Articles

Green Book vs Heard on the Street: Choose a Quant Interview Book for Your Target Role

Compare Green Book and Heard on the Street for quant trading, research, and developer interviews, with edition checks and an original probability exercise.

Choose between the Green Book and Heard on the Street by the interview you need to prepare for. Based on its published subject coverage, the Green Book is a reasonable starting point for organizing practice across mathematical topics. Heard on the Street is worth considering when you want a finance-interview question collection. Neither title should determine your study plan before you know whether the role emphasizes trading, research, or software development.

Those recommendations are editorial judgments, not employer endorsements. This comparison uses verified book records and publicly available publisher descriptions; it is not a claim that we purchased and tested every edition. We do not use candidate anecdotes to predict a target employer’s question set. The worked problem below is original and does not reproduce either book's questions or solutions.

A book can guide what to study; an unfamiliar question helps reveal whether you can use it.

Which Green Book are we comparing?

The quant interview Green Book here means A Practical Guide to Quantitative Finance Interviews by Xinfeng Zhou, not another publication with a similar nickname. The Google Books record identifies a 2008 Lulu edition, ISBN 9781435715752. Its publisher description lists probability, calculus, linear algebra, stochastic topics, finance, and programming.

Bibliographic fact: Open Library also records a 2020 edition with ISBN 9781735028804. A title alone does not identify the edition you will receive, so compare its ISBN before ordering. Confirm the ISBN, edition statement, format, and table of contents before buying or borrowing.

Heard on the Street means Heard on the Street: Quantitative Questions from Wall Street Job Interviews by Timothy Falcon Crack. A publisher description carried by Harvard Book Store identifies a revised 24th edition published in 2023, ISBN 9781991155467. It describes a collection of quantitative questions from finance interviews and notes that the old option-pricing appendix became a separate book.

That is an edition-specific source, not a statement that the 24th edition is the newest. Retailers can display different editions under similar titles, and inventories change. We do not give a universal current price, a ranking based on sales, or a claim that more pages necessarily mean better preparation.

What can each book help you do?

Editorial inference from the published scope: the Green Book's broad mathematical topic coverage makes it useful as a map for finding gaps. If you can solve elementary probability questions but cannot explain a derivative or a matrix calculation, a subject-based reading plan can expose the imbalance.

Heard on the Street's stated focus on finance-interview questions suggests a different use: choose a problem, attempt it without the answer, and practice defending your method. Its interview context can help structure rehearsal. It does not establish that your next interviewer will reuse a question from the collection.

We have not compared the clarity of every worked solution across editions. Use a legally available sample or library copy to judge the explanation style for yourself. Can you follow the reasoning without importing a result you have never learned? Does the answer show enough steps for you to identify where your approach failed?

If both books cover a topic, compare one accessible explanation before committing to both. Buying a second book is worthwhile only if it addresses a demonstrated gap: missing prerequisites, a different explanation you can understand, or additional unfamiliar practice. Otherwise, you may recognize an answer on the page and still freeze when an interviewer changes how the observation was selected.

Choose chapters from role evidence, not the cover

The following original role-to-reading matrix is a selection tool. Its chapter areas are topics to look for in the edition you legally access, not promises about identical chapter numbering or every employer's interview.

Target role and evidence	Areas to select first	What to practice beyond the books
Trading invitation names probability and numerical reasoning	Conditional probability, expected value, counting, and estimation problems	Spoken reasoning, arithmetic under the stated time limit, and responding to changed assumptions
Research role names statistics or model development	Probability, statistics where covered, linear algebra, and relevant mathematical foundations	Research discussion, data leakage, model checks, and coding analysis
Quant developer role names programming and systems	Algorithms or numerical methods that match the role	Runnable code, complexity, debugging, concurrency, and systems trade-offs
Derivatives role explicitly names pricing mathematics	Relevant finance, calculus, and stochastic sections	The required model assumptions, numerical implementation, and instrument-specific reasoning

Official employer example: Jane Street's trading page describes collaborative problem solving and says finance knowledge is not required for those interviews. Its interviewing overview distinguishes trading, research, technology, and machine learning. That is a useful warning against treating “quant interview” as one syllabus; it is not a rule for all trading firms.

For a software-focused invitation, advanced stochastic-calculus reading may have a lower immediate return than implementing and testing an unfamiliar problem. For a pricing role that explicitly requires those concepts, skipping them because another firm's trading guide does not require finance would be equally mistaken.

Before choosing chapters, write the next round's stated skills and one uncertainty to resolve with the recruiter. Then classify each planned reading session as directly required, a prerequisite, or optional breadth. This helps you preserve depth without spending the entire preparation window on a prestigious but unrelated chapter.

An original exercise: expectation survives dependence

Use the following original problem to test how you study. A bag contains three blue tokens and one orange token. Draw two tokens uniformly without replacement. Let X be the number of blue tokens drawn. Find E[X], then find the probability that X equals two.

Define an indicator for whether each draw is blue. Each position has marginal probability 3/4 of being blue, so the expected count is 3/4 + 3/4 = 3/2. The draws are dependent, but adding expectations does not require independence.

For two blue tokens, count the unordered pairs. There are six equally likely pairs of distinct tokens, and three contain two blue tokens. Therefore P(X = 2) = 1/2. You can also multiply 3/4 by 2/3 to obtain the same result.

Now change the rule to drawing with replacement. The expectation remains 3/2 because both draws still have blue probability 3/4. The probability of two blues becomes 9/16. The expected number of blue tokens stays the same, but the chance of two blues changes.

This is where a memorized solution can fail under follow-up. “The draws are independent” would be false in the first version, even though the expected-value arithmetic happens to be correct. State which assumption is needed for each step instead of attaching one assumption to the entire solution.

For a trading-oriented rehearsal, explain why an expected outcome alone does not determine a suitable position size or acceptable loss. For research, discuss the distribution and observation mechanism. For development, enumerate outcomes and check the calculation with runnable code. These are original extensions, not claims about either book's exercises.

Change the observation, and change the conditional probability

Keep the original draw without replacement. Someone uniformly selects one of the two drawn tokens, shows it to you, and it is blue. What is the probability that the other drawn token is also blue?

There are six equally likely unordered pairs. Three are blue-blue and three are blue-orange. The chance of seeing blue is one for a blue-blue pair and one-half for a blue-orange pair. Weighting those cases gives a posterior probability of two blues equal to 3/(3 + 3/2) = 2/3.

Now suppose you are only told that the pair contains at least one blue token. That statement is always true in this bag, because there is only one orange token and the draw has no replacement. It provides no new information. The probability of two blues remains 1/2.

The observation was selected differently, so conditioning on the word “blue” is insufficient. In a rehearsal, your partner should ask how the token was chosen before you calculate. This is a small test of whether you can recover the model from a description rather than recognize a familiar puzzle label.

Turn a solution into an explanation you can defend

Study one problem until you can explain it and handle a changed condition. First write the sample space and assumptions. Attempt a method without consulting the solution. Then compare the argument, identify the first unsupported step, and try again with a changed condition.

For the token exercise, the first unsupported step might be multiplying two unconditional probabilities for a dependent draw. Repair that step using conditional probability or pair enumeration. Do not merely replace the final number. Your error record should state why the original calculation failed.

Revisit the problem later without looking at your notes. Explain it in a short conversation, including a check and a limitation. If you can state the result but cannot explain why replacing a token changes one answer and preserves another, the reading has not yet become usable skill.

Difficulty also needs diagnosis. If the solution's first line invokes a distribution or theorem you cannot explain, pause for the prerequisite. If you understand the mathematics but become silent during follow-up, rehearse with a person who can interrupt. A harder book will not automatically fix a communication gap.

Avoid turning question counts into an invented readiness score. A solved problem with a changed assumption is stronger evidence of transfer than ten recognized answers. There is no validated number of completed book questions here that guarantees an interview pass.

Keep coding and research evidence in the plan

For a developer path, implement the token model as a small program that enumerates all distinct pairs. Check that the six pair probabilities sum to one and the expected blue count equals 3/2. Then change to ordered draws or replacement and explain why the outcome representation changes.

For a research path, add a simulation and compare its estimate with the exact answer. State the number of trials, random seed, and sampling mechanism. A reproducible simulation with a wrong model remains wrong; agreement on one aggregate is not enough to validate every conditional calculation.

For a trading path, practice presenting a calculation, checking a boundary, and revising it when a constraint changes. If your invitation includes a timed arithmetic assessment, rehearse that separately. We do not infer a timing format from a book's title.

Books can supply mathematical practice, but your own project or research discussion needs different preparation. Be ready to explain what you implemented, what you measured, and what would falsify your conclusion. That evidence cannot be purchased as another answer collection.

Five practice questions to test transfer

Question	What to test after reading
Probability That One Fair Die Beats Another	Count outcomes and check symmetry without forgetting ties.
Probability a negative review came from a lazy reviewer	State the evidence mechanism before applying Bayes' rule.
Probability That the Sum of 1,000 Fair Die Rolls Exceeds a Threshold	Explain an approximation and its assumptions.
Trading Game: Expected-Value Betting, Kelly Sizing, and Arbitrage Under Time Pressure	Separate expected value, sizing, and time-pressure decisions.
Find the Probability That Two Uniform Variables Have Product Above One-Half	Turn a verbal condition into a continuous probability region.

Start with one legally accessible book section and one Statistics and Math practice question. Attempt both without answers, explain one changed assumption, and use the resulting gap to decide what to read next. That gives you a concrete reason to choose a book, keep using it, or supplement it.