The thing that got me interested in regional premiums is that they should not exist. In a market with free capital flow, if Bitcoin trades higher on a Korean exchange than on an offshore one, someone buys offshore, sells in Korea, pockets the spread, and the two prices converge within minutes. That is the whole mechanism that keeps prices honest across venues. The reason the Korean premium can sit positive for weeks, sometimes climbing to double digit percentages, is that the arbitrage is blocked. And when arbitrage is blocked, price stops being a clean consensus number and starts carrying information about who is buying where. That is the part worth paying attention to.
What each premium actually measures
The kimchi premium is the gap between the price of a coin on Korean won exchanges and its price on offshore dollar exchanges, expressed as a percentage. It exists because getting money in and out of Korea to trade crypto is genuinely hard. Korean banking rules make it difficult to wire large sums to foreign exchanges for the purpose of buying crypto, and there is no easy way to buy in Korea, move the coin out, and sell abroad at scale without tripping compliance or eating fees and delays that swallow the spread. So Korean retail demand gets trapped inside a walled market. When that demand runs hot, won buyers bid the local price up and nothing offshore comes in to cap it. The premium is basically a pressure gauge on isolated Korean retail appetite.
The Coinbase premium is a different animal that people often lump into the same bucket. It is the gap between the price on Coinbase, which skews toward US based and institutional flow, and the price on a large offshore exchange like Binance. The friction here is not capital controls, it is who trades where and in what size. US institutions and funds that need a regulated, dollar native, compliant venue tend to route through Coinbase. When that flow is aggressive, Coinbase trades at a small premium to the offshore reference. When it goes quiet or turns to selling, Coinbase slips to a discount. The gap is usually tiny, a fraction of a percent, which is exactly why it is useful as a directional read rather than a tradable spread.
So one premium tells you something about isolated retail, the other about a specific slice of institutional demand. Neither is a price prediction. Both are structure.
How to track them without overthinking it
You do not need anything fancy. Both premiums are just a ratio between two price feeds. The workflow I use is roughly this:
- Pick a clean pair for each. For the kimchi premium, a major won exchange against a major dollar exchange, both quoting the same coin. For the Coinbase premium, Coinbase USD against a large offshore venue.
- Convert to a common currency so you are comparing like with like. The kimchi premium needs the won to dollar rate baked in, otherwise you are just measuring the exchange rate moving, not real demand.
- Express the gap as a percentage and watch the direction and the rate of change, not the absolute number. A premium going from small positive to large positive matters more than whether it is two percent or three on any given read.
- Look at it in context with price. A premium widening while price rises tells a very different story than a premium widening while price is flat or falling.
Plenty of dashboards publish both already, so most people never compute them by hand. The value is in checking them habitually, not in the arithmetic. On Blockcircle we treat these as two of the standing structure indicators alongside the whale and disclosure feeds, because they answer a question raw price cannot, which is who is doing the buying.
What widening and flipping have historically signaled
The classic read on a large positive kimchi premium is that Korean retail is euphoric and buying with both hands while the rest of the market is calmer. Historically, a premium that blows out to an extreme level has tended to show up near local tops rather than near the start of a move. It is a crowd indicator, and crowds are usually loudest late. That is not a timing signal you can trade mechanically, but a kimchi premium screaming near a high is a reason to check your own conviction rather than add to it.
A negative kimchi premium, a discount, is the rarer and in some ways more interesting reading. It usually means Korean sellers are trying to get out and the trapped nature of the market now works against them, or that offshore demand has run far ahead of local appetite. Persistent discounts have historically clustered around fear rather than greed.
The Coinbase premium works as a near real time proxy for whether the US institutional side is leaning in or out. A Coinbase price sitting steadily above the offshore reference during a rally is often read as confirmation that the buying has a US institutional component rather than being purely offshore leverage. When a rally is running but Coinbase is trading at a discount, the honest interpretation is that the move may be thinner than it looks, driven by offshore flow that can reverse fast. I have watched moves fizzle precisely because the Coinbase premium never confirmed them.
The caveats that actually bite
Here is where people get burned. The first mistake is treating the kimchi premium as a global signal when it is a local one. It measures Korean retail, full stop. Reading it as a verdict on the whole market is how you end up short into strength or long into a top because one isolated crowd was excited.
The second is forgetting the exchange rate. A chunk of what looks like a moving kimchi premium can just be the won to dollar rate shifting. If you are not currency adjusting properly, you are watching forex noise and calling it sentiment.
The third is confusing correlation with a trigger. These premiums are context, not entries. A premium can stay extreme far longer than you can stay solvent betting against it, and it can sit near zero through a violent move that had nothing to do with regional flow. Use them to shade your read, to raise or lower conviction on a thesis you already have. Do not build a position on the premium alone.
The last one is stale plumbing. If one of your two feeds lags, delists the coin, or has a withdrawal issue, the premium you are reading is an artifact of a broken pipe, not real demand. Whenever a premium does something genuinely strange, check that both venues are trading and settling normally before you believe the number.
Used carefully, the two of them give you a cheap, always on read on the two groups that raw price hides. Korean retail on one side, US institutional flow on the other. Neither will tell you what to do. Both will tell you who is in the room, and that is usually enough to keep you from mistaking one crowd's excitement for the whole market's.