Analyze trading RFQ competitiveness data
Company: Jane Street
Role: Data Scientist
Category: Machine Learning
Difficulty: medium
Interview Round: Take-home Project
# RFQ Competitiveness Analysis in Excel
You are given an Excel file containing a few months of trading RFQ (request-for-quote) data for a sales & trading desk. The data includes RFQs from 3 clients on 3 different assets.
Each row in the dataset corresponds to a single RFQ and has (at least) the following columns:
- `timestamp` – time of the RFQ
- `client_id` – which of the 3 clients sent the RFQ
- `asset_id` – which of the 3 assets the RFQ is for
- `rfq_side` – buy or sell
- `rfq_size` – requested notional/quantity
- `our_quote` – the price (or spread) your firm quoted
- `market_reference_price` – the mid‑market or reference price at RFQ time
- `is_competitive` – Boolean flag: whether our quote was considered competitive (e.g., inside some benchmark or relative to competitors)
- `won_trade` – Boolean flag: whether your firm ultimately won the RFQ (the client traded with you) or not
You are told only to “analyze the data”; there is no further guidance. You can use only Excel (including functions, pivot tables, charts, etc.).
**Task:**
Describe, in detail, how you would approach this analysis to help the trading desk understand and potentially improve its quoting and bidding strategy. In particular:
1. What questions would you try to answer from this dataset (e.g., about win rates, pricing competitiveness, client behavior, asset behavior)?
2. What data cleaning or preprocessing steps would you perform in Excel?
3. What descriptive statistics, aggregations, or pivot tables would you build, and why?
4. How would you quantify the relationship between quote competitiveness and probability of winning a trade (overall and by client/asset)?
5. What simple models or rules of thumb (they can be statistical but must be implementable/inspectable in Excel) might you build to:
- predict the chance of winning given an RFQ and a chosen quote; and
- suggest how aggressive the quote should be for different clients/assets/sizes?
6. How would you summarize and present your findings and recommendations back to traders or sales in a clear, actionable way?
Quick Answer: This question evaluates a candidate's skills in exploratory data analysis, Excel-based feature engineering, basic statistical modeling, and interpretation of RFQ competitiveness and win-rate relationships.