High Frequency Data
Traditional finance uses data that is spaced evenly in time (e.g., Daily closing prices). High Frequency (HF) finance uses data that arrives at irregular intervals—every time a trade happens or a quote changes.
What is HF Data?
High Frequency data records market activity tick-by-tick.
- Low Frequency: Daily, Weekly, Monthly.
- High Frequency: Milliseconds, Microseconds, Nanoseconds.
Note
Analogy:
- Low Freq: Checking the score of a cricket match every hour.
- High Freq: Watching the match ball-by-ball. You see much more detail, but it's also much noisier.
Key Characteristics
1. Irregular Time Spacing
In normal time series, t increases by 1 day or 1 minute. In HF data, trades happen whenever they happen. You might have 1000 trades in one minute (market open) and 0 trades in the next minute (lunchtime).
- Implication: We often work in Event Time (Trade 1, Trade 2...) rather than Clock Time (9:15, 9:16...).
2. Microstructure Noise
At very high speeds, prices bounce around due to the Bid-Ask Bounce. This is "noise" rather than true value change.
- If a buyer buys at Ask (100.05) and a seller sells at Bid (100.00), the price looks like it dropped 0.05, but the value didn't change.
3. Massive Volume
HF data is huge. A single stock can generate millions of data points per day.
Types of HF Data
Loading comparison…
Why Study HF Data?
- Algorithmic Trading: Algos react to microseconds. You need HF data to build them.
- Market Microstructure: To understand how liquidity works and how prices are actually formed.
- Realized Volatility: You can measure volatility much more accurately using 5-minute intraday data than using daily data.
Loading quiz…