From flash crashes to fake reviews, bad data and worse intentions can shake financial markets, damage reputations and destroy trust. This investigation examines 12 unethical data practices , with real-world examples and verified sources , for financial professionals, data scientists, regulator and anyone who depends on data integrity.
On May 6, 2010, trader Navinder Singh Sarao's spoofing behavior of creating fake orders with no intent to execute contributed to the Flash Crash, erasing billions in minutes. WSJ
Oobah Butler's fake restaurant "The Shed" climbed to #1 on TripAdvisor through fabricated reviews. The FTC has since banned the sale and purchase of fake reviews. FTC
Advisers recommending higher-commission products without disclosure breach fiduciary trust. The SEC has brought marketing-rule cases against advisers for undisclosed endorsements. SEC
In 2018, the SEC charged robo-advisers for presenting back tested results as live performance. Backtests ignore slippage and trading costs, often overstating real-world outcomes. SEC
Diluting contaminated data with cleaner records without disclosure creates a false sense of reliability. Honest analysis must document contamination and remaining limitations.
Excluding outliers can make a model appear stronger while erasing the very cases where it fails. Ethical reporting requires explaining every exclusion and its implications.
Smoothed returns conceal volatility and risk. If a fund's real experience includes sharp losses, investors deserve to see that truth not a prettier curve.
Running enough tests until one looks statistically significant then publishing only that result is selective reporting. Ethical research requires disclosing the full process.
A strategy can look excellent if you choose a favorable date range. The most honest presentation includes both strong and weak periods so readers can see the full pattern.
Data from narrow demographics or scraped sites may not represent the broader population. Undisclosed collection methods make results appear scientific while remaining unbalanced.
Privacy violations occur when companies bury consent in unreadable terms. Regulators increasingly emphasize meaningful consent asking clearly, explaining plainly and limiting collection. Davies
Unauthorized sharing exposes people to additional harm. Once data passes to third parties, the original user may lose control over how it is used, stored or sold.
Question performance claims: Ask whether results are back tested or live. Demand disclosure of conflicts, time periods and methodology before making decisions.
Disclose methods. Label simulations clearly. Report the full analytical process: not just the results that worked. A polished dataset with hidden defects is always less ethical than an honest one with documented flaws.
The Dark Side of Data: 12 Unethical Practices That Could Cost You Everything