-hour clock in Eastern time, with midnight at the bottom and noon at the top. Bigger wedges mean busier hours. Each dial has its own scale, so they show when things peak but can’t be compared to each other.
Each row is a day and each column is an hour, in Eastern time. Darker means more chat. You can see the late nights on the right.
YES, IT DOESN’T MATCH TWITCH. MOUSE OVER ANY ROW FOR THE DONO AUDIT. The two boards start counting at different times: Twitch’s Top Supporter Event opened at 7:30 PM ET, thirteen minutes after the subathon did. Switch between them here.
Ranked by subs gifted to other people, in gift bombs or one at a time. Your own sub or resub doesn’t count.
Counted from the gift sub messages Twitch posts in chat.
YES, IT DOESN’T MATCH TWITCH. MOUSE OVER ANY ROW FOR THE DONO AUDIT. Twitch’s Monthly Top Cheerers counts every cheer in August and resets on the 1st; this board counts the subathon only, from 7:17 PM ET on Aug 27. Anyone who cheered earlier in the month reads higher there than here.
Who has cheered the most bits. Every cheer counts, even 1-bit cheers that look like emote spam.
Counted from the cheer amount in each message.
Who has added the most time to the clock with song and sound requests. Every dollar adds 10 seconds, the same as 100 bits.
From the StreamElements request history.
Subs, resubs, gift subs and raids, counted from the messages Twitch posts in chat.
Counts of announcements. A single gift bomb is one event carrying many subs.
This puts all kinds of giving on one scale, counted in gift subs. Bits and mediashare are converted by what they cost. Each bar is ten times bigger than the one before, because some people gave less than one sub and some gave thousands.
Each bar is a number of people. Everyone who gave is counted, even people too small for the leaderboard. Hover over a bar to see the number.
How much of all the giving came from the biggest givers. If everyone gave the same, the curve would follow the straight diagonal line. The higher the curve bulges above it, the more of the total came from a few big givers.
Like the chart in the December report. To read it, find 20% at the bottom, go up to the curve, and look across. That’s how much of all the giving came from the top 20% of givers.
How many distinct people have contributed at least once, cumulatively. A flattening curve means the same people are giving again rather than new ones arriving.
How many minutes the clock gained or lost in each hour of the day, in Eastern time. It counts everything: subs, bits, donations and any time added by hand.
Chat messages per minute, ZChum’s and Omid’s chats together. Spikes usually mean something happened on stream.
Who has sent the most messages. Bots are left out, so the percentages are out of messages from real people.
How the conversation splits between the two chats being recorded.
Every time the timer was paused since the subathon started. The clock doesn’t count down while it’s paused.
Loading…
We check the subathon clock every 10 seconds. This shows how often that worked. A gap means we missed a check.
How far our clock is off from the timer’s clock, in thousandths of a second. All the charts use the timer’s clock, so this doesn’t change anything.
Corporate Captivity: Overgrown
ZChum × OmidLive. This desktop is an unofficial stats page for the subathon,
built and run by a viewer.
Where the numbers come from.
The clock and the subscriber count are read from ZChum’s own timer, sampled every
10 seconds. Chat is recorded live from both Twitch channels. Viewer counts, the
current game and the stream title come from Twitch’s public API once a minute.
If a chart has no data yet, it stays empty.
What is stored.
Public chat messages and the public stream data above. The page shows totals from them, like counts, rates and leaderboards.
Archive 2025.
The icon in the corner opens the previous subathon: 583,745 messages from 7,102 chatters
between 29 October and 9 December 2025, reduced from 1.3 GB of logs.
Subscriptions are missing from it because Twitch’s VOD chat export omits them.
Games.
Scores are saved in your own browser only. There is no shared leaderboard and nothing
is sent anywhere.
Credits.
Art by DevilsCoast. Site by DrJellybear.
The Streamer Awards: nominations are open
ZChum would love a nomination for Best Roleplay and Best Marathon Stream, and any other category you think fits.
Nominate ZChum at thestreamerawards.com
Trout is here and on stream tonight at 10pm ET.
Loading the goals…