# Research Review and Peer Mentoring Review another learner's evidence without substituting confidence for competence. Use this workbook alongside the course. Write your answers before opening the solutions. Practical work is self-reviewed; scored knowledge checks are in the Academy. ## 1. Review foundations ### Separate learner and claim Review the claim and evidence rather than judging the learner's worth or confidence. A respectful review can still identify a material error. Separate a correctable calculation defect from a broad statement about someone's ability. ### Use explicit criteria Use explicit criteria for reasoning, data, calculations and limitations. A rubric helps keep reviews consistent, but it should not conceal uncertainty or reward presentation more heavily than correctness. ### Ask reproducible questions Ask questions another person can answer with records or a reproducible calculation. What observation supported this entry is more useful than why were you not confident. The former can be checked against the information set. ### Recognize limits of expertise Recognize the limits of your expertise and evidence. A peer review is not professional accreditation or a guarantee of strategy quality. Escalate specialist questions or mark them unresolved rather than inventing authoritative answers. ### Worked example A reviewer dislikes a losing trade but cannot identify a rule violation. The appropriate review asks whether the predefined process was followed, rather than treating the negative outcome as proof of incompetence. ### Independent exercise Write feedback that distinguishes the trade's outcome, the observed process and one unanswered evidence question. My inputs and assumptions: My calculation or decision: Evidence that would change my conclusion: ## 2. Technical review ### Check units and accounting Check units and accounting before interpreting sophisticated statistics. A missing multiplier or duplicate fill can invalidate every later chart. Begin with a small independently verified example and aggregate reconciliation. ### Challenge data availability Challenge data availability at the claimed decision time. Ask how revised releases, completed bars and later annotations entered the analysis. A convincing model score cannot repair an unavailable feature. ### Inspect sample selection Inspect the sample-selection process and exclusions. A reviewer should be able to trace raw observations to reported trades. Missing losing cases require investigation, not automatic acceptance of the author's summary. ### Review execution assumptions Review execution assumptions alongside the strategy logic. A signal's historical return is not equivalent to fillable P&L. Check quote side, delay, quantity and costs, and state which elements the available data cannot verify. ### Worked example A polished report has a contract multiplier omitted from every P&L calculation. The statistical analysis may be internally consistent but describes the wrong cash exposure. ### Independent exercise Write a prioritized review finding and specify which outputs need to be regenerated after correction. My inputs and assumptions: My calculation or decision: Evidence that would change my conclusion: ## 3. Feedback ### Prioritize material defects Prioritize defects by their effect on the conclusion or operating behavior. A wrong unit or lookahead error matters more than a minor formatting preference. Explain the impact so the learner understands the order of work. ### Explain errors with examples Use a small worked example to demonstrate the error and correction. Concrete arithmetic or an event trace is more useful than an unexplained statement that the model is wrong. ### Set a bounded correction task Set a bounded remediation task with an observable acceptance criterion. Asking the learner to improve everything at once makes it hard to assess whether the core issue was understood. ### Verify the correction Verify the correction on an unseen or neighboring case. Repeating only the original example may show memorization rather than transfer. Preserve unresolved questions for later review. ### Worked example A learner corrects a long-trade formula but still computes short P&L with the same sign. A second example with a winning short reveals that the underlying direction concept remains incomplete. ### Independent exercise Design a remediation task containing one long, one short and one invalid-input case, with independently calculated expectations. My inputs and assumptions: My calculation or decision: Evidence that would change my conclusion: ## 4. Teaching practice ### Adapt explanations to level Adapt explanations to the learner's prerequisite knowledge. Define units and terms before introducing advanced notation. Simpler language need not remove the important assumptions or limitations. ### Use worked and unseen examples Use both worked examples and independent tasks. The example demonstrates a method; the task checks whether the learner can apply it without copying the same numbers or event sequence. ### Assess transfer of understanding Assess transfer by changing the context while preserving the underlying concept. A learner who can calculate one memorized position size may still struggle when multiplier, currency or direction changes. ### Document unresolved questions Document unresolved questions and the next learning step. A review can end with a clear partial understanding rather than an inflated claim of mastery. Completion records should not imply competence beyond the evidence collected. ### Worked example A learner correctly repeats a ten-unit multiplier example but fails when the multiplier is twenty. The next lesson should address dimensional reasoning rather than simply provide more identical examples. ### Independent exercise Write a short teaching plan with a definition, worked example, changed-context task and follow-up criterion. My inputs and assumptions: My calculation or decision: Evidence that would change my conclusion: ## Course project Produce a structured peer review and a targeted remediation plan. ### Self-review rubric - Concepts and reasoning: 25% - Calculations, data and evidence: 30% - Process and risk controls: 25% - Limitations and communication: 20% Record one correction and one next practice task. This rubric is not automatically graded. ## Worked solutions ### Exercise 1 State the loss factually, identify which rules were or were not evidenced, and ask for the missing timestamp, sizing calculation or other specific record. Avoid inferring a process failure solely from P&L. ### Exercise 2 Identify the unit error as material, correct the calculation independently and regenerate P&L, sizing, drawdown and dependent statistics. Reassess the conclusion rather than merely fixing one table label. ### Exercise 3 Require correct direction handling for both valid cases and an explicit failure for the invalid case. Ask the learner to explain the units and sign, then verify on different numbers rather than only the demonstrated example. ### Exercise 4 Introduce how price-per-unit converts to cash through quantity and multiplier, demonstrate one case, vary the multiplier and direction independently, and require the learner to explain the cancellation of units in the final result. ## Further reading - https://www.itl.nist.gov/div898/handbook/ - https://www.cmegroup.com/education