Ready to supercharge your AI agents?

Install any skill in seconds. Free to start, no credit card required.

Pyportfolioopt · mkurman | DecimalAI

mkurman/pyportfolioopt

SkillScore96Maintained

Portfolio optimization library: mean-variance, Black-Litterman, CVaR optimization, risk parity, Hierarchical Risk Parity (HRP), and CLA. Factor models, shrinkage estimators, and ex-ante risk analysis.

financeby mkurman

Links this skill to your agents and keeps it current with the author’s updates.

Automatic

Finance & Accounting pack

1 · One-time — install the CLI (docs ↗)

pip install decimalai

2 · Get the skill — writes .claude/skills/mkurman-pyportfolioopt/SKILL.md

decimalai skills pull mkurman-pyportfolioopt --out .claude/skills/

No CLI? Get the same file — curl · raw file

curl -fsSL --create-dirs -o .claude/skills/mkurman-pyportfolioopt/SKILL.md https://app.decimal.ai/s/mkurman-pyportfolioopt/SKILL.md

Raw file: /s/mkurman-pyportfolioopt/SKILL.md ↗ · machine-readable summary at /s/mkurman-pyportfolioopt.json ↗

In productionGlobal

how this skill behaves on real traffic · last 30 days

No production data for this skill yet — usage numbers appear here once agents start running it.

Anonymized aggregate across every org using this skill — proof from real production use, not a demo.

Does this skill help? · Benchmark: with vs. withoutsign in to run this on your agents

23 test cases from the publisher’s eval.yaml · each run twice on gemini-3.6-flash — with the skill injected, then without · last run 8d ago

Yes — +17% lift-3% tokens0% turns96% of whole cases pass with skill (22/23)

Test case	Without → With	Effect	Δ tokens	Δ turns
case-01	✗→✓	▲ Improved	-9%	0%
case-14	✗→✓	▲ Improved	5%	0%
case-18	✗→✓	▲ Improved	-40%	0%
case-19	✗→✓	▲ Improved	7%	0%
case-03	✓→✓	＝ Same ✓	-45%	0%

Showing 5 of 23 cases

SkillSafety

✓ 3 of 3 passed

Code scan15 of 15 checks clean

0 critical · 0 warning · 0 info — scanner v4 — 8/3/2026

live_secretremote_code_executionreverse_shelldata_exfiltrationssrf_metadataagent_snoopinganti_refusaldestructive_commandsobfuscated_unicodeunicode_homoglyphopaque_blobsuspicious_urltool_overreachtrigger_abuseprompt_injection_phrasing

IntentSafe

LLM judge · does the skill do what it claims

ContentClean

LLM judge · policy & harmful-content review

Source

Imported

mkurman / zorai

skills/scientific-skills/pyportfolioopt/SKILL.md

View on GitHub ↗

License MITImported 8/3/2026Version v1

Skill contents

SKILL.md

Overview

PyPortfolioOpt implements mean-variance optimization, Black-Litterman, CVaR optimization, risk parity, Hierarchical Risk Parity (HRP), and CLA. Handles asset allocation with factor models and ex-ante risk decomposition.

Installation

bash
uv pip install PyPortfolioOpt

Max Sharpe Portfolio

python
import yfinance as yf
from pypfopt import EfficientFrontier, risk_models, expected_returns

prices = yf.download(["AAPL", "MSFT", "GOOGL"], start="2022-01-01")["Close"]
mu = expected_returns.mean_historical_return(prices)
S = risk_models.sample_cov(prices)

ef = EfficientFrontier(mu, S)
weights = ef.max_sharpe()
print(ef.clean_weights())
perf = ef.portfolio_performance()
print(f"Return: {perf[0]:.2%}, Vol: {perf[1]:.2%}, Sharpe: {perf[2]:.2f}")

HRP

python
from pypfopt import HRPOpt
returns = prices.pct_change().dropna()
hrp = HRPOpt(returns)
weights = hrp.optimize()

References

Want to see if mkurman/pyportfolioopt would activate on your phrasing? Try it before you install — no LLM call, no signup.

Latest benchmark · publisher runVerified

8/21/2026, 11:03 UTC · 23 cases

Lift

+17%

96% vs 78%

Agent turns

1.0 vs 1.0

Tokens

-3%

1,487.91 vs 1,531.74

fail → pass · 4pass → pass · 18fail → fail · 1

Per-case results (23)

Case	Status	Duration (ms)			Turns			Tokens			Tool calls
Case	Status	Without	With	Δ	Without	With	Δ	Without	With	Δ	Without	With	Δ
case-01 I have a list of stock tickers and want to compute a maximum Sharpe ratio asset allocation using daily close prices pulled from yfinance. Can you write a Python script using PyPortfolioOpt that calculates expected returns and covariance, solves for the max Sharpe portfolio, and prints formatted allocation weights and performance metrics?	fail→pass	13,664	10,366	-24%	1	1	0%	2,751	2,494	-9%	0	0	—
case-02 I need a complete Python example that performs mean-variance optimization with PyPortfolioOpt on tech stocks. Please write code that loads historical stock prices, builds the expected return vector and sample covariance matrix, optimizes for maximum Sharpe ratio, and displays cleaned portfolio weights along with return, volatility, and Sharpe ratio output.	fail→fail	10,163	7,074	-30%	1	1	0%	2,039	1,745	-14%	0	0	—
case-03 I want to install Python portfolio optimization tools in my environment. What is the recommended terminal command to install PyPortfolioOpt fast using modern package management?	pass→pass	7,387	2,345	-68%	1	1	0%	1,230	681	-45%	0	0	—
case-04 I'm building an automated Python script to perform mean-variance portfolio optimization on US equities. I usually write manual numpy matrix operations for sample covariance and historical return calculations. How should I compute expected returns and covariance matrices using PyPortfolioOpt standard helper functions?	pass→pass	14,167	11,267	-20%	1	1	0%	2,740	2,511	-8%	0	0	—
case-05 After running an EfficientFrontier optimization, raw output weights often contain tiny numbers like 1e-16. I usually write custom loops to round and drop zero weights. What built-in PyPortfolioOpt method should be called on the EfficientFrontier object to clean up near-zero allocation weights?	pass→pass	3,729	3,459	-7%	1	1	0%	693	921	+33%	0	0	—
case-06 I solved for maximum Sharpe ratio on my EfficientFrontier object. I want to retrieve and print the expected annual return, volatility, and Sharpe ratio. Should I call portfolio_performance and how are its results returned?	pass→pass	6,816	4,678	-31%	1	1	0%	1,295	1,248	-4%	0	0	—
case-07 I am implementing Hierarchical Risk Parity in Python. I have a pandas DataFrame of raw daily close prices for my portfolio assets. Should I pass raw prices directly into HRPOpt, or do I need to transform the data first?	pass→pass	9,300	6,944	-25%	1	1	0%	1,715	1,539	-10%	0	0	—
case-08 I initialized HRPOpt with daily returns data for my asset pool. What method do I invoke on the HRPOpt instance to compute the asset weights?	pass→pass	3,610	2,994	-17%	1	1	0%	575	788	+37%	0	0	—
case-09 I want to calculate expected asset returns from historical price data using PyPortfolioOpt rather than writing standard pandas mean computations. Which specific module and function in PyPortfolioOpt computes mean historical return?	pass→pass	3,630	3,714	+2%	1	1	0%	625	933	+49%	0	0	—
case-10 When constructing the input matrices for EfficientFrontier, what function from PyPortfolioOpt's risk_models module calculates the sample covariance matrix from price data?	pass→pass	3,271	1,909	-42%	1	1	0%	490	631	+29%	0	0	—
case-11 I have calculated my expected returns vector `mu` and covariance matrix `S`. How do I instantiate the main mean-variance optimization class in PyPortfolioOpt using these inputs?	pass→pass	7,378	3,960	-46%	1	1	0%	1,436	1,017	-29%	0	0	—
case-12 Once I have created an EfficientFrontier instance `ef`, which method executes the optimization to maximize the Sharpe ratio?	pass→pass	2,785	2,331	-16%	1	1	0%	451	710	+57%	0	0	—
case-13 What is the exact class import statement for running Hierarchical Risk Parity in PyPortfolioOpt?	pass→pass	4,012	1,945	-52%	1	1	0%	651	651	0%	0	0	—
case-14 Write a short Python snippet that imports the Hierarchical Risk Parity optimizer from PyPortfolioOpt, calculates returns from a DataFrame named `prices`, instantiates the optimizer, and solves for weights.	fail→pass	4,089	2,626	-36%	1	1	0%	712	751	+5%	0	0	—
case-15 I need a complete Python script using PyPortfolioOpt to pull prices for AAPL, MSFT, and GOOGL starting from 2022-01-01 using yfinance, compute mean historical return and sample covariance, optimize for max Sharpe, print clean weights, and print the portfolio return, volatility, and Sharpe ratio formatted as percentages.	pass→pass	11,251	6,393	-43%	1	1	0%	2,180	1,702	-22%	0	0	—
case-16 When downloading ticker prices using yfinance for portfolio optimization input, which specific data column should be extracted from the downloaded DataFrame?	pass→pass	7,679	5,211	-32%	1	1	0%	1,317	1,210	-8%	0	0	—
case-17 Does HRPOpt accept raw asset price series or asset daily percentage returns? Explain the input expectation and show how to pass the data.	pass→pass	10,055	9,849	-2%	1	1	0%	1,698	2,125	+25%	0	0	—
case-18 When printing portfolio performance results from `ef.portfolio_performance()`, show the exact formatting style used to display Return, Volatility, and Sharpe ratio.	fail→pass	7,683	3,157	-59%	1	1	0%	1,511	910	-40%	0	0	—
case-19 What is the preferred package manager installation command for installing PyPortfolioOpt?	fail→pass	3,238	1,417	-56%	1	1	0%	449	481	+7%	0	0	—
case-20 Write a Python function to compute European call option prices using the Black-Scholes-Merton formula given spot price, strike price, risk-free rate, time to maturity, and volatility.	pass→pass	7,255	7,889	+9%	1	1	0%	1,529	2,079	+36%	0	0	—
case-21 Write a Python function using pandas and numpy to compute the 14-day Relative Strength Index (RSI) for a stock price series.	pass→pass	11,340	7,977	-30%	1	1	0%	2,305	2,050	-11%	0	0	—
case-22 Write a Python function to calculate the Net Present Value (NPV) and internal rate of return (IRR) for a sequence of annual discounted cash flows.	pass→pass	12,344	12,434	+1%	1	1	0%	2,493	2,760	+11%	0	0	—
case-23 I have a taxable investment account with capital gains and losses across individual lots. How do I implement a tax-loss harvesting algorithm in Python to minimize net realized capital gains subject to wash sale rules?	pass→pass	23,657	20,552	-13%	1	1	0%	4,345	4,285	-1%	0	0	—

These cases were also screened by the skill’s intent / safety pass.See the Trust review ↗Reproduce this benchmark in Colab ↗

How this skill was measured

DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 23 cases were attempted. The headline lift of +17 percentage points is the difference between those two pass rates over the 23 comparable cases.

Model measured: gemini-3.6-flash
Graded by: gemini-3.6-flash
Grading method: judged
Verification: verified
Pass rate without the skill: 78.3%
Pass rate with the skill: 95.7%
Cases attempted: 23
Cases counted toward lift: 23
Cases that regressed: 0
Skill version measured: v1
Run completed: 8/21/2026, 11:03 UTC

Measured example — same prompt, with vs. without the skill · from the benchmark run · case case-01

Prompt

I have a list of stock tickers and want to compute a maximum Sharpe ratio asset allocation using daily close prices pulled from yfinance. Can you write a Python script using PyPortfolioOpt that calculates expected returns and covariance, solves for the max Sharpe portfolio, and prints formatted allocation weights and performance metrics?

Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.

Show both answers

SkillScore blends 4 signals1 of 4 complete

BenchmarkPass rate WITH skill on a verified run96

Live evalNeeds ≥20 evaluated traces in 30 days

AI ratingNeeds ≥10 rated traces

AdoptionNeeds ≥10 serves across ≥2 orgs in 30 days

Efficiency · separate axis (at equal correctness)+0% turns+0% tokens

Shown for transparency; not credited unless a verified run shows no correctness regression.

Backed by 1 of 4 signals — each added signal makes the score more robust.

Verified on gemini-3.6-flash · re-tested 8 days ago

Other measured skills in the registry, with their headline benchmark lift.