| Course | GB792 Advanced Topics in Finance |
|---|---|
| Unit | Unit 6 |
| Paper type | Seminar reflection |
| Length | About 860 words, 6 pages double-spaced |
| Format | APA 7 student paper |
| School | Purdue University Global |
| Program | MBA |
| Updated | October 2026 |
Free sample paper for GB792 Unit 6
Shrinking Is Not the Same as Vanishing: A Unit 6 Seminar Reflection on Published Anomalies
[Student Name]
Purdue University Global
GB792: Advanced Topics in Finance
Unit 6 Seminar Reflection
[Instructor Name]
[Date]
The classmates are composite; the studies discussed are real.
What I Believed
I wrote it in my notes before the session in almost these words: "Returns to published predictors fall after publication, so most of them were statistical flukes found by searching enough data." I had read Harvey et al. (2016), who counted hundreds of variables claimed to predict returns and argued that, given how many had been tried, the usual threshold for statistical significance was far too low. If researchers test enough variables, some will appear to work by chance, and those are the ones that get published. When the chance pattern meets new data, it disappears. That seemed to me the obvious reading of the decline in returns after publication.
What My Classmate Said
A classmate who works in quantitative equity research said that my reading made a prediction I had not checked. If predictors were flukes, their returns should collapse as soon as the data moved outside the period the original authors studied, whether or not anyone had read the paper. McLean and Pontiff (2016) designed their study to test exactly that. They took 97 variables that published studies had shown to predict returns and divided each one's history into three stretches: the original sample period, the period after the sample ended but before the paper was published, and the period after publication.
The results separated the two explanations. Returns were about 26 percent lower in the second period than in the original sample, and about 58 percent lower after publication. If predictors were pure flukes, the drop should have come entirely at the end of the original sample. Instead, most of the decline came after publication, when investors could read about the predictor and trade on it. The authors interpreted the gap between 26 and 58 percent as the effect of publication-informed trading, and the 26 percent as an upper bound on data mining.
My Objection
I raised one objection, and I still hold it. Publication dates are fuzzy. Working papers circulate on SSRN and at conferences for years before journal publication, and practitioners read them. If investors began trading on a predictor before its official publication date, some of the decline attributed to the period after the sample but before publication was really publication-informed trading under another name. That would make the 26 percent figure partly a trading effect rather than a data mining effect.
My classmate agreed that the dates were imperfect but pointed out that the objection cuts in an unexpected direction. If early circulation moved some trading into the middle period, then the true data mining effect is smaller than 26 percent, not larger. The objection weakens my original claim further rather than rescuing it.
My objection about dates was correct, but it implied that even less of the decline was due to flukes, which is the opposite of what I wanted it to show.
What Moved and What Did Not
My view moved in two ways. I no longer think the decline after publication shows that most anomalies were flukes. The three-period design shows that much of the decline follows publication, which is what one would expect if the patterns were real mispricings that investors learned about and then traded away. McLean and Pontiff also found that declines were larger for predictors with higher original returns and that returns persisted more in stocks that are costly to trade, both of which fit a story of informed trading held back by the costs and risks that limit arbitrage (Shleifer & Vishny, 1997).
What did not move is my concern about multiple testing. Harvey and colleagues' argument is not answered by McLean and Pontiff, because their sample consists of predictors that were published, and the problem Harvey describes is the many unpublished tests that never reach a journal. The 26 percent figure is an estimate for published predictors; it cannot say how many false predictors were never published, or how many published ones are false but have not yet been tested out of sample.
The Claim I Now Defend
The smaller claim I now defend is this: the returns to published predictors decline mostly after publication, which suggests that many of them reflected real mispricing that trading has partly removed, but the evidence cannot tell us what share of all claimed predictors are statistical artifacts, since the selection into publication is unobserved. Its limit is the one my objection exposed: the timing of publication is approximate, and any estimate that depends on it inherits that uncertainty. I would now describe an anomaly's post-publication decline as evidence about how markets learn, not about whether the original finding was real.
If I gave my original argument again, I would start by asking what each explanation predicts about timing, which is the question my classmate asked me. I had treated a single number, the decline after publication, as if it could speak for itself.
References
Harvey, C. R., Liu, Y., & Zhu, H. (2016). ... and the cross-section of expected returns. Review of Financial Studies, 29(1), 5-68. https://doi.org/10.1093/rfs/hhv059
McLean, R. D., & Pontiff, J. (2016). Does academic research destroy stock return predictability? The Journal of Finance, 71(1), 5-32. https://doi.org/10.1111/jofi.12365
Shleifer, A., & Vishny, R. W. (1997). The limits of arbitrage. The Journal of Finance, 52(1), 35-55. https://doi.org/10.1111/j.1540-6261.1997.tb03807.x
Reading the GB 792 Unit 6 instructions
Unit 6 in GB792 wants an account of one seminar exchange that shifted your view of a research question. Quote what you believed beforehand and the source that led you there. Describe the challenge precisely, including the design or evidence a classmate used. Explain what that evidence predicts under each competing explanation and what it found. Record your own objection honestly and what happened to it under discussion. State what moved in your view and what did not, with reasons for each. Give the revised claim you now defend, with its limit attached. Reflect on how you would argue the original point differently. Write in the first person, identify classmates only by role, and keep the length near 650 to 800 words. Cite the studies you discuss accurately in APA 7.
Inside the GB792 Unit 6 example
The reflection opens with the candidate's note that published predictors were statistical flukes, grounded in Harvey and colleagues. A quantitative researcher in the seminar explains McLean and Pontiff's division of each predictor's history into three periods and reports the 26 and 58 percent declines. The candidate's objection about early circulation of working papers is stated, and the classmate's answer, that it shrinks the data mining estimate further, is captured in the highlighted sentence. A section on what moved cites the larger declines for high-return predictors and the role of costly arbitrage, with Shleifer and Vishny. The multiple testing concern is kept because it concerns unpublished tests. The revised claim and a note on arguing from timing close the reflection.
Where the points sit in the GB792 Unit 6 rubric
GB792 instructors mark the reflection on the clarity of the change in reasoning. A quoted starting belief, tied to a source, earns credit because it makes the change visible. The challenge should be described with enough detail that a reader understands what it tests. Explaining what each explanation predicts, and what the evidence found, is weighted heavily. Recording an objection and its fate shows intellectual honesty, especially when the objection backfires. Separating what moved from what did not earns more than a complete reversal, since it shows the student kept valid concerns. A revised claim with its limit attached is expected. Accurate reporting of the studies, including their main figures, completes the grade.
Common GB792 Unit 6 mistakes, and how to avoid them
Many GB792 reflections summarize the seminar and announce a new view without saying what the old one was. Students often report the McLean and Pontiff figures without explaining why the three-period design separates data mining from informed trading. Objections are frequently dropped once challenged rather than followed to their conclusion. Some reflections abandon a valid concern, such as multiple testing, along with the mistaken belief. The revised claim is commonly vague. Quote your starting point, describe the design, say what each explanation predicts, follow your objection through, keep what still holds and state the narrower claim with its limit. When your seminar argued over a different study, your session notes are enough for a writer to start a reflection of this kind.
Get a GB792 Unit 6 example written to your instructions
What finding came under pressure in your Unit 6 seminar, and where did your view land afterward? Describe the exchange and add the reflection instructions with the rubric. Back comes a reflection that quotes your starting claim plainly, presents the strongest challenge in design terms and pins down the point you now give up. Your first custom sample is free, normally within 24-48h. The paper above is an original model document written by our desk, not a submitted student paper and not an official Purdue University Global document.
More GB792 papers and MBA samples
- GB792 Unit 1: Discussion Board Post
- GB792 Unit 2: Empirical Study Critique
- GB792 Unit 3: Event Study Appraisal
- GB792 Unit 4: Asset Pricing Memo
- GB792 Unit 5: Capital Structure Analysis
- GB792 Unit 7: Replication Review
- GB792 Unit 8: Valuation Model Critique
- GB792 Unit 9: Research Design Proposal
- GB792 Unit 10: Finance Research Paper
- GB791 Unit 6: Seminar Reflection
- GB740 Unit 6: Seminar Reflection
- GB730 Unit 6: Seminar Reflection
- GB720 Unit 6: Seminar Reflection
GB792 Unit 6 questions, answered
What if the seminar was replaced by a written alternative?
The shape barely changes. Instead of a classmate's challenge, the reflection responds to an assigned reading or an instructor's question, and says so in its opening lines. Starting position, challenge, counter, concession and revised claim still structure the piece, and the written version is judged on the same movement in reasoning as the live one.
Does this reflection conclude that markets are efficient?
It avoids that leap. Returns falling after publication fits investors learning from research, but it also shows the predictors survived for years before anyone traded them away. The reflection treats the evidence as bearing on how quickly prices absorb published findings, a narrower question than efficiency as a whole, and says so in its closing paragraph.
Why keep an objection instead of conceding fully?
Because a reflection in which every original belief collapses reads as rehearsed. The circulation objection is real: many predictors appeared as working papers years before journal publication, so the design's middle period is not as clean as its labels suggest. Keeping one argued objection shows the candidate weighed the challenge rather than simply adopting a classmate's view.
Where can I find a free GB792 Unit 6 sample paper?
A complete GB792 Unit 6 seminar reflection appears here, recording how McLean and Pontiff's evidence on 97 predictors narrowed a candidate's view that published anomalies were flukes.
What did McLean and Pontiff find?
Returns on 97 published predictors were about 26 percent lower after the original sample period ended and about 58 percent lower after publication, suggesting that investors learn from research and trade on it.