The Disposition Effect in Crypto: What the 2017 Bitcoin Boom-Bust Revealed

It’s December 2017. Bitcoin is up double digits in a week, and a smaller-cap position you bought on a whim in the spring — the one you’d almost forgotten about — is suddenly worth three times what you paid. You sell it. It feels good. It feels like a decision.

Meanwhile, another position sits in the same wallet, down 40% since summer. The story you told yourself when you bought it hasn’t come true, but it hasn’t been disproven either, not exactly. You hold. Then Bitcoin turns in January, and by March that losing position is down 70%. You’re still holding it in June.

If this sequence sounds familiar, it should. It’s one of the most reliably observed footprints in trading data, in any asset class where prices move and people hold positions long enough to have feelings about them: sell the position that’s up, keep the position that’s down. It has a name — the disposition effect — and it’s one of the few behavioral patterns that’s been specifically documented inside crypto markets, not just borrowed from equities research and assumed to translate.

A pattern that doesn’t need a ticker tape to show up

The disposition effect describes an asymmetry in exit timing: positions that have appreciated get closed sooner, relative to some reference point, than positions that have declined. It was first mapped in stock brokerage data, where tax incentives, dividend timing, and analyst coverage could be pointed to as possible explanations layered on top of the psychology.

Crypto strips a lot of that away. There’s no dividend to wait for. Until relatively recently, tax-loss harvesting wasn’t a meaningful factor for most retail holders in most jurisdictions. There’s no earnings calendar, no analyst price target, no fundamentals in the traditional sense to anchor a “fair value.” A coin is worth what the last trade said it was worth, twenty-four hours a day, seven days a week, with no closing bell to create a natural pause.

Given all of that, you might expect the disposition effect to be weaker in crypto — a pattern that needs institutional and tax scaffolding to survive, and collapses without it.

That’s not what the research on the 2017 boom-bust found. Bitcoin disposition-effect studies covering that cycle observed the pattern intensifying, not fading, as the market moved from euphoria into collapse. If anything, the absence of the usual structural explanations makes the finding more interesting, not less: whatever is driving the asymmetry, it isn’t primarily about tax timing or dividend calendars. It’s closer to the underlying psychology itself, showing up with less interference.

Why the boom-bust shape matters

The 2017 cycle is a useful natural experiment because it has two distinct phases with opposite emotional textures, and the disposition effect showed up differently in each.

During the boom — prices climbing, sentiment euphoric, new entrants arriving daily — the pattern looked like impatience with winners. Positions that had appreciated were closed relatively quickly, almost as if the upside move itself created pressure to lock it in before it evaporated. This tracks with a broader idea in behavioral finance: an upward move feels less certain than it looks, and the discomfort of a possible reversal outweighs the appeal of letting it run.

During the bust — prices falling fast, sentiment shifting from euphoria to denial to fatigue — the pattern flipped into something that looked more like reluctance with losers. Positions that had declined were held longer, sometimes much longer, than positions of similar size that had declined less. The reference point traders seemed to be measuring against wasn’t the current price; it was the price they remembered from the top.

Put those two halves together and you get the shape the studies describe: a market where winning positions get sold into strength and losing positions get carried through weakness, at exactly the moments when that combination compounds the asymmetry rather than correcting it.

The 24/7 market doesn’t cancel the pattern — it just removes the excuses

One of the more counterintuitive things about the crypto findings is how they complicate a common assumption: that a market without institutional structure, without market-open rituals, without a “professional class” gatekeeping order flow, would behave more randomly, more rationally, or at least differently from equities.

It didn’t. If the disposition effect were mostly a byproduct of institutional plumbing — brokerage tax defaults, portfolio manager mandates, quarterly reporting — you’d expect crypto, with almost none of that plumbing during the 2017 cycle, to show a muted version of the pattern. Instead, the asymmetry appeared to intensify. That’s a meaningful data point for anyone studying bitcoin investor psychology specifically, and crypto trading biases more broadly: the pattern doesn’t need a stock exchange to exist. It needs a person, a reference point, and a decision about when to exit.

That’s worth sitting with, because it reframes the disposition effect as less of a market-structure artifact and more of a default setting — something closer to how people process upside and downside moves regardless of the wrapper the asset comes in. Crypto, precisely because it lacks so many of the traditional scaffolds, ends up being one of the cleaner places to observe it.

What this looks like in a real trade history

None of this requires introspection or self-diagnosis to see. It shows up as a distribution — a skew in how long winning and losing positions are held before being closed, measured in hours or days, across enough trades to matter. A single trade doesn’t tell you anything; the shape across dozens or hundreds does.

That’s the layer most journals and P&L trackers don’t reach. A log of entries and exits tells you what happened. It doesn’t tell you whether your holding-time distribution for appreciating positions looks structurally different from your holding-time distribution for declining ones — which is the actual signature of the disposition effect, and the thing the 2017 Bitcoin research was measuring in the first place.

The behavior is already in the data. Most of us just haven’t looked at it in a shape that would let us recognize it.