Analyst Scorecard Q3 2026: Who Actually Beat the Market?

How we score

Each pick is measured from the closing price on the day the teaser first ran. We compare against the S&P 500 over the same window.

Headline results

Only two of the eight writers we track beat the index over 365 days. Short-term pops after a campaign launch often fade within 90 days.

What the headline can support

The existing summary says two of eight tracked writers beat the index over a year. That describes the article's stated result, but the supplied material contains no underlying observation table, writer identities, or calculation files. Readers cannot independently reproduce the ranking from this page alone. It would be inappropriate to add names, exact returns, or claims of statistical significance. The useful addition is to explain what a defensible scorecard needs before anyone treats the headline as evidence.

An eight writer sample is not the entire newsletter industry. Selection criteria matter. If the list includes only publishers whose advertisements were easy to find, it may exclude quieter services or closed campaigns. If unsuccessful writers disappeared before the study began, the sample could look better than the true population. State the sample boundaries and distinguish a description of tracked promotions from a general conclusion about all paid research.

Define the unit being measured

A writer can promote the same stock several times, sometimes through different publications. Counting each mailing as a new pick can overweight one company and make the sample appear larger than it is. A scorecard needs a rule for repeated promotions, shared authorship, and overlapping campaigns. It should also distinguish a publicly teased idea from an explicit recommendation inside a paid model portfolio.

The stated approach starts at the closing price on the first teaser date. That is a reproducible convention only if the first date can be established. Preserve dated advertisements, email timestamps, and supporting records. A page that remains online for months is not evidence of its original launch day. When the first public date is uncertain, mark the uncertainty rather than quietly choosing a price that makes the eventual return look better or worse.

Keep the return windows honest

A thirty day observation, a ninety day observation, and a yearlong observation require separate endpoints. Decide whether those are calendar days or trading days and how weekends and market holidays are handled. Apply the same convention to the stock and benchmark. A newly launched campaign cannot have a completed yearlong result merely because the scorecard is published at the end of a quarter.

For an October 2026 publication, many teasers launched during the preceding year would still lack a full 365 day history. Those observations should remain pending for that window, not be treated as zero returns or silently excluded without disclosure. Report the eligible sample size for each horizon. Otherwise, readers may mistake a comparison of different groups of picks for a progression in the same group's performance.

Use comparable total returns

A stock's closing price alone does not capture the full shareholder result when dividends or other distributions occur. If the benchmark includes reinvested dividends, the stock comparison should do the same. Alternatively, label both series as price returns. Adjust for splits and investigate spinoffs, mergers, special distributions, and delistings. Corporate actions can turn an apparently simple price calculation into a misleading result if handled mechanically.

The S&P 500 is a reasonable familiar reference for US equities, but it is not a perfect risk match for every teased security. A portfolio of speculative small companies can behave very differently from large established businesses. Show the broad market comparison without implying that it removes sector, size, currency, or volatility differences. Additional benchmarks can provide context if their selection is consistent and not chosen after seeing which one produces the preferred conclusion.

Average, median, and hit rate answer different questions

The average return summarizes the arithmetic result across observations, while the median describes the middle result. A few extraordinary winners can lift the average even when most picks disappoint. The percentage beating the index adds another perspective, but it says little about the magnitude of wins and losses. A useful scorecard presents these measures together and explains whether each pick receives equal weight.

Writer level results also need a weighting rule. Giving every writer equal influence differs from pooling all picks, which gives prolific writers more influence. Neither choice is inherently dishonest, but they answer different questions. A reader evaluating a subscription might care about the typical writer, the typical recommendation, or a realistic portfolio of simultaneous holdings. The method should identify which experience the reported ranking is intended to approximate.

Costs and implementation can change the outcome

A closing price convention is useful for comparison, not a promise that every subscriber could transact at that level. An advertisement delivered after the market closes creates a different opportunity from one delivered before the opening bell. Bid ask spreads, thin liquidity, order size, and market impact can make actual entry prices worse than the recorded close. For overseas securities, currency conversion and local trading hours add further complications.

Subscription fees should be shown separately from stock performance unless there is a clearly defined account size and allocation method. The same annual fee has very different effects on a small account and a large one. Commissions, borrowing costs, and taxes may also matter. Avoid subtracting an arbitrary cost estimate and presenting the result as universal. Explain the assumptions so readers can judge whether the comparison resembles their circumstances.

Do not turn a retrospective ranking into a forecast

A writer who outperformed in one period may have benefited from a sector tailwind, a concentrated bet, or chance. Multiple picks in related companies are not fully independent observations. Rankings can change substantially when the measurement window moves. Comparing many writers and highlighting the best also increases the possibility that a standout result arose without a repeatable forecasting advantage.

Evidence of skill would be stronger with transparent records across different market conditions, consistent methods, and results that persist beyond the period used to identify winners. Even then, past returns do not guarantee future outcomes. Readers should also consider whether recommendations were practical, whether losing ideas remained visible, and whether the writer explained risk honestly. A performance table is one component of evaluating research quality, not a complete verdict.

FAQ and a practical reading checklist