This time I’m meeting Optiver SDE post, three rounds in a row, the pace is very fast, but the overall experience can be described in two words: hardcore. Unlike traditional major companies, although the position is Software Engineer, the interview content is obviously not just pure back-end development. Transaction understanding, mathematical intuition, and expression skills will all be explored in depth. Let’s break down the entire process by round.
Round 1: Technical Interview (45min)
The first round is generally focused on technical expression + project implementation capabilities.
Let me start by introducing a technical project. I chose a trading signal backtesting system I built during my internship, and used Python to build the data pipeline, signal generation logic and PnL calculation module. I focused on architectural splitting: how the data layer, strategy layer, and execution layer are decoupled, and how to ensure that backtest results are reproducible.
The interviewer did not stop at what he had done, but immediately asked: What if the amount of data doubled? What if latency requirements change from seconds to milliseconds? I talked about vectorization optimization, reducing unnecessary dataframe copies, extending hotspot functions with C++, and parallel processing of historical data. This is obviously looking at engineering thinking rather than the ability to answer questions.
Then a dimensionality reduction expression problem arises: explain what a trading algorithm is in words that non-technical people can understand. I compare the logic of market making to the dynamic pricing of a milk tea shop. If there are more people, the price will be slightly increased, and if there are less people, coupons will be issued. The essence is to dynamically balance supply and demand and earn the price difference. The interviewer nodded, but also reminded not to use jargon.
Finally, a little question about option sensitivity: If Call's Delta is 0.5 and the stock price rises by 1 cent, how much will the option rise? I explain directly using proportions instead of writing formulas. The emphasis is on logical clarity.
Second round (60min)
The intensity increased significantly in the second round, starting to get closer to the heart of the trading firm.
It begins with a Market Making scenario discussion. Assume that an asset is currently quoted at $40,000. As a market-making system developer, how would you design the automatic order logic? How does your system respond if a large sell order suddenly appears? I talked about it from a system perspective: how the quotation engine dynamically adjusts the bid/ask spread based on inventory risk, how to expand or reduce the spread based on changes in order flow, and how to control risk exposure through the position limit mechanism. You are not required to be able to trade, but you must understand the matching logic and risk control ideas.
Then comes a Fermi estimation problem: estimate how many milk tea shops there are in Shanghai. I am not in a hurry to give numbers, but to dismantle them layer by layer from the perspective of population size, consumption frequency, and single store service radius. The interviewer is more concerned with whether the assumptions are reasonable than with the final number. Then enter a discussion of partial quantification: implementing the Black-Scholes pricing model from scratch. As soon as I finished writing the closed-form solution formula, the interviewer immediately asked me to re-derive it using the binary tree method. I can only build the risk-neutral probability step by step and then discount it back. After getting stuck for a while, he asked whether Ito's lemma could be used to simplify a certain step.
Round 3: Behavioral + Culture Fit (45min)
The final image is more detailed.
First up is the Why company. After doing research in advance, I mentioned their market-making expansion in the Asia-Pacific region, their technology-driven culture, and their emphasis on engineering efficiency based on public information. There must be no general talk here.
Then there are conflict experiences. I talked about the disagreements I had with the PM about the selection of strategy parameters during my internship. The debate was very intense at the time. In the end, the plan was verified by comparing the backtest data, and the strategy income increased by about 15%. The structure is strictly based on "conflict background-taking action-result impact".
Finally, I asked, what should I prepare before joining the job as a fresh graduate? The interviewer's answer is very interesting: less models and more mental arithmetic and game thinking. This actually highlights the core culture of the trading firm-response speed and decision-making logic.
Want to be more confident in passing the Trading Firm SDE interview?
We have been doing special mocks and mocks for Trading / Quant / SDE positions for a long time.Interview assistance . It has helped many students pass the technical aspects of trading companies such as Optiver, IMC, and Jane Street. If your goal is a trading firm instead of a traditional large factory, the preparation method must be different.
