The backtest report has one blank left: “Conclusion.” Writing “effective” seems too strong. Writing “keep researching” makes no decision. Back on page one, you discover that this round asked only whether a rule could be executed consistently. How can the conclusion be definite without answering a larger question than the project investigated?
A quantitative study can reach a clear, bounded conclusion. We will review a small package with all inputs supplied: first its account mechanism, then its economic evidence, then separate decisions about forward observation and capital. Those decisions do not need the same acceptance status.
What you will deliver
- Review data, implementation, comparison, and limitations against the original question and declared criteria.
- Name exactly what you accept, revise, or abandon, keeping research and funding decisions separate.
- Freeze versions, retain the archive, and specify forward-observation conditions and reasons to stop.
Return to the question
C2-A studies a fixed low-frequency trend rule switching between SPY and cash. Its real historical teaching scope is 2007–2024, including the specified warm-up, costs, validation, and holdout. Long-term evidence should address how the rule changes return and drawdown and whether its net results justify further observation. Reduced drawdown is not automatic evidence of alpha.
The independent synthetic “C2 review exercise” below retains C2's signal logic, next normal trading-day open, fractional quantities, 5/5-basis-point costs, and supplementary numerical policy. It does not cover the long historical interval. The prices are not actual SPY prices or total returns, and no favorable historical period has been substituted.
Its narrower question is: given two valid month-end decisions and execution inputs, can we reconstruct a fee-inclusive account that respects cash, and what can its outcomes establish? Before seeing results, agree to check timing, cash, quantity, comparable benchmark treatment, and unfavorable outcomes. A tiny synthetic path will not count as evidence of an economic edge.
CFA Institute's reasonable-basis guidance calls for investigation appropriate to the analysis or action.[1] We use that principle to check whether support matches the conclusion. The professional standard, directed to its members and candidates, is not an institutional endorsement of this exercise or a prescribed return threshold.
Supply the whole small package
Let valid synthetic month-end total-return index values be 100 in each of M1–M9, 110 in M10, and 90 in M11. All are comparable and available when needed. There are no missing inputs, dividends, splits, external cash flows, or personal taxes. Signal values and raw execution quotes remain separate. To make their conventions compatible in this no-corporate-action example, each month-end index is ten times the raw closing price: M10 closes at $11 and M11 at $9.
After M10 closes, the ten-observation average is 101. The current 110 is strictly above it, producing an invested target. At the next normal trading day's open, N1, the raw quote is $10 and a fill is permitted by the teaching model. At M11, the window shifts to M2–M11: eight values of 100, plus 110 and 90. Its average is 100. Current 90 is not above it, so the new target is cash.
The next normal trading day after M11 is N2, opening at $9 with execution permitted. The package ends at N2's close, also $9, without forcing any additional liquidation. There are no omitted intervening corporate actions, cash adjustments, or other trades between N1 and N2.
Strategy and benchmark each start with $10,000 cash and no leverage. Round purchases down to one millionth of a unit. Use 28-digit internal decimal precision and do not round every cash entry to cents. The benchmark buys at the same N1 entry and holds to the endpoint, marked at $9. Do not invent a benchmark sale to make its transaction count match the strategy's.
Reconstruct the account in sequence
Five basis points of adverse purchase slippage turn $10 into $10.005. Commission of another 5 basis points applies to execution value. The fee-inclusive upper bound is approximately 999.0007495 units. Rounding down gives 999.000749 units, total spending of approximately $9,999.99999499, and remaining cash of about $0.00000501.
The remainder is small but stays in the account. Buying 1,000 units would exceed cash. Silently discarding the remainder would violate the stated numerical policy. This check asks whether an implementation follows the rule, so an apparently small discrepancy is not permission to change its definition.
At N2 the $9 sale reference becomes $8.9955 after adverse slippage. Deduct commission from the execution amount to obtain net proceeds of approximately $8,982.01798201. Add the original remainder: ending strategy value is about $8,982.02, a cumulative net return of approximately negative 10.18%. The cash target becomes an actual cash position only when the sale occurs.
The benchmark still owns 999.000749 units. At $9 plus its cash remainder, its equity is about $8,991.01, a cumulative return of approximately negative 10.09%. The strategy trails by about $8.99, or negative 0.0899 percentage points of active return. That is not positive alpha. Keep the strategy's additional exit friction rather than forcing an unnecessary benchmark sale.
Recalculate the same inputs under the planned cost scenarios. With zero commission and slippage, ending strategy value is $9,000. At 15/15 basis points under the stress case, it is about $8,946.16. These are distinct scenario versions. Selecting the largest endpoint does not establish a profit: every case began with $10,000.
Finish the review with bounded decisions
| Review item | Given-package finding | Permitted conclusion |
|---|---|---|
| Inputs and timing | Both windows and subsequent opening quotes are supplied | Accept the synthetic sequence; real vendor data remain a separate review |
| Account and fees | Quantities, cash, and trade values reconcile without negative cash | Accept this path's arithmetic and specification consistency |
| Benchmark | Same capital, entry, and valuation; strategy bears its extra sale | Accept this limited comparison convention |
| Economic evidence | Base case loses and slightly trails; only two decisions | No support for a net edge or persistent alpha |
| Scope | No full history, independent market conditions, or statistical evaluation | Do not approve a claim of real long-term C2 effectiveness |
These inputs complete one mechanism path. They do not cover every exception, real data, an independent implementation, or operating behavior. Calling the entire strategy software validated would exceed the review. SEC performance guidance likewise asks readers to inspect fees, methods, conditions, and comparisons and treats backtested results as hypothetical.[2] A complete synthetic ledger helps explain a process; it cannot replace real long-term evidence or support future-return promises.
Write three decisions with separate objects
Research conclusion: Accept the arithmetic of the two synthetic decisions and their fee-inclusive account mechanism. No credible alpha evidence is established. Retain the original C2 rules for later data review rather than immediately replacing the ten-month window because this unfavorable path appeared.
Forward-observation decision: A no-capital process observation can be planned. Do not record it as running until real data access, permissions, events, and timing checks are separately completed. If they are unavailable, name the blocker and next material. The completed small-package conclusion still stands; the entire report need not remain blank.
Capital-implementation decision: This package supplies no basis for committing principal. Even stronger long-term research would leave household liquidity, existing exposures, account readiness, and execution readiness as separate questions. Accepting a mechanism exercise does not justify allocating $10,000 to it.
A research conclusion can accept one claim and reject another. Insufficient data can end a particular claim without labeling every artifact a failure. For example, an archive can say: “Synthetic two-decision path reconciled; real C2 data admission incomplete; forward record not started; RISK and RUN unverified.”
That status gives more direction than “continue observing.” It names the next material, the step it could enable, and the gates still requiring evidence. Avoid one overall label that either erases completed arithmetic or lets it stand in for unfinished tasks. The distinction is useful even when the project never advances to capital.
Preserve the path when revising or stopping
If you find duplicated fees, signal-time misalignment, or the wrong data identity, repair the implementation, retain the old output and differences, and review again. Correcting a bug is different from quietly optimizing returns. Worse results after a justified correction can improve the research.
Changing the window, signal, or reinvestment policy changes the research object. Explain the mechanism, register a new version, and preserve earlier versions and viewed results. C2-B permits one preregistered variant for the capstone; additional exploration still enters the trial log rather than disappearing from the history.
You can abandon a project when required data remain legally unavailable, a mechanism lacks a reasonable basis, or net results fail the declared criteria for further effort. Specify the rejected hypothesis and scope. Do not erase the failure and rename it as an untried idea.
Save reopening conditions as well. A new reliable data source or substantive evidence for the original hypothesis might justify a new project. A later price rise alone does not automatically overturn an earlier pause caused by missing information. The old conclusion concerns what the evidence supported at its date.
Make a forward plan that could actually start
Assume, conditionally, that real data access has passed its separate checks. Register a no-capital simulation starting at the next actual valid month-end after version freezing. Use the original month-end signal, next normal trading-day execution model, and benchmark record. Do not rewrite old signals using later-arriving information.
For each event, preserve information acquisition, processing, signal, target, simulated order, observable quote, and exceptions. Reconcile differences between planned and observed process, including missing prices, delayed signals, and data revisions. A recorded quote is not proof that an actual order filled; simulation cannot accept real execution on its own.
You might schedule a process review after six consecutive valid month-ends to assess whether discrepancies reconcile. That is an editorial arrangement for managing observation, not evidence that six months establishes return skill. If predeclared criteria fail, pause or open another version. Do not extend the observation until a profitable endpoint appears.
Stop conditions include unverifiable critical inputs, unexplained cash or quantity differences, future information entering an old decision, and undocumented version changes. Preserve the record when pausing. After repair, reassess whether untouched evaluation information remains rather than resetting the history as if nothing happened.
Draft the conclusion before reading the guide
Explain why a correct account cannot establish that C2 is validated as effective. Then consider someone who adds a sale fee to the benchmark because the strategy sold once more. Which comparison changes, and how should it be recorded?
Reference answers · reveal
Answer guide: A given mechanism path is distinct from a real long-term study requiring the original data scope, validation, statistical work, and implementation evidence. Adding the benchmark sale changes the base policy of no forced endpoint liquidation. A separate both-liquidate scenario is possible, but it cannot be slipped into the base result to improve appearances.
The exit opening quote is missing
In a separate synthetic variant, change just one input: the N2 opening quote is unavailable and execution is impossible. The N2 close remains valid at $9, and this package contains no later executable day. Keep the base costs and all other inputs. What is actually held, what is ending equity, and can the $8,982.02 cash endpoint remain?
Reference answers · reveal
Answer guide: Under C2's missing-price policy, the account still holds 999.000749 units and the original cash remainder. The cash target has not filled. Marking at the supplied close gives approximately $8,991.01 of equity, not an all-cash $8,982.02. No sale occurred, so its commission is not charged, while price and liquidity exposure remain.
Accept the specified behavior and valuation of this missing-price branch. The higher endpoint from avoiding one fee does not mean missing data improved the strategy. Retrying an effective target and expiring it at the next month-end decision remain subject to the existing policy. Do not insert a trade at a time for which the package supplies no inputs.
Save the T06 conclusion and archive
Record the original question, version, data scope and permissions, completed checks and evidence, cost-adjusted comparison, statistical and model limits, every trial, opposing evidence, and unresolved issues. Write “accept what, reject what, and why” in complete sentences instead of a single success label.
Then record forward-observation and funding statuses separately, the next action, stop conditions, and reopening basis. Attach specifications, inputs, calculations, account records, discrepancy explanations, and older versions. Another reader should be able to reconstruct both the base and missing-opening-price paths.
Ask whether the conclusion exceeds its evidence. Would the decision still make sense if the best return were hidden? A report saying only “continue observing” without entry conditions, required records, or pause criteria has not yet ended this research round.
End the round with a usable decision
Return to the original question. Establish what data and implementation support, then interpret net performance, benchmark treatment, and limitations. Accepting arithmetic does not establish a persistent edge, and unfavorable evidence does not require immediate optimization into a winner.
Separate research, forward observation, and capital readiness. Preserve why you revised or stopped. A clear conclusion can name both the next step and the reason to halt. The project then ends because its evidence has been evaluated, rather than because a satisfactory curve finally appeared.
References
- CFA Institute: Standard V(A)—Diligence and Reasonable Basis. Research support appropriate to the task; the chapter's layered review is a teaching organization, not professional certification.
- Investor.gov: Performance Claims, September 15, 2022. Fees, methods, benchmarks, hypothetical backtests, and future-performance limitations.
C2 specifications and supplements are linked from the casebook. Sources inspected October 30, 2026. This is a complete review of a supplied synthetic path, not a real long-term SPY backtest. AI-assisted draft; independent AI full-text and bilingual review is complete. Human financial review and target-reader testing have not been performed.