Two Sigma OA, 3 Questions Passed: 2D Data Points + Pandas Filtering + Asset Returns via Linear Regression

Overview

• Format: 3 questions, 100 minutes
• Result: passed ✅

Question	Topic
Q1	Processing 2D data points
Q2	Pandas data filtering
Q3	Linear regression coefficients to compute asset returns

The question bank is small right now and the bar is relatively low. Happy to discuss!

Q2: Pandas data filtering — common patterns

Two Sigma's data questions care a lot about Pandas fluency. Practice these until you don't need the docs:

import pandas as pd

def filter_trades(df: pd.DataFrame, min_price: float, start: str) -> pd.DataFrame:
    """Keep valid trades with price >= min_price and date >= start, sorted by date"""
    df = df.dropna(subset=["price", "volume"])
    mask = (df["price"] >= min_price) & (pd.to_datetime(df["date"]) >= pd.Timestamp(start))
    return df.loc[mask].sort_values("date").reset_index(drop=True)

def daily_volume_by_symbol(df: pd.DataFrame) -> pd.DataFrame:
    """Daily trading volume per symbol"""
    return df.groupby(["symbol", "date"], as_index=False)["volume"].sum()

Note: combine conditions with & / | and wrap each condition in parentheses — and / or won't work.

Q3: Linear regression coefficients → asset returns

A common setup regresses asset returns y on market returns x with simple linear regression y = α + βx:

• β (beta) = Cov(x, y) / Var(x): the asset's sensitivity to market moves
• α (alpha) = mean(y) − β · mean(x)
• Returns usually come from prices via r_t = P_t / P_{t-1} − 1

def returns(prices):
    return [prices[i] / prices[i - 1] - 1 for i in range(1, len(prices))]

def regression(x, y):
    """Simple least squares; returns (alpha, beta)"""
    n = len(x)
    mx, my = sum(x) / n, sum(y) / n
    cov = sum((a - mx) * (b - my) for a, b in zip(x, y))
    var = sum((a - mx) ** 2 for a in x)
    beta = cov / var
    return my - beta * mx, beta

Edge cases: handle price series shorter than 2 and zero variance in market returns separately.