A CDS is a bilateral contract between two counterparties. The protection buyer is buying insurance: he/she pays premiums in exchange for a payoff in case there is a CREDIT EVENT (a trigger)
A balance sheet CDO transfers credit risk from the bank (originator) to investors. A key aspect of a CDO is that investors have different (tranched) securities.
OLS minimizes the residual sum of squares (RSS). RSS is the sum of each squared residual (residual = the observed Y minus the predicted "on the line" Y). Also, about the OLS: the average residual is always zero, and the line passes through the point (average X, average Y)
The small sample is a 10-day series of Google's daily periodic returns. The question is, with 95% confidence, what is the true (population) average return? This is the essence of statistics, based on sample statistics (sample mean, sample variance) we are trying to infer population parameters (population mean).
Covariance is a measure of relationship (or co-movement) between two variables. Correlation is just the translation of covariance into a UNITLESS measure that we can understand (-1.0 to 1.0)
The binomial is one of the basic distributions, yet surprisingly common in risk and quant finance. Here I take a look at its key properties and compare the formula to Excel's built in =BINOMDIST()
In a linear regression, you often see the R-squared quoted. To explain the R-squared (coefficient of determination), I compare it to the standard error of estimate (a measure of the line's accuracy) and the correlation (the square root of the coefficient of determination). All three, loosely speaking, are measures of the line's fit to the data
Both count the ways that (r) objects can be taken from a group of (n) objects, but permutations are arrangements (sequence matters), while combinations are selections (order does not matter). For example, how many ways can you seat people at a table? That's permutation. How many poker hands are available in five-card draw? That's a combination
A linear regression gives us a best-fit line for a scatterplot of data. The standard error of estimate (SEE) is one of the metrics that tells us about the fit of the line to the data. The SEE is the standard deviation of the errors (or residuals)
I illustrate the confidence interval construction with an example: the P/E ratio of 28 companies. The point is to say with confidence (e.g., 95%) that the "true" population lies within an interval.
The key difference between a cash and synthetic CDO is: instead of selling the reference portfolio (loans), the originator (bank) purchases credit protection with credit default swaps (CDS)
![Student\](http://img.youtube.com/vi/pqtG1vXg_f8/2.jpg "Student\")
