Direct Answer
Bipower variation estimates the continuous part of price variance from the products of adjacent absolute returns. Because a single jump affects only one return at a time, jumps barely contribute, so the measure is robust to them. Comparing realized variance with bipower variation separates jump risk from diffusive volatility: when realized variance is clearly larger, a jump probably contributed.
Bipower Variation: Formula, Meaning, Signals, Examples and How to Use It
How Is Bipower Variation Calculated?
Bipower variation estimates the continuous part of price variance from the products of adjacent absolute returns. Because a single jump affects only one return at a time, jumps barely contribute, so the measure is robust to them.
Formula: BV = (pi / 2) x sum from i = 2 to n of |r_i| x |r_(i-1)|
Inputs:
- A series of intraday or daily returns r_i
- Number of returns n
Worked example. Four returns of 1.0%, -2.0%, 1.5% and -0.5% give a bipower variation of 0.000903. The inputs are illustrative numbers, not market data.
Reading it. Comparing realized variance with bipower variation separates jump risk from diffusive volatility: when realized variance is clearly larger, a jump probably contributed. The measure is in variance units, so take the square root for volatility.
Limits. It is designed for high-frequency data, and with few observations it is very noisy. Some versions add a small-sample scaling factor, and microstructure noise at very fine sampling can bias it.
What is Bipower Variation?
Bipower variation estimates the continuous part of price variance from the products of adjacent absolute returns. Because a single jump affects only one return at a time, jumps barely contribute, so the measure is robust to them.
How to use Bipower Variation
Comparing realized variance with bipower variation separates jump risk from diffusive volatility: when realized variance is clearly larger, a jump probably contributed. The measure is in variance units, so take the square root for volatility. A reading alone is never a buy or sell signal; it needs context from price structure, volume, volatility and the broader market regime.
Strengths
- Multiplying neighboring absolute returns means one jump enters only through a product with a normal return, which limits its effect on the sum.
- The pi over 2 constant scales the sum so that, without jumps, it estimates the same integrated variance as realized variance.
- Reproducible when the formula and data convention are fixed.
Weaknesses and limitations
- Two jumps in consecutive intervals multiply together and are not filtered out, so clustered jumps can still contaminate the estimate.
- Because it uses products of adjacent returns, it is sensitive to serial correlation in returns, which can bias the estimate.
- Correlated indicators can create false confidence by repeating the same underlying information.
Editorial and risk note
This page is educational content, not individualized investment advice. It should not imply guaranteed prediction, accuracy or outperformance.
Frequently Asked Questions
Is Bipower Variation a buy or sell signal?
No. Comparing realized variance with bipower variation separates jump risk from diffusive volatility: when realized variance is clearly larger, a jump probably contributed. It is designed for high-frequency data, and with few observations it is very noisy. A trade built on it still needs a trigger, an invalidation rule, a position size and realistic execution assumptions.
What is the best setting for Bipower Variation?
Bipower Variation takes these inputs: a series of intraday or daily returns r_i; number of returns n. There is no universal best setting for the tunable ones, so test them across instruments, regimes and out-of-sample periods and favor stable parameter regions over one historical winner.
Can Bipower Variation be used by itself?
It is designed for high-frequency data, and with few observations it is very noisy. Some versions add a small-sample scaling factor, and microstructure noise at very fine sampling can bias it. For that reason it is read next to measures of direction, participation, valuation, liquidity and risk, not alone.
Does Bipower Variation work on every timeframe?
The calculation (BV = (pi / 2) x sum from i = 2 to n of |r_i| x |r_(i-1)|) can be run on any bar size, but behavior is not portable: a reading on a five-minute chart describes a very different market window from the same reading on a daily chart. Validate the exact timeframe before relying on it.
Why does Bipower Variation give false signals?
Two jumps in consecutive intervals multiply together and are not filtered out, so clustered jumps can still contaminate the estimate. Because it uses products of adjacent returns, it is sensitive to serial correlation in returns, which can bias the estimate. Lag, noise and regime changes add to this, because market participants react to new information after the reading is calculated.
How should Bipower Variation be backtested?
Rebuild Bipower Variation from a series of intraday or daily returns r_i using point-in-time data, write the rules before testing, include delisted securities where relevant, model realistic fills and costs, reserve a validation sample and compare with a simpler baseline.