Michael Sullivan built cohort-level sentiment analysis of Bitcoin Twitter and the data is unambiguous: pleb anger is at its longest persistent streak in years, OG conviction is diverging sharply upward, and the narratives that matter most are being completely ignored.
↓ Jump to the video and timestamps
I've been saying for months that 2026 feels more like 2015 than any other bear market I've lived through. The sentiment is terrible. People genuinely think Bitcoin is dying. And I've never been more bullish.
Michael Sullivan came on the show this week with actual data to back that read up. He's spent hundreds of hours building cohort-level sentiment analysis of Bitcoin Twitter, not the aggregated Reddit-scraping nonsense that powers most fear and greed indexes, but individualized account-level analysis that tracks how real people's language evolves over time. What he's found confirms everything my pattern recognition has been signaling: pleb anger is at its longest persistent streak in recent memory, OG conviction is diverging sharply upward, and the bullish narratives that would have trended for days in a euphoric market are getting zero engagement. That combination, historically, is what bottoms look like.
I've been through five of these. I top-ticked my first buy in late 2013, rode it all the way down through the Mt. Gox collapse, and watched the price wick to around $180 in what I believe was July 2015. Bitcoin Twitter barely existed. The people who were around mostly went quiet. It felt like it was over. It wasn't. This feels exactly like that.
I got into Bitcoin in late 2013. My first buy was basically the top tick right before Mt. Gox collapsed. I rode that thing down over the following 18 months to what I believe was around $180. I think it wicked down there sometime around July 2015.
The sentiment at the time was genuinely awful. Bitcoin Twitter barely existed, and the small pocket that did exist mostly went quiet. People thought it was legitimately over.
Then we started climbing back in late 2016. 2017 happened. And here we are.
The reason I keep anchoring to 2015 right now is that the emotional texture is the same. Not the specifics. The FTX era had its own character, and the 2018 bear market was more boring than angry. But 2026 has that particular flavor of people being genuinely, persistently furious and convinced the thesis is broken. That's what 2015 felt like.
Sullivan's data confirms it isn't vibes. The OG cohort in his system (accounts that have been in Bitcoin for ten or more years and posting consistently) is showing rising conviction. Not wild euphoria, just a steady divergence upward from where the newer cohort sits. The people who've been through this before are not bailing. They're doing the opposite.
I'll say it straight: I've never been more bullish. That's not a hedge. That's my read after 13 years and five bear markets.
Most sentiment indexes, including the fear and greed indexes that get passed around constantly, pull from a relatively small pool of top posts or aggregate Reddit activity. The problem is obvious once you think about it: you're capturing engagement bait, bots, and whatever the algorithm decided to surface that day. The signal gets polluted fast.
Sullivan built something different. He went and individually identified real accounts (anonymous or not, it doesn't matter as much as they're verifiably human and posting consistently over time) and then ran their entire tweet history through AI models trained to detect emotional language patterns. Every emotion has its own vocabulary fingerprint. Hedging language like "would" or "maybe" reads as lower conviction. Declarative, exclamation-heavy language reads as higher conviction. Anger, optimism, fear: all of them have associated word patterns that trained models can detect at scale. The key move is comparing each person's current language against their own historical baseline, then cohorting people together to smooth out individual noise.
Someone going quiet for a month because they had a kid doesn't tank the whole OG cohort reading. But when a large group of experienced Bitcoiners starts moving in the same emotional direction simultaneously, that's a signal worth paying attention to.
He also draws a sharp distinction from the way fear and greed indexes work. Sullivan's characterization of those tools is that they lean on top posts and Reddit aggregation, an approach that disproportionately captures engagement-optimized content and bot-polluted feeds. His individualized approach sidesteps a lot of that noise by watching how real people's language actually evolves, regardless of what the algorithm chose to amplify that day.
Sullivan's output is a multi-dimensional picture: conviction, optimism, anger, and fear tracked separately, by cohort, over time. Conviction is the one I find most interesting right now, because it's non-directional on its own. High conviction during euphoria means people are certain good things will happen. High conviction during a drawdown means people are certain bad things will happen.
What makes the current OG picture notable is that their conviction is rising while their anger stays relatively contained, which reads like experienced buyers growing more confident in their thesis, not more fearful about the outcome.
The chart Sullivan shared during the conversation is the one I keep thinking about. Pleb conviction has fallen off dramatically. OG conviction has done the opposite. It's risen through the same period, creating one of the larger divergences Sullivan has observed.
His explanation for why this happens maps exactly to what I'd say from lived experience. Conviction drops when your story disintegrates. If you came into Bitcoin around April 2021, your mental price anchor is roughly $60K. You look up five years later, price is in the same neighborhood, and if you never DCA'd, you feel like you got nowhere. The story you told yourself (that this was the asset that goes up forever, that the ETFs and the strategic reserve and the government validation changed the game permanently) has been challenged. So you start hedging your language. You start looking for explanations. You get angry.
Hand up: I was one of the people saying on the podcast that I could squint and see a supercycle coming. The government is on our side, the ETFs are here, nation-states are buying. Maybe the four-year cycle is done. I was wrong, and I've been around long enough to know when to say that and move on.
Bitcoin does what it does. It beats to its own drum. The four-year cycle appears to be intact, and the sooner you internalize that instead of projecting blame onto Saylor or whoever else is handy, the better your decisions get.
The OGs who are growing more convicted right now are the ones who've made that peace. They're not looking for a villain. They're looking at debt levels, interest expense, central bank behavior, the weaponization of the financial system, and they're saying, yeah, all of this still points at the same place it always pointed. The conviction rises because the thesis gets stronger every day, even when the price doesn't cooperate.
Sullivan collaborated with James Check on a piece called "Peak Apathy" a few weeks back, exploring how flat and disengaged sentiment had become. After the most recent sell-off, Sullivan ran his full pipeline again to see how moods had shifted. The divergence that was beginning had continued to grow (across anger levels, optimism, and conviction) between the OGs and the plebs. That's the data behind the gut read.
Sullivan has a framework for what anger does to decision-making in markets, and it's worth sitting with. His analogy: someone cuts you off in traffic and within seconds you've invented a whole character for them (their whole deal, why they're a bad person, what their life is probably like). Anger manufactures villains. It's not a deliberate process; it's just what the emotional state does.
Bitcoin Twitter right now is doing the same thing. Someone posts that Saylor sold a small number of bitcoin and the ratio is immediate. People desperately, desperately wanted somebody to blame for negative price action. Whether that specific sale was even meaningful is beside the point. The crowd needed a villain and Saylor was available.
The STRC situation over the recent weekend is another example. The instrument dropped and briefly traded significantly below par, and within hours Twitter had it as the next Terra Luna, a blow-up in progress, a systemic threat. Then it reclaimed near par. The catastrophe narrative evaporated almost immediately.
Sullivan's data shows that Stretch has actually been one of the more durable narratives he's tracked, growing consistently since its launch around the end of July, with the volatility around it being noise on top of a persistent uptrend in mentions. The villain story lasted a weekend. The underlying narrative has been building for months.
When you see that pattern (anger seeking a target, catastrophe narratives arising and collapsing quickly, ratio'd posts about negligible events), that's the tell. That's what bottoms have looked like every time I've been through one. That pattern is diagnostic of people being at an emotional extreme, not of something fundamentally wrong with Bitcoin. Emotional extremes end.
This is the part that I find most practically useful about Sullivan's work. He tracks not just how people feel but what they're talking about, and the divergence between what's actually happening and what's getting attention is striking right now.
The strategic bitcoin reserve is the clearest example. At the end of 2024, when price was ripping, SBR talk was everywhere. People were treating it like a guaranteed perpetual nation-state buying program and pricing in unlimited euphoria.
Now Scott Bessent has explicitly confirmed that the administration is working on the strategic bitcoin reserve, that it will take time, that they want to do it right. Active, on-record government commitment to the thing. Sullivan's mention rate chart for SBR is basically flat.
We posted about the Bessent clip ourselves. It got 184,000 impressions. 478 likes. 61 retweets. That's the engagement you'd expect on a mid-tier observation, not on a sitting Treasury Secretary explicitly reaffirming a bitcoin reserve policy. The same clip in December 2024 would have gotten 10,000 likes.
The news didn't change. The emotional context around it did. People are underweighting genuinely bullish information because they're angry, and angry people don't want to hear bullish stories.
The CLARITY Act is another one Sullivan flags as underweighted. It's getting talked about more, it represents real positive regulatory movement, and it's generating almost no excitement in the current mood environment. On the other side, the quantum threat narrative has been elevated, exactly the kind of fear-amplified story that gets disproportionate attention when sentiment is poor. The paper bitcoin narrative tracked similarly: it spiked alongside anger as people looked for something to be scared of or blame. Sullivan is careful to note that doesn't mean there's nothing there, but the timing of narrative adoption matters. Stories get latched onto based on emotional environment at least as much as based on truth.
Saylor recently wrote a piece categorizing different ideological groups within Bitcoin (the capitalists, the technologists, the fundamentalists, among others). Sullivan has already started tracking those cohorts separately. The capitalist cohort is actually highly convicted right now, talking heavily about instruments like STRC.
The fundamentalists are showing much worse sentiment across the board. These groups aren't even interacting that much, partly because of algorithmic siloing. They're living in different emotional realities, driven by the different narratives their feeds are serving them.
I admitted the supercycle thesis above. Here's the fuller version: when Trump ran as the Bitcoin president, when the ETFs got approved, when the strategic reserve narrative picked up steam, I could squint and see a world where the four-year cycle got disrupted by a completely new class of buyer. Nation-states accumulating. Institutional vehicles running 24/7. A US government that, for the first time, was explicitly supportive.
I was wrong. The four-year cycle has held. Someone I follow on X, whose handle I'd rather not misattribute here given how names can get garbled in transcription, had been saying publicly throughout last year that the cycle was intact, that we'd drift lower, and that the bottom would come in its own time. That read has proven out.
Sullivan's point about the four-year cycle narrative is one of the more interesting analytical frames in the whole conversation. So many people are now locked into an October bottom thesis that it's become its own source of fragility. When that many people are highly convicted about a specific story (that the bottom happens on a particular timeline, that the cycle works a specific way), any evidence that challenges it creates extreme volatility. Narratives don't die gracefully. They collapse, and the collapse tends to be rapid.
His read: if positive price action emerges before the "expected" bottom, the October thesis unravels fast. People who were locked into one story realize they're wrong, conviction falls, they pivot, and those pivots tend to be violent. I'm not calling a bottom here, but I am saying that the fragility of that narrative is itself something to watch.
Sullivan's closing position was unambiguous: he's buying bitcoin hard right now. I'm with him.
One thing worth pulling out from the conversation is the cold storage angle. The paper bitcoin anxiety that's been elevated alongside the anger spike has an interesting secondary effect: it's driving people back toward self-custody. People asking whether BlackRock's ETF actually holds the bitcoin, whether the treasury companies are fully custodied, whether Coinbase holds what it says it holds. Some of that is paranoia amplified by bad sentiment, but the behavior it produces is structurally healthy.
If you pull bitcoin off exchanges and into cold storage, you're tightening actual supply. Proof-of-keys day used to be a real event (January 3rd, Trace Mayer originated it) and then it faded as the culture got more comfortable with custodial exposure. The anxiety that's driving re-emergence of that impulse right now, whatever its emotional origin, produces a result I'd characterize as a feature, not a bug.
We can say with a high degree of certainty that there's blood in the streets right now. That's typically when the best buying opportunities arise. I've been through this five times. The OGs in Sullivan's system are more convicted than they've been in a while, and the plebs are angry and looking for villains.
The bullish narratives are being ignored. The four-year cycle held. I know where I stand.
Bitcoin sentiment analysis uses AI models trained on large sets of social media data to detect emotional signals in language (things like conviction, optimism, anger, and fear) by identifying the specific vocabulary patterns associated with each emotion. Sullivan's approach runs individual accounts' full tweet histories through these models, compares current language to each person's own historical baseline, and then aggregates by cohort to surface group-level emotional trends. The result is a multi-dimensional picture of how different groups of Bitcoin holders feel at any given time, rather than a single fear-or-greed number.
Sullivan's data points to a few converging factors: the AI trade and gold have captured a lot of attention and produced visible gains while Bitcoin has underperformed relative to expectations, Bitcoin treasury companies have disappointed a cohort of newer holders, and people who bought near the 2021 highs and never dollar-cost averaged are sitting at roughly the same price years later. The anger is compounded by algorithmic feeds that amplify negative content and by the basic psychology of loss, which drives villain-seeking behavior. The anger is real, but it's more diagnostic of emotional exhaustion than of a broken thesis.
Sullivan defines OGs as accounts that have been in Bitcoin for ten or more years and posting consistently. Plebs are the newer cohort, generally accounts from the 2020-2021 era forward. The current divergence is significant: pleb anger is at its longest persistent streak in Sullivan's dataset, pleb conviction has fallen sharply, while OG conviction is rising and their anger, while not absent, is comparatively contained. The OGs have seen multiple cycles and their thesis is anchored in fundamentals rather than price entry points.
The standard fear and greed indexes aggregate top posts and Reddit activity, which means they disproportionately capture engagement-optimized content and bot activity rather than the genuine sentiment of real, consistent participants. Sullivan's individualized cohort approach is designed to sidestep that problem by tracking real accounts over time and comparing their current language to their own baselines. The result surfaces things the aggregated approach misses, particularly the divergence between different cohorts, which a single-number index can't capture by design.
The more interesting observation right now is less about what the SBR means for price and more about how little attention genuinely bullish SBR news is getting. Sullivan's mention-rate data for the strategic bitcoin reserve has collapsed even as the policy has active government backing. Scott Bessent has explicitly confirmed the administration is working on it.
In a euphoric market, that clip would trend. Right now it barely registers. That gap between the news and the attention it receives is itself a signal. Bullish narratives get suppressed when sentiment is poor, which historically tends to mean the market is underpricing them.
Anger breeds villain-seeking and catastrophe narratives, which leads to overreaction to negative news and under-reaction to positive news. When a crowd is angry enough to ratio someone for a negligible bitcoin sale, or to declare a weekend price dip as the next Terra Luna, those reactions are telling you more about the emotional state of the market than about the underlying fundamentals.
Historically, that kind of sustained emotional extreme precedes recoveries because it reflects maximum pessimism, not maximum risk. The fundamentals haven't changed. The emotional context around them has, and emotional contexts revert.
The four-year cycle is the observed pattern of Bitcoin price moving through roughly four-year periods of accumulation, bull run, blow-off top, and drawdown, loosely correlated with the halving schedule. A meaningful subset of the market, including me at one point, thought the 2020-2021 institutional adoption wave might have broken the pattern by introducing a persistent new class of buyer. That thesis didn't hold. The cycle appears intact.
Sullivan's analytical point is that the narrative around the cycle is now itself a source of market fragility: if enough people are convinced the bottom comes in October and price moves before then, the narrative collapse could be violent and fast.
Michael Sullivan is an engineer and writer who has been building cohort-level Bitcoin sentiment analysis infrastructure over the past several months. He came into Bitcoin around 2021, has an engineering background of roughly 13 years, and is also known in the Bitcoin community for writing a Bitcoin novel. His sentiment analysis work is published on his Substack, searchable under his name, and he posts regularly on X under the handle @sullymichaelvin. He collaborated with James Check on a piece called "Peak Apathy" examining current Bitcoin sentiment conditions.
Michael Sullivan is an engineer with roughly 13 years of experience and a writer of nearly a decade. He entered Bitcoin around 2021 and is known in the community for writing a Bitcoin novel. Over the past several months he has built a cohort-level sentiment analysis system that runs individual Bitcoin Twitter accounts through AI models trained to detect emotional language patterns, tracking how different groups (OGs, plebs, capitalists, technologists) feel over time.
He collaborated with Glassnode analyst James Check on a piece examining peak apathy in current Bitcoin markets. He publishes his analysis on Substack and posts on X at @sullymichaelvin.
