LessonThis is a lesson restated in new words. Any figure in it is an example from the source, not this desk's record.
qfin is a quantitative-finance package installed with pip install qfin. The public piece at the time of this overview prices European calls and puts from the 1973 Black-Scholes closed form, reports the main greeks, and simulates geometric Brownian motion plus the Heston stochastic-variance model from 1993. Monte Carlo routines then price vanilla, binary, barrier, Asian, and extendable options on whichever of those two processes the inputs select.
Why it matters
The same tasks show up in separate scripts: a closed-form European price, a greek, a path simulator, an exotic payoff. A library puts them behind import paths so a notebook can call a model instead of retyping it. This overview is a map of what was already importable, and a list of what the authors said was still unfinished. Bonds, futures, and swaps were not in the release they were showing. A web scraper and natural-language sentiment models for equity research were being packaged and were described as the next release's focus.
The idea in plain words
Black-Scholes, in this package, is the closed form for a European call and a European put when the underlying follows geometric Brownian motion. You construct a call or a put, pass the five usual inputs, and read a price. From the same object the library can return delta, the sensitivity to the underlying price; gamma, the second-order sensitivity to that price; vega, the sensitivity to volatility; and theta, the sensitivity to time left. The overview does not walk through rho.
Two simulators are offered, with more described as work in progress. Geometric Brownian motion is the single-volatility process. The Heston model lets variance itself move randomly. The overview's plain contrast is that a newcomer who only knows geometric Brownian motion can think of the Heston step as adding a second source of randomness, the one inside volatility. Bates and Merton style jump or stochastic-volatility extensions are named as later work, not as importable models in this talk.
Exotic prices use those processes inside Monte Carlo. The supported payoffs named here are vanilla, binary, barrier, Asian, and extendable. Which process runs is inferred from the parameters you pass. Supply only the geometric-Brownian-motion inputs and the simulation uses that process. Supply the further Heston inputs and the simulation uses Heston. Under a parameter set that matches the closed form, the overview says a Monte Carlo vanilla price can be seen converging toward the Black-Scholes price. It does not print an error table.
How to use it
Install with pip install qfin so the dependencies come along. Import the Black-Scholes call or put, construct it with the five inputs, and read the price and the greeks you need. For a path-dependent contract, import the payoff you want from the simulations area of the package. Pass a parameter set. A short set selects geometric Brownian motion. A set that also includes the Heston parameters selects that model. Compare a vanilla Monte Carlo price with the closed form when you want a check that the paths and the formula are aimed at the same contract. The overview treats that check as a baseline before trusting the same engine on a barrier or an Asian payoff.
The unfinished sentiment tools are not part of this usage. The authors locate them in equity-alpha research and say the next version is where that work lands. Pricing code in the release they are demonstrating does not depend on them.
What usually goes wrong
Extra Heston inputs switch the path driver to Heston. Omitting them keeps geometric Brownian motion. A vanilla Monte Carlo price that will not approach the closed form is, in this overview, a parameter mismatch.
Bonds, futures, and swaps are listed as unreleased. Greeks are attached to the Black-Scholes objects. Path-dependent prices are simulated prices, and no numeric tolerance is given for the vanilla convergence. A short run is a noisy neighbor of the formula.
Questions people ask
How do you install it?
pip install qfin. The overview says the other dependencies install with it.
Which exotics can the Monte Carlo price?
Vanilla, binary, barrier, Asian, and extendable options, driven by geometric Brownian motion or by Heston, depending on the inputs.
What was not released yet?
Bond, futures, and swap pricing, plus the scraper and sentiment models aimed at equity-alpha research. Bates and Merton extensions were also still ahead.