OIPD v2.0.4

We made a no-code web interface!

• Check open-lemma.com
• And revamped the documentation to be much clearer

Thanks to our new contributors 👇 👇

• NewMark Risk
• Vivek Rao

A lot of bug fixes!

Probability Domain and CDF Improvements

• Added full-domain probability grids for ProbCurve and ProbSurface,
  including smart grid_points=None sizing and explicit full_domain=True
  exports/plots.
• CDFs now use the 1st derivative of call-price instead of integrating
  the PDF (which is itself a 2nd derivative)
• density_results() and plot() now default to compact, smooth view domains;
  points controls view resolution, while grid_points controls native
  numerical resolution.

Lazy Probability Materialization

• Added lazy probability materialization for ProbCurve and ProbSurface.
  Probability objects store the fitted-volatility recipe first and
  build price/PDF/CDF arrays only when queried, exported, or plotted.
• implied_distribution() now accepts grid_points=None and
  cdf_violation_policy="warn".

Warning Diagnostics

• Added warning_diagnostics events and summaries on public curve and surface
  objects, with concise grouped warnings instead of repeated row/expiry noise.
• Added warn-and-repair CDF monotonicity handling via
  cdf_violation_policy="warn"; strict users can choose "raise".
• Added diagnostics for stale quotes, price fallback, SVI butterfly risk,
  skipped expiries, CDF repairs, and fan-chart skips.

Market Data, Forward Inference, and SVI Robustness

• Improved Black-76 put-call parity forward inference: nearest-ATM selection,
  selected-subset outlier filtering, richer diagnostics, and no influence from
  far-from-ATM pairs that merely pass validation.
• Temporarily disabled the default bid/ask relative-spread gate; explicit
  max_bid_ask_relative_spread values still apply.
• Added bid/ask-only ingestion, raw yfinance
  spot handling, and lowercase volume preservation.
• Fixed SVI weighting so reliable bid/ask spreads take precedence over volume,
  with diagnostics for the chosen auxiliary weight source and fallback reason.
• Simplified the public volatility/probability API by hiding legacy
  Black-Scholes, Newton solver, and explicit dividend-input knobs; default fits
  remain parity-forward Black-76.

OIPD v2.0.3

What's Changed

• Updated valuation_date and expiry date to accept timestamps, instead of just calendar dates
• This provides support for 0DTE options

Case study: spotting advanced knowledge trading 18min prior to Trump tweet

Trump announced a 90-day pause on most reciprocal tariffs on April 9, 2025. Reuters reports indications of advanced knowledge trading

• Trump announced tariffs at 1:18pm, moving the SPY price substantially
• we see abnormal implied distributions starting from 1pm, prior to his tweet

Notes:

• implied distributions of 0DTE April 9 2025 options
• 1m frequency SPY price downloaded using Alpaca free API
• free 1m options data from London Strategic

OIPD v2.0.2

Minor updates:

• Added DataFrame export methods for fitted results:
  • ProbCurve.density_results(domain=None, points=200)
  • ProbSurface.density_results(domain=None, points=200, start=None, end=None, step_days=1)
  • extended VolSurface.iv_results(domain=None, points=200, include_observed=True, start=None, end=None, step_days=1)

OIPD v2.0.1

OIPD v2: Overview of new capabilities

Previously in v1, OIPD generated the probability distribution of a future asset on a single future date.

OIPD has advanced in 2 directions:

1. It computes the market's expectations about the probable future prices of an asset, based on information contained in options data.

2. It provides an opinionated, end-to-end volatility surface fitting pipeline:

v2.0.1 Updates:

• Improved the theory of interpolating the probability surface. Probabilities can't be interpolated over time directly, only the volatility smiles can be interpolated linearly in total-variance space. So we've made the ProbSurface use the fitted VolSurface to directly
• Bug fixes: Made the t input consistent everywhere (same domain requirements and same type requirements)

Special thanks everyone who contributed to the development of v2:

Thanks to everyone who has contributed code:

And special thanks for support on theory, implementation, or advisory:

• integral-alpha.com
• Jannic H., Chun H. H., and Melanie C.
• and others who prefer to go unnamed

OIPD v2.0.0

OIPD v2: Overview of new capabilities

Previously in v1, OIPD generated the probability distribution of a future asset on a single future date.

OIPD has advanced in 2 directions:

1. It computes the market's expectations about the probable future prices of an asset, based on information contained in options data.

2. It provides an opinionated, end-to-end volatility surface fitting pipeline:

Special thanks everyone who contributed to the development of v2:

Thanks to everyone who has contributed code:

And special thanks for support on theory, implementation, or advisory:

• integral-alpha.com
• Jannic H., Chun H. H., and Melanie C.
• and others who prefer to go unnamed

1.0 release!

What's Changed

• Improve put-call parity preprocessing by @tyrneh in #64

New Contributors

• @elainechang362 made their first contribution in #63

Full Changelog: v0.0.6...v1.0.1