Provide simpler-to-use extension methods for time series data

A financial time series library I was looking at had extension methods such as

series.MovingAverage()
series.MovingAverage(range)
series.Correlation(otherSeries, ...)
series.CrossVolatility(...)

etc. I wonder if Deedle should come with these pre-defined. I felt they were much easier to use than having to think through how to do this with windows, Series.mean etc., and allowed me to think at a higher level of abstraction.

cheers
don

[hmansell]

We discussed time series functionality in our initial scope meeting for Deedle, and wanted to make sure the design of Deedle was generic but TS functionality can be naturally layered on top. We originally thought it should be a separate library but I am not so sure now.

I think we have some of these implemented in our internal libraries so I will look to see if we can open-source them.

[tpetricek]

We added some of those in the Stats module and there are also corresponding extension methods.

Adding additional functionality like Correlation and CrossVolatility would still be nice. This could either be done in Deedle (if the implementation does not have dependencies) or as an extension in FsLab (http://github.com/tpetricek/FsLab) in case it has dependencies (like Math.NET Numerics, which we do not require in Deedle, but do in FsLab).

[zyzhu]

@tpetricek I'd like to revisit this issue. Pandas library comes with built-in functions such as df.cov to calculate covariance matrix. Pandas has dependency on numpy.
Shall we add dependency of Math.Net on Deedle out-of-box so that more mathematics functions can be added to Stats module? I wonder what's the pros and cons if we add that dependency.

[dsyme]

/repo-assist research the status of this issue after recent changes and advise

[github-actions]

🤖 This is an automated response from Repo Assist.

Status of This Issue After Recent Changes

This is a good opportunity to review how much has been addressed since this issue was filed in 2013. Here's a summary:

---

✅ Largely Addressed

Feature	Original Request	Current Status
Moving Average	series.MovingAverage()	Available as Stats.movingMean(n, series) / Stats.ewmMean(decay, series) with C# extension .MovingMean()
Correlation	series.Correlation(otherSeries)	Stats.corr(s1, s2) in core; full Pearson/Spearman corrMatrix in Deedle.Math
Covariance matrix	(implied)	Stats.covFrame, Deedle.Math.Stats.covMatrix
Volatility	CrossVolatility(...) (partial)	Deedle.Math.Finance has ewmVolStdDev, ewmVolRMS, ewmVol (single-series)
EWM functions	Not originally requested	Full suite: ewmMean, ewmVar, ewmCov, ewmCorr, ewmCorrMatrix
Extension methods (C#)	Explicit ask	SeriesStatsExtensions.fs and FrameStatsExtensions.fs expose these as .MovingMean(), .MovingStdDev(), etc.

The question about adding MathNet.Numerics as a dependency (from the 2018 comment) was resolved by creating the separate Deedle.Math NuGet package, which brings in MathNet and provides advanced statistics, linear algebra, regression, and PCA — without adding the dependency to the core package.

---

❌ Still Missing

• CrossVolatility between two series: There's no cross-volatility / cross-variance function. Deedle.Math.Finance only computes volatility for a single series or columns of a frame. A pairwise cross-volatility (EWMA covariance between two return series) would be a natural addition to Deedle.Math.Finance.

---

Recommendation

Given the above, this issue is substantially resolved. The main outstanding gap is pairwise cross-volatility, which is a narrow, specific ask.

Options for the maintainer:

1. Close this issue as largely resolved, and open a focused new issue for cross-volatility if it's a desired feature.
2. Keep open with an updated scope focused only on cross-volatility.

The Deedle.Math package is the right home for any new financial-math additions, since it already depends on MathNet.Numerics and has the Finance.fs module.

---

Repo Assist — automated analysis. Please verify before acting on this.

Generated by Repo Assist for issue #81 · ◷

To install this agentic workflow, run

gh aw add githubnext/agentics/workflows/repo-assist.md@30f2254f2a7a944da1224df45d181a3f8faefd0d

[dsyme]

/repo-assist

Regarding this:

CrossVolatility between two series: There's no cross-volatility / cross-variance function. Deedle.Math.Finance only computes volatility for a single series or columns of a frame. A pairwise cross-volatility (EWMA covariance between two return series) would be a natural addition to Deedle.Math.Finance.

please imlplement this