Market Risk

Market risk is the foundation for understanding how AI is used in finance—from forecasting portfolio losses to reading earnings calls. Once you grasp how price movements create losses, you'll be ready to explore how AI models like FinBERT gauge sentiment, how backtesting validates strategies, and how point-in-time data prevents lookahead bias. This concept unlocks the entire 'AI in Finance' branch of your learning library.

Imagine you own a small shop that buys spices wholesale and sells them retail. Your profit depends on the price you pay for spices. If wholesale prices jump, your costs rise; if they fall, you can either make more profit or hold stock that has lost value. That uncertainty—will prices move against you?—is market risk. It's the risk that the things you own (or owe) change in value because of movements in prices, interest rates, currency exchange rates, or how much prices bounce around (volatility). In finance, market risk is not about good or bad—it's about the fact that prices move, and movements can cause losses. A trader with a portfolio of stocks faces market risk every day: if the market falls, the portfolio falls. A bank that lent money in dollars but borrows in rupees faces currency risk: if the rupee strengthens, the loan becomes more expensive to repay. Even a company that buys raw materials in one currency and sells products in another carries market risk. AI enters because we can use models to estimate how much money might be lost—not by predicting the future, but by measuring how much the value of a portfolio could swing.

Think of a ship sailing through a storm. The ship's captain knows that waves will lift and drop the vessel, but the captain can't predict each wave. What the captain can do is measure the sea's 'roughness'—how high waves typically get, how big the biggest waves can be, and how often they occur. That measurement helps decide whether to sail or dock, how much cargo to load, and how many lifeboats to prepare. Market risk is the sea: prices are the waves, and a portfolio is the ship. Value at Risk (VaR) is like a measure of 'how high will the biggest wave be 95% of the time?'—it gives a number: 'We expect that 95% of days, the portfolio will lose no more than ₹10 lakh.' Expected shortfall goes further: 'On the worst 5% of days, the average loss will be ₹25 lakh.' But the analogy breaks down: the ocean's waves follow physical laws we can measure perfectly; financial markets are driven by human behaviour, which can change abruptly. A model fitted to calm seas won't prepare you for a tsunami. That's the limitation the notes warn about: tail events—extreme losses—are systematically underweighted by models built from ordinary times. The ship's captain has a map of the sea; the risk manager only has the past, which may not repeat.

Market risk is the potential for loss due to adverse movements in market prices, interest rates, exchange rates, commodity prices, or volatility—the degree of price fluctuation.

You haven't studied other topics yet, so this is your first. But think of it as the entry point to 'AI in Finance'. Later, you'll see how market risk connects to backtesting (does a strategy that worked in the past still work?—but beware of lookahead bias), to earnings call analysis (when a company's tone becomes evasive, volatility often rises, which is market risk), and to the specialized language models like FinBERT and BloombergGPT that process financial text. In your library's vocabulary, market risk is the baseline: every trading decision, every portfolio, every AI-driven signal is ultimately about managing this risk. Even the quote from 'Large Language Models in Finance'—that volatility is not simply bad—reflects that market risk has two sides: losing money when prices move against you, and making money when they move for you.