Paper Replication

A repeatable way to replicate published finance papers: a target sheet of every number, a written protocol and tolerances, a discrepancy log, out-of-sample tests and a write-up of what held.

Free · Mac (Apple silicon) and Windows · v1.0.5 · Other platforms

Then open the Marketplace in the app, search for Paper Replication, and click Install. It goes into your own files, and works offline from then on.

Before you start: this plan is general information to help you organise decisions about money. It is not financial, investment, tax or legal advice, and it does not recommend any product, investment or deal. Rules differ between countries, so check decisions that matter with a qualified, licensed adviser in yours. Investing and trading can lose money, and raising or managing other people's money is regulated in most countries.

What it installs.

1

Pillar

1

Area

50

Projects

204

Starter tasks

🧮 Quantitative Finance & Trading

Paper Replication

Replicating published finance papers: getting the data, matching the original results, testing out of sample, and writing up what held and what did not.

  1. Choosing the first paper to replicate
  2. Target sheet of every published number
  3. Close reading of the method section
  4. Variable definitions crosswalk from text to code
  5. Data requirements map for the replication
  6. Academic and free data access set-up
  7. Replication project skeleton and environment
  8. Replication protocol written before running code
  9. Tolerance rules for matching published results
  10. Contacting the original authors
  11. Weekly replication work session
  12. Discrepancy log for mismatched numbers
  13. Data vintage and extract record
  14. Pipeline of candidate papers to replicate
  15. Monthly replication review with a peer
  16. Comments, corrigenda and replies watch
  17. Quarterly rerun on the latest data vintage
  18. Replication package tidy-up after each milestone
  19. Portfolio sort mechanics as papers use them
  20. Reproducing reported standard errors
  21. Reading empirical asset pricing tables fluently
  22. Accounting data timing and look-ahead conventions
  23. Sample filters and delisting returns
  24. Event study replication technique
  25. Time series predictability regression replication
  26. Toy data checks for each replication step
  27. Exact versus conceptual replication decision
  28. Free data substitute versus licensed database test
  29. Out-of-sample extension after the paper's sample end
  30. International sample test of a single-market result
  31. Published anomaly net of realistic trading costs
  32. Sensitivity grid on the paper's key choices
  33. Go or stop decision on further research
  34. Replication write-up of what held and what did not
  35. Presenting a replication at a reading group
  36. Submitting a replication for publication
  37. Public replication package release
  38. Replication as a portfolio piece for applications
  39. Master's dissertation built on a replication
  40. First-year doctoral replication assignment
  41. Replication on a laptop with free data only
  42. Team replication in a reading group
  43. Replicating a paper for an investment committee note
  44. Supervising a student replication
  45. Restarting a stalled replication
  46. Replicating a machine learning return prediction paper
  47. Replicating a high-frequency result with tick data
  48. Replicating an option-implied information paper
  49. Post-publication decay study across replicated anomalies
  50. Writing a comment on a result that did not replicate

Where it starts.

Each project arrives with its purpose, milestones and first tasks written out. Here is how the first three read.

Choosing the first paper to replicate

Picking a paper that is too ambitious is the usual reason a first replication stalls in week three. A good first target has a clear method section, data you can actually get, a sample you can run on a laptop and tables with enough detail to check against. Scoring three or four candidates on those points takes an evening and saves months.

First tasks

  • List three or four papers you would like to rebuild and note their main tables
  • Score each one for data access, method clarity, compute needs and table detail
  • Check whether each paper has an online appendix or published code
  • Record the chosen paper and the reason it won in your replication folder

Target sheet of every published number

You cannot say a result replicated unless you know exactly which numbers you are trying to hit. Copying every coefficient, t-statistic, average return, sample count and date range from the target tables into a structured sheet turns a vague goal into a checklist and makes the later comparison mechanical rather than impressionistic.

First tasks

  • Mark every table and figure in the paper you plan to match
  • Type each reported number into a sheet with table, row, column and units
  • Add the sample period and observation counts as their own target rows
  • Add columns for your replicated value, the difference and a status flag

Close reading of the method section

Published method sections compress months of choices into a few paragraphs, and the details that matter most often live in footnotes or the appendix. Reading the paper three times with a specific question each pass, and writing down every choice the authors made or left unstated, gives you the list of decisions your code must take.

First tasks

  • Write a five-sentence summary of the paper's claim and main test
  • List every explicit method choice, including those buried in footnotes
  • List each unstated choice and write the default you will use
  • Read the online appendix and note anything that contradicts the main text

Make it yours.

Install Paper Replication, then tell Aurum the date you're aiming for. It checks whether the plan fits the weeks you actually have.

Free · Mac (Apple silicon) and Windows · v1.0.5 · Other platforms

Blueprints are general information to help you organise your own work. They are not medical, financial, legal or other professional advice. See the Terms.