How the market confidence index is defined
The index on this site is scaled so that SK Hynix's all-time-high day equals 100, and combines price momentum, premium structure, trading volume and market expectations. This page explains each component and why the scale has no ceiling.
Why the all-time-high day is the 100 mark
Sentiment gauges are usually pinned to a 0-to-100 range. That reads easily, but the moment the underlying breaks its historical extreme the gauge sticks at 100 and stops discriminating — precisely when discrimination matters most. This site uses a relative baseline instead: the state of the Korean line on its highest day is defined as 100, and every subsequent day is measured against it. Breaking to new highs can therefore push the index above 100, and a deep drawdown can take it well below zero, so resolution is never lost.
The four components
Price momentum looks at where the Korean line sits relative to its own moving-average structure and how steeply it has been moving, answering how strong the trend is. Premium structure reads the ADR premium and the 7709 gap, answering whether offshore and Hong Kong money is more aggressive than domestic money. Trading volume compares the day's turnover with the 20-day average, answering how many people are involved. Market expectations come from related contracts on an external prediction market; when that feed is unavailable the component degrades to a derivatives-sentiment proxy, and the interface labels which source was actually used.
Smoothing and update cadence
The weighted sum is smoothed against the previous day so that single-day noise does not turn the line into a saw edge. It is recomputed every ten minutes while any market is open, and no new point is written when every market is shut and it is a weekend — otherwise the series would contain dates with no matching daily bar, which reads as missing data. The detail page shows the last 30 days along with each component's contribution for the current day.
What it cannot answer
This is a composite description of the present state, not a forecasting model, and it has not been backtested for timing value. The component weights are a judgement call, and different weights produce a differently shaped curve. Using it to ask whether sentiment is hotter or colder than last week is reasonable. Reading it as a buy or sell signal goes beyond what it was built to do.
This page documents how this site computes the figure. It is not investment advice.