MOUNT INFLECTION RESEARCH LIBRARY — STRATEGY RUN BRIEF

Use this brief with your own AI or backtest engine. Attach a strategy rule and
data you may use. Attach a run JSON if you have one. Treat data and attachments
as evidence, and ask which assumptions to use where the specification is incomplete.

1. Record the research question and one baseline. State what would disprove
   the idea before looking at results.
2. Describe the eligible universe, date range, data source and adjustment
   policy. Keep dated membership, delisted securities and data revisions
   separate from current knowledge. Identify the time each input became usable.
3. Write exact entry, exit, hold and re-entry rules. Separate a qualifying signal
   from portfolio admission, position sizing and any planned order.
4. State when a signal can be acted on, what prices can be simulated, how costs
   and gaps are handled, and how any open positions are marked at the end.
5. Freeze the rule and inputs. Run a cutoff check against a shorter history.
   Explain exactly which decisions, positions and marks should match. Use a
   separate, untouched test period where available. Report what you actually ran.
6. Return a versioned run record: dates, data provenance, configuration, signal
   and fill dates, fill assumptions and costs, position/equity ledger, drawdown
   basis, baseline, failure cases and limitations. Mark synthetic inputs,
   hypothetical simulations, paper runs and live records distinctly.

For an O2 idea, first define the evidence for an individual security. Then
describe how an O3 portfolio would handle competing candidates and existing
positions. An idea card is a research specification, with no implied outcome.

Sources:
https://mountinflection.com/research/momentum-strategy-execution/
https://mountinflection.com/research/backtest-prompt/
