Skills

Build a Risk-Managed Portfolio Model

5 weeks · 0 milestones

Estimate risk and correlation, build a portfolio tested against equal weights out of sample, and stress-test it through real crises and a made-up shock.

Part of: Business & Management · Proves for: Career

Milestone map

3 milestones

Estimate risk and correlation

1–2 weeks

For a set of at least 5 assets, estimate expected returns, volatility and correlations from historical data, and show how unstable these estimates are over time.

Proof required

Submit the notebook with the estimates and a chart showing how correlation or volatility changes across time windows.

What gets checked

5+ assets with sources
Estimates shown for more than one window
Instability noted, not ignored

Common mistakes

One window treated as truth
Too few assets
No data source

Powstik Guide

Risk estimates with their uncertainty shown.

Steps

Pick 5+ assets.
Compute returns.
Estimate mean, volatility and correlation.
Repeat on rolling windows.
Chart the changes.

Template

Assets:
Windows:
Most unstable estimate:

What gets sent back

Single window only.
Fewer than 5 assets.
No source.

Resources

Foundationstart here

Depthgo deeper

What a verifier looks for

Checked by the verifiers you invite to your panel (up to 3). Pick people who can judge the work, not friends.
Check the time windows used
Ask which estimate is least reliable

You'll sign in first, then come straight back here.

Build and test a portfolio

2 weeks

Build an optimised portfolio and compare it with equal weights out of sample. Measure risk with value at risk and drawdown, and include costs.

Proof required

Submit out-of-sample results for your portfolio and equal weights, with VaR, max drawdown and turnover.

What gets checked

Compared to equal weights
Tested on later data
Risk shown three ways

Common mistakes

In-sample optimisation shown as results
No equal-weight comparison
Ignoring turnover costs

Powstik Guide

An optimised portfolio tested fairly.

Steps

Optimise on the early period.
Hold weights into the test period.
Build an equal-weight comparison.
Compute VaR, drawdown, turnover.
Compare honestly.

Template

Metric | optimised | equal weight
Return
Volatility
VaR (95%)
Max drawdown
Turnover

What gets sent back

In-sample only.
No equal-weight benchmark.
Missing risk measures.

Resources

Foundationstart here

Depthgo deeper

What a verifier looks for

Checked by the verifiers you invite to your panel (up to 3). Pick people who can judge the work, not friends.
Confirm weights were set before the test period
Ask why the optimiser beat or lost to equal weights

You'll sign in first, then come straight back here.

Stress test and defend it

1–2 weeks

Run the portfolio through at least two historical stress periods and one made-up shock. Present the full model and defend it live to a reviewer with finance or quant experience.

Proof required

Submit the stress-test table and the review log, including the reviewer's toughest question and your answer.

What gets checked

Two real stress periods and one scenario
Limits of VaR named
Reviewer has real finance or quant background

Common mistakes

Only calm periods tested
Treating VaR as a worst case
A reviewer without the background

Powstik Guide

A portfolio model that faced stress and questions.

Steps

Pick two stress periods.
Design one shock scenario.
Run all three.
Present to a qualified reviewer.
Log the questions.

Template

Review session log
Date:            Duration:
Reviewer name:
Reviewer role and experience:

Challenge 1 (portfolio risk under stress):
  What they asked:
  My answer:
Challenge 2:
  What they asked:
  My answer:

Change I made after the session:

What gets sent back

No stress test.
No scenario.
Unqualified reviewer.

Resources

Foundationstart here

Depthgo deeper

Masteryfor the dedicated

What a verifier looks for

Checked by the verifiers you invite to your panel (up to 3). Pick people who can judge the work, not friends.
Pick a stress period for them to run
Ask what VaR misses

You'll sign in first, then come straight back here.

Part of

Quant Modeling →

Related outcomes

Milestone map

3 milestones

Estimate risk and correlation

1–2 weeks

For a set of at least 5 assets, estimate expected returns, volatility and correlations from historical data, and show how unstable these estimates are over time.

Proof required

Submit the notebook with the estimates and a chart showing how correlation or volatility changes across time windows.

What gets checked

5+ assets with sources
Estimates shown for more than one window
Instability noted, not ignored

Common mistakes

One window treated as truth
Too few assets
No data source

Powstik Guide

Risk estimates with their uncertainty shown.

Steps

Pick 5+ assets.
Compute returns.
Estimate mean, volatility and correlation.
Repeat on rolling windows.
Chart the changes.

Template

Assets:
Windows:
Most unstable estimate:

What gets sent back

Single window only.
Fewer than 5 assets.
No source.

Resources

Foundationstart here

Depthgo deeper

What a verifier looks for

Checked by the verifiers you invite to your panel (up to 3). Pick people who can judge the work, not friends.
Check the time windows used
Ask which estimate is least reliable

You'll sign in first, then come straight back here.

Build and test a portfolio

2 weeks

Build an optimised portfolio and compare it with equal weights out of sample. Measure risk with value at risk and drawdown, and include costs.

Proof required

Submit out-of-sample results for your portfolio and equal weights, with VaR, max drawdown and turnover.

What gets checked

Compared to equal weights
Tested on later data
Risk shown three ways

Common mistakes

In-sample optimisation shown as results
No equal-weight comparison
Ignoring turnover costs

Powstik Guide

An optimised portfolio tested fairly.

Steps

Optimise on the early period.
Hold weights into the test period.
Build an equal-weight comparison.
Compute VaR, drawdown, turnover.
Compare honestly.

Template

Metric | optimised | equal weight
Return
Volatility
VaR (95%)
Max drawdown
Turnover

What gets sent back

In-sample only.
No equal-weight benchmark.
Missing risk measures.

Resources

Foundationstart here

Depthgo deeper

What a verifier looks for

Checked by the verifiers you invite to your panel (up to 3). Pick people who can judge the work, not friends.
Confirm weights were set before the test period
Ask why the optimiser beat or lost to equal weights

You'll sign in first, then come straight back here.

Stress test and defend it

1–2 weeks

Run the portfolio through at least two historical stress periods and one made-up shock. Present the full model and defend it live to a reviewer with finance or quant experience.

Proof required

Submit the stress-test table and the review log, including the reviewer's toughest question and your answer.

What gets checked

Two real stress periods and one scenario
Limits of VaR named
Reviewer has real finance or quant background

Common mistakes

Only calm periods tested
Treating VaR as a worst case
A reviewer without the background

Powstik Guide

A portfolio model that faced stress and questions.

Steps

Pick two stress periods.
Design one shock scenario.
Run all three.
Present to a qualified reviewer.
Log the questions.

Template

Review session log
Date:            Duration:
Reviewer name:
Reviewer role and experience:

Challenge 1 (portfolio risk under stress):
  What they asked:
  My answer:
Challenge 2:
  What they asked:
  My answer:

Change I made after the session:

What gets sent back

No stress test.
No scenario.
Unqualified reviewer.

Resources

Foundationstart here

Depthgo deeper

Masteryfor the dedicated

What a verifier looks for

Checked by the verifiers you invite to your panel (up to 3). Pick people who can judge the work, not friends.
Pick a stress period for them to run
Ask what VaR misses

You'll sign in first, then come straight back here.