Hey, welcome to VoraPrep Audio. If you're studying for the FRM exam, you know that some topics feel more abstract than others. Probability can definitely feel that way. But on exam day, and in your actual job, it’s anything but abstract.
Imagine you're a risk analyst staring at a new structured product. Its payoff depends on a bunch of different things happening—or not happening. Maybe a default event occurs. Or maybe there's a big interest rate hike. You have to figure out the chance of those things happening together. That's how you price the risk correctly. That’s not abstract. That's money. And that’s what we’re talking about today.
Let's start with the basics: probability. It's just a number, from zero to one. Zero means impossible. One means it's a sure thing. Everything else is somewhere in between. The set of all possible outcomes is called the sample space. An event is just one of those outcomes, or a group of them.
Now, the exam loves to test how you combine probabilities. The two big rules are the addition rule and the multiplication rule.
The addition rule is for "OR" questions. What’s the probability of A or B happening? You add their individual probabilities. But then you have to subtract the part where they overlap—that’s the chance they both happen at the same time. If they can’t possibly happen together, like a coin flip being heads and tails at the same time, we call them mutually exclusive. In that case, you just add them up. Simple as that.
The multiplication rule is for "AND" questions. What's the probability of A and B happening? This depends on whether the events are independent. If they are, like two separate coin flips, you just multiply their probabilities. Easy. But what if they aren't? What if event A happening changes the odds of event B happening?
That's where conditional probability comes in. The formula is written as P of A, given B. This just means "the probability of A, given that B already happened." So the full multiplication rule is: the probability of A and B is the probability of B, times the probability of A given B.
Next, let's talk about random variables. It's a fancy name for a simple idea. It’s just a process that spits out a number. And we can't be sure which number it'll be. There are two main types: discrete and continuous.
Think about the number of defaults in your loan portfolio over the next year. You could have zero defaults, one, two… a specific, countable number. You can't have one-point-seven-three defaults. That's a discrete random variable. We use something called a Probability Mass Function, or PMF, to find the probability of getting exactly a certain number.
Now think about tomorrow's interest rate. It could be four-point-five percent. Or four-point-five-one. Or four-point-five-one-two. It can take on any value within a range. It’s not a set of distinct, countable steps. That’s a continuous random variable. For these, we use a Probability Density Function, or PDF. And here’s a key trap to watch for: with a continuous variable, the probability of it being exactly one specific number is zero. It’s just too precise. We can only talk about the probability of it falling within a certain range.
For any random variable, we want to know two key things: its expected value and its variance. The expected value is the average outcome you’d expect if you ran the experiment a million times. It's a weighted average of all possible values, weighted by their probabilities. The variance tells you how spread out the results are around that average. A low variance means the outcomes are all clustered tightly around the expected value. High variance means they’re all over the place. And the square root of that variance… that's your standard deviation. A term you've probably heard a thousand times.
Okay, so what happens when you have two random variables, say, the returns of two different stocks? We need to know how they move together. That's where covariance and correlation come in. Covariance tells you the direction of the relationship. If it’s positive, they tend to move in the same direction. If it’s negative, they move opposite. But the number itself is hard to interpret. So we standardize it, and that gives us correlation. Correlation is always between negative one and positive one. Positive one means they move in perfect lockstep. Negative one means they are perfect opposites. Zero means no linear relationship. This is huge for portfolio risk.
Now for Bayes' Theorem. This sounds intimidating, but the idea is actually very intuitive. It’s a formal way to update your beliefs when you get new information.
Think of it like a doctor’s diagnosis. Let's say there's a rare disease. Your initial belief—the prior probability—that a random person has it is very low. Say, one percent. That's P of A. Now, the person takes a test, and it comes back positive. That's the new information, our event B. The test isn't perfect, but it's pretty good. We know the probability of the test being positive if you have the disease. That’s the likelihood, or P of B, given A. What Bayes' Theorem lets us do is flip that around. We can calculate the probability you actually have the disease, given the positive test result. That's the posterior probability, P of A, given B. We started with a belief, got new evidence, and ended up with a new, updated belief. That’s all Bayes' Theorem is. It's a recipe for learning from experience.
Let’s try a real exam-style problem. A bank is looking at a loan. The initial, or prior, probability of default is five percent. Let's call default 'D', so P of D equals zero-point-zero-five. The probability of the economy going into a recession, 'R', is ten percent. So P of R equals zero-point-one-zero. And we're told that if a recession happens, the probability of default jumps to thirty percent. That's a conditional probability: P of D, given R, equals zero-point-three-zero.
This is a classic Bayes' Theorem setup. The formula is: P of R, given D, equals P of D, given R, times P of R, all divided by P of D. We have all the pieces.
P of D, given R, is zero-point-three-zero. P of R is zero-point-one-zero. And P of D is zero-point-zero-five.
So in the numerator, we have zero-point-three-zero times zero-point-one-zero, which is zero-point-zero-three. In the denominator, we have zero-point-zero-five. So, zero-point-zero-three divided by zero-point-zero-five gives us zero-point-six-zero. Sixty percent.
So, if we see that company default, there's a sixty percent chance we're in a recession. Notice how that posterior probability, sixty percent, is way higher than the prior probability of a recession, which was only ten percent. We got new information—the default—and it drastically updated our belief. Also, notice the trap here. The question asks for P of R, given D, but it gives you P of D, given R. It's so easy to mix those up under pressure. Always write down what you're given and what you need to find.
Let's do one more, this time with a portfolio. You have two assets, A and B, weighted fifty-fifty. Asset A has an expected return of ten percent and a standard deviation of fifteen percent. Asset B has an expected return of twelve percent and a standard deviation of twenty percent. The correlation between them is zero-point-five. What's the portfolio's expected return and variance?
First, the expected return. This one's straightforward. It's just the weighted average of the individual expected returns. So, fifty percent times ten percent, plus fifty percent times twelve percent. That’s five percent plus six percent, which gives us eleven percent.
Now, for the variance. The formula is a bit longer, but it's just a recipe. You take the first weight squared, times the first variance. Add the second weight squared, times the second variance. Then add two, times the first weight, times the second weight, times the covariance between the two assets.
Let's plug in the numbers. First, we need the variances, which are the standard deviations squared. For Asset A, variance is zero-point-one-five squared, which is zero-point-zero-two-two-five. For Asset B, it’s zero-point-two-zero squared, which is zero-point-zero-four.
Next, we need the covariance. The formula is correlation, times the standard deviation of A, times the standard deviation of B. So, zero-point-five times zero-point-one-five times zero-point-two-zero, which gives us zero-point-zero-one-five.
Now we put it all into the big portfolio variance formula. First, we have zero-point-five squared, times zero-point-zero-two-two-five. Then, we add zero-point-five squared, times zero-point-zero-four. Finally, we add two, times zero-point-five, times zero-point-five, times zero-point-zero-one-five.
Working that through, you get zero-point-zero-zero-five-six-two-five, plus zero-point-zero-one, plus zero-point-zero-zero-seven-five. Add those all up, and you get a portfolio variance of zero-point-zero-two-three-one-two-five. So the answer is an expected return of eleven percent and a variance of zero-point-zero-two-three-one-two-five.
So, let's recap the key points. First, know your basic probability rules for "OR" and "AND" events. Second, really understand the difference between discrete and continuous random variables. Third, be able to calculate expected value and variance. And finally, for Bayes' Theorem, just remember it’s a tool for updating your beliefs. You start with a prior, get new evidence, and then calculate a posterior.
You're building a mental toolkit here. Each of these concepts is a tool, and with practice, you'll know exactly which one to pull out for any given problem. Keep up the great work, and I'll talk to you in the next session.
VoraPrep is a full exam-prep app — lessons, practice questions, and an AI tutor in one place. Get started at Vora Prep dot com.