Quantitative Finance: A Day in the Life + Moving Average Explained

A day in the life of a quant researcher

A quant researcher at a hedge fund like Two Sigma or D.E. Shaw spends most of their day in code and data — not in meetings. This is a key difference from banking or PE.

A realistic day:

8:30am — Review overnight P&L (profit and loss) from the live trading strategies. Scan for anomalies.
9:00am — Continue a research project: testing whether earnings announcement patterns predict short-term returns in small-cap stocks. Pull 5 years of data, clean it, run the initial regression.
12:00pm — Brief team meeting: each researcher presents one new finding or question. 30 minutes total.
1:30pm — Iterate on the model. Add a control variable (market cap) to the regression. Results improved.
3:00pm — Code review with a senior researcher. They find an error — your data was not adjusted for stock splits correctly. Fix it. Results change.
5:00pm — Write up findings in a research note for the team. Leave by 7pm.

Hours: 50-65/week. Much better than banking. The work is intensive but the environment is generally calmer.

Backtesting

Testing a trading strategy on historical data to see how it would have performed in the past. A strategy that "looks good" in a backtest may not work in live trading — markets change, and data-mining produces false positives. Good quants test for robustness.

The moving average crossover: a complete example

This is one of the most well-known quantitative trading signals. Walk through it completely — it teaches you how quants think.

The setup:

Compute the 50-day moving average (MA50) — average of the last 50 closing prices
Compute the 200-day moving average (MA200) — average of the last 200 closing prices
Buy signal: When MA50 crosses above MA200 (called the "Golden Cross")
Sell signal: When MA50 crosses below MA200 (called the "Death Cross")

The logic: When the short-term average rises above the long-term average, recent prices are higher than the historical average — a potential sign of upward momentum.

A numerical example:

On Day 200, MA50 = $48 and MA200 = $51. MA50 < MA200 → No position.
On Day 201, a rally occurs. MA50 = $52, MA200 = $51. MA50 > MA200 → Buy signal.
You buy at $52. Over the next 3 months, the stock rises to $68. The MA50 stays above MA200.
On Day 280, MA50 = $65, MA200 = $66. MA50 < MA200 → Sell signal.
You sell at $65. Profit: $65 - $52 = $13 per share.

Real-world example

In Python with Pandas, the entire calculation is a few lines:

df['MA50'] = df['Close'].rolling(50).mean()

df['MA200'] = df['Close'].rolling(200).mean()

df['signal'] = (df['MA50'] > df['MA200']).astype(int)

df['position'] = df['signal'].diff()

A "1" in the position column means a buy signal. A "-1" means a sell. You can then calculate returns for every buy-sell pair. This is a backtest in its simplest form.

Why this signal is interesting — and limited

What it gets right: Trend-following strategies work in trending markets. When assets have momentum (which historically they do, at least in some periods), this approach captures gains.

What it gets wrong:

Lag: Moving averages are by definition backward-looking. The signal fires after the move has started.
Whipsaws: In choppy, sideways markets, the MA50 crosses above and below the MA200 repeatedly — generating buy signals followed by quick losses.
Overfitting risk: If you test 1,000 parameter combinations (50/200, 40/180, 60/220...), some will look amazing in backtests purely by chance. A good quant tests out-of-sample, on data not used to design the strategy.

Transaction costs: Each trade has costs — commissions, bid-ask spread, market impact. A backtest that ignores these will always look better than live reality.

Fun fact

The "Turtle Traders" experiment in the 1980s: Richard Dennis and William Eckhardt bet on whether trading could be taught. They recruited 23 people with no trading experience and taught them a trend-following system (related to moving average concepts). Over 4 years, the "turtles" collectively earned over $175 million. Systematic, rules-based trading works — when the rules are well-designed.

Quant developer day vs. researcher day

Quant Developer:

9am: Code review — review a colleague's Python code for a new data pipeline
10am: Debug why the C++ execution engine is misaligning order timestamps with price data
2pm: Build a new module to calculate portfolio risk in real time
5pm: Write unit tests for the new module

Quant devs are primarily software engineers. They typically earn less than researchers but more than typical software engineers at most firms.

Scenario

Your moving average strategy looked great in backtests

Your MA crossover strategy returned 25% per year in your 10-year backtest. You show it to a senior researcher who asks three questions. What do they expose?

What does a 'Golden Cross' signal mean in moving average analysis?

What is the main danger of 'overfitting' in backtesting a quantitative strategy?

Quant work is writing code, analyzing data, and testing models — not client meetings. The moving average crossover is a complete example of quant thinking: build a signal, backtest it, question every assumption, test out-of-sample. Python makes all of this accessible starting today.

Why do moving average strategies perform poorly in sideways, choppy markets?