Using Yield Curve PCA to Construct Curvature-Neutral Butterfly Trades

Summary

The exchange describes a common fixed-income application of principal component analysis: analyzing changes in yields across maturities. The first three components are characterized approximately as level, slope, and curvature, and are said to capture most of the yield variation. This gives traders a way to describe curve moves with a small set of factors.

For butterfly trades involving contracts at different maturities, PCA can provide risk weights that neutralize exposure to the first two components. A trader can then construct a position aimed at curvature while limiting exposure to broad level and slope shifts. The answer names examples of Treasury and interest-rate futures structures and points to a comprehensive reference, but supplies no exact weights, signal rules, or performance evidence. The outline explains portfolio construction and risk balancing, not a complete or proven trading strategy.

Key ideas

• Yield curve PCA components are commonly interpreted as level, slope, and curvature.
• The first three components capture most yield variation, according to the answer.
• PCA-derived risk weights can balance a butterfly against level and slope exposures.
• A balanced structure can focus more directly on yield curve curvature.
• The exchange gives a strategy outline but no trading signals, exact weights, or performance results.

Tags

# Principal Component Analysis and Yield Curves


# Principal Component Analysis and Yield Curves












I've been tasked with researching trading strategies relating PCA to trading fixed income futures instruments. Apparently PCA is frequently used in this area. I'm just looking for some references for obtaining a basic idea of what a strategy might look like. I'm not looking for a winning strategy -- just an outline of how PCA might be useful in generating trading signals. I understand the mathematics behind PCA and have used it in other areas, but its applications to finance are new to me.

## Answer by Helin (score 7, accepted)

https://quant.stackexchange.com/a/16110

One of the best pieces ever written on this topic is Salomon's "Principles of Principal Components," which is readily available on the Internet. I won't go into the details, since this paper is ridiculously comprehensive, but the fundamental idea is straightforward -- if you run a PCA based on yields, the first three components capture most of the variances, with the three factors roughly interpreted as the level, slope, and curvature of the curve.

The most widely used application for PCA is butterfly trading (e.g., you may buy the TY contract against FV and WN; or you may buy EDZ6 against EDZ5 and EDZ4). PCA allows you t compute the "risk weights" needed so that the structures are neutral to the first two principal components. This allows you to focus on trading the curvature of the yield curve, without taking on level/slope risks.

Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.