Look at two numbers that will make you sit up.
First number: $300,000.
This is the base salary for new graduates written in black and white on the official website of the top quant firm Jane Street—note, this is just the base salary, excluding bonuses. Including bonuses, the total package for a newcomer at a top quant firm is generally between $300,000 and $500,000. The average annual salary for the entire company at Jane Street (including administration and logistics) is $1.4 million.
Second number: 88%.
This is the growth rate of AI/ML job postings in the finance industry over one year. This industry not only pays well but is also expanding rapidly.
What surprises ordinary people even more is a quote on Jane Street's recruitment page: "No prior knowledge of finance or economics is required."
Yes, you read that right. This industry, which offers a $300,000 base salary to new graduates, does not require you to understand finance.
So what does it require? Can an ordinary person take this path? If so, what is the correct order?
In this article today, I will explain this path thoroughly. No piling up formulas, no academic jargon—just principles, methodology, and a clear roadmap.
First, Correct a Misconception: Quant Trading is Not "Stock Picking"
Most people think quant trading is: studying stocks, having an opinion on Tesla, or predicting earnings reports.
All wrong.
The essence of quant trading is mathematics, not stock selection.
What a Quantitative Trader (Quant) does is: look for statistical patterns, pricing deviations, and structural loopholes in the market. Why do these opportunities exist? Because the market is a complex system run by humans, and humans make systematic errors—panic selling during fear, chasing highs during greed, and hesitating at round numbers.
Quant traders don't predict "whether Tesla will rise tomorrow." They ask a different type of question:
"When A happens, what is the probability that B follows? Is this probability worth betting on?"
Remember this difference. It determines everything you need to learn next—not financial report analysis, not K-line charts, but a set of thinking patterns regarding uncertainty.
This Path is Like a Video Game: You Can't Skip Levels
The entire learning path is like a game where you cannot skip levels. Every level's concepts are built upon the previous one. The result of skipping levels is that you won't understand anything later.
The good news is: if you put in real effort (not watching mindless financial videos, but actually solving problems), it takes about 18 months to go from zero to being able to knock on the door of this industry.
Here are the five levels. I won't talk about formulas, only what each level actually trains in your thinking and why it's indispensable.
Level 1: Probability—Learning to "Think Conditionally"
Everything in quantitative finance ultimately boils down to one question:
"What are the odds? Are these odds in my favor?"
This is probability. The core of this level's training is a way of thinking that ordinary people rarely possess—conditional thinking.
Ordinary people think in "absolutes": something is either true or false.
Quant traders think in "conditions": Given the information I already know, how likely is this event?
For example, a stock rises on 60% of trading days—this is "base probability," it's crude and mostly useless. But if you find that on days when trading volume is higher than average, it rises 75% of the time—this "conditional probability" is the real money-making information.
At this level, you also need to learn one thing: updating your judgment in real-time based on new information (known as Bayesian updating). You originally thought a stock was worth $50, but the earnings report shows revenue exceeded expectations by 3%—how much should you adjust your judgment upward? The person who adjusts the fastest and most accurately takes the money.
Finally, there are two lifelong friends: Expected Value and Variance. Expected value is your win rate; variance is your risk. Remember the ultimate mantra of this level in one sentence:
If your strategy has a positive expected value and you can withstand the volatility—you will likely make money.
(Self-study this level for 3-4 weeks, 2 hours a day. The classic textbook is Harvard's free probability course, combined with writing some simulations—like simulating a coin toss 10,000 times to see the average converge to 0.5. Seeing it with your own eyes is a completely different level of understanding than just hearing about it.)
Level 2: Statistics—Installing a "Nonsense Detector"
After speaking the language of probability, you must learn to hear what the data is saying. This is statistics.
The first and most important lesson statistics teaches you is:
The vast majority of things that "look like patterns" are actually noise.
Here's a painful example. You develop a strategy, and backtesting shows an annualized return of 15%. Is it real?
The statistical approach is to first assume "this strategy is actually useless," and then calculate: If it really were useless, what is the probability of achieving such a good result? If this probability is very small, only then are you qualified to say "it might be real."
But there's a trap here where beginners get wiped out: If you randomly test 1,000 strategies, by pure luck, about 50 will look "significantly effective." You think you've found a gold mine, but you've just rolled several consecutive sixes.
So please accept a reality in advance: The first 10 strategies you find will likely be pure noise. Accepting this now will save you a lot of real money.
A core concept here is "Alpha" (α): subtract all known market factors from your strategy's return; the remaining unexplained excess return is your true skill. If nothing is left after subtracting—your "unique skill" was just a disguised way of "following the market."
Level 3: Linear Algebra—Learning to "See 500 Stocks Simultaneously"
This level sounds the most boring, but it is the engine for everything that follows: portfolio construction, risk management, and machine learning all rely on it.
Don't be afraid; its core idea can be explained in one sentence:
When you need to handle the relationships between hundreds of stocks simultaneously, you need a mathematical tool that can "batch process"—matrices.
For 500 stocks, there are over 120,000 pairwise relationships. Matrices allow you to pack these 120,000 relationships into one object and handle them in one operation.
The most magical moment at this level is when you first do "Principal Component Analysis" (think of it as taking an X-ray of a complex system): you will find that while 500 stocks seem to move independently, the first 5 "hidden factors" explain 70% of all fluctuations—the rest is mostly noise.
At that moment, you will suddenly understand: the market is not as complex as you thought; it is driven by a few forces. By learning to find these forces, you transform from a "person looking at 500 screens" into a "person looking at 5 dashboards."
(The golden resource for this level is MIT's Gilbert Strang Linear Algebra open course; it's free and globally recognized as the best.)
Level 4: Calculus and Optimization—Learning to "Find the Best Solution Under Constraints"
Everything in finance is changing: prices, volatility, correlations—they change every second. Calculus is the language for describing "change."
But for ordinary people, what's truly valuable at this level is optimization thinking:
Finding the optimal asset allocation under a set of constraints (limited money, risk caps, position limits).
This is what all "Robo-advisors" do behind the scenes. You don't need to derive formulas by hand, but you need to understand this framework—most decisions in life are essentially optimization problems with constraints.
Level 5: Stochastic Calculus—The Watershed Between "Enthusiast" and "Professional"
A quote from the original text says it well:
"Before learning stochastic calculus, you are a data scientist who likes finance. After learning it, you are a quant expert."
This level is the hardest and takes 6-8 weeks, but I can tell you what it does in plain English: building mathematical models for "pure randomness."
It ultimately leads to a trillion-dollar achievement—the Black-Scholes option pricing formula. This formula supports the operation of the entire global derivatives market.
There is an insight here that is chillingly profound when you first grasp it:
In the process of deriving option prices, the variable "how much this stock is expected to rise" is actually perfectly canceled out mathematically. In other words—the fair price of an option has nothing to do with whether you think the stock will go up or down.
This counter-intuitive conclusion means: you don't need to predict the future to accurately price future uncertainty. This is the essential difference between quant finance and "stock picking"—stock pickers bet on direction; quants price uncertainty.
The Four Types of People in This Industry
After completing these five levels, you can take four paths:
Quant Researcher—The Pattern Finder. Digging through massive data to find predictable patterns and design strategies. The highest barrier to entry (usually requires PhD-level math/stats/ML or extremely outstanding undergraduate performance). Researchers at top firms often call upon tens of thousands of GPUs.
Quant Developer—The Machine Builder. Turning the researchers' models into trading systems that can actually place orders. Requires solid programming skills (C++/Rust/Python) and low-latency system experience. For programmers, this is the smoothest entry point for a career change.
Quant Trader—The Trigger Puller. Managing money, managing risk, and making real-time decisions. Compensation fluctuates the most—it can reach eight figures in a good year, but you might get nothing in a bad one.
Risk Quant—The Brake Applier. Validating models, stress testing, and compliance. The ceiling is lower, but the career is the most stable.
The fastest-growing is the fifth type: AI/ML Quant—using deep learning for signal mining. Recruitment volume has grown by 88% in a year.
The full compensation landscape (based on 2026 data): New graduates at top firms (Jane Street, Citadel, HRT) have a total package of $300k-$500k; those who stay in the game until the 5th year have a median of $800k-$1.2M; star traders make $3M to $30M.
But note the words "stay in the game"—this is the median for survivors. A large number of people are eliminated over 5 years. The money is high because the screening is brutal.
What Do Interviews Test? Not Finance
To enter this industry, the interview process is roughly: Resume screening → Online assessment (mental math + logic) → Phone interview (probability questions, betting games) → Final round (3-5 consecutive rounds of simulated trading, programming, whiteboard derivations).
One interesting detail: Jane Street will deliberately give you a difficult problem that one person cannot solve alone—they are not testing whether you can do it, but how you use hints and collaborate with others.
Another data point shows what the industry wants: Among Jane Street's recent interns, two-thirds studied Computer Science, one-third studied Math—almost none had a finance background.
This industry buys your way of thinking, not your financial knowledge.
Three Key Lessons
First, the real enemy is "Estimation Error."
All mathematical models work perfectly under the premise that "parameters are real." But you never get real parameters—you only have parameters estimated from historical data, carrying errors. The gap between theory and practice is always estimation error. The best quant traders are not those with the best math, but those who respect error the most.
This also applies to ordinary people's investing: any "precisely calculated" return prediction is worth being wary of—algorithms can be precise, but inputs are always crude.
Second, tools have been democratized, but "conviction" has not.
Today, anyone can use top-tier quant libraries, data interfaces, and machine learning frameworks for free. Technology is necessary, but it's no longer scarce. The real advantage (known as the edge) only exists in three places: unique data, unique models, or unique execution—not in having more software packages than others.
Third, math is the moat.
AI can already write code and give strategy suggestions. So what value is left for a quant trader?
What remains is: the ability to understand "why." Knowing why a formula holds, under what conditions a model fails, and where a seemingly smart strategy hides a landmine—this mathematical intuition determines whether you are a "creator of advantage" or a "borrower of advantage."
And borrowed advantages always expire.
Three Insights for Ordinary People
Even if you don't plan to actually become a quant, there are three universal truths in this path:
First, the real barriers to high-paying industries are often different from what you imagine.
Everyone thinks you need to know finance to enter quant trading, but they explicitly state "no finance knowledge required"—what they want is probabilistic thinking, statistical literacy, and problem-solving skills. The real barrier to many seemingly unreachable industries is a learnable way of thinking, not the industry's "jargon."
Second, "no skipping levels" is a common law for all hard skills.
The five levels are interconnected; skip any one, and everything follows will collapse. This is why "18 months of hard work" can beat "3 years of fragmented learning"—if the order is right, time is compound interest; if the order is wrong, time is a loss.
Third, in the AI era, the value of "understanding principles" is skyrocketing, not shrinking.
The more popular tools become and the more capable AI is, the scarcer "knowing why" becomes. This industry tells you with the clearest price signals: those who use tools earn $10,000 a month; those who understand principles earn $1 million a year. The difference isn't the tool; it's that layer of "why."
Finally
Back to the original question: Can an ordinary person become a quant trader?
Yes. This path doesn't look at your origin, background, or even if you understand finance—it only looks at whether you can settle down and break through the five levels one by one.
It's not easy. 18 months, two hours a day, doing real problems, writing real code, and deriving real logic—there are no shortcuts, no quick fixes.
But look at it another way: a path with transparent rules, a clear roadmap, incredible returns, and an explicit announcement that "we don't look at your past" is rare in today's world.
Most people will bookmark this article and then continue scrolling through their timeline.
A few will open Claude or Codex tonight and use AI to write down the first strategy they've ever practiced.
18 months later, the lives of these two types of people will not be on the same track.
