This InsightMeter guide is for readers who already collect screenshots of winning calls—and who still treat a highlight reel as proof of skill. Your reader job is bias literacy: name survivorship bias, ask “out of how many?”, separate hit rate from magnitude, demand minimum sample sizes, and understand how InsightMeter-style scoring tries to keep failed names and unresolved outcomes in view. Sibling guides cover forecast track records, calibration versus confidence, backtest versus live scoring, falsifying a smart-money thesis, and reading forecast scorecards. Soft product note later: Free may show short educational teasers while Pro can unlock longer history and CSV—never a hard sell.
A track record is a claim about past predictions or past picks under stated rules. A trophy shelf is a curated gallery of winners. Retail tip culture loves trophy shelves because they are shareable. Education prefers track records with denominators: how many forecasts or picks entered the sample, how many exited by delisting or silence, and what would count as a miss.
Write one primary question per note. Good questions: “What was the full starting universe?” “How many calls were made, and how many are shown?” “What horizon and benchmark were fixed before looking?” Bad questions: “This guru was right three times—should I copy the next trade?” Stop if the question is an order ticket.
Soft product note: InsightMeter scorecards and research tables are built to support explicit definitions—horizons, universes, and documentation—not vibes. Free teasers teach vocabulary on short samples; Pro can unlock longer history and CSV within stated windows. Soft invitation: after the checklist, you may create a free account (utm_campaign=checklist30s)—Pro is optional and never a return promise.
Survivorship bias happens when the sample you see is filtered toward what survived—successful funds, continuing commentators, non-delisted tickers, or remembered hits—while failures quietly drop out. The surviving set looks smarter than the full starting set. The bias is about missing losers, not about insulting winners.
Classic retail shapes: (1) a newsletter that republishes only correct calls; (2) a “smart money” thread that lists famous managers’ winning holdings without closed funds or trimmed names; (3) a backtest that drops delisted stocks; (4) a social collage of three accurate predictions without the twenty that failed. Each shape answers a different question than “is this person skillful under rules R?”
Survivorship also hides in self-selection. People with lucky streaks keep posting; people with cold streaks stop. The public feed becomes a museum of luck plus skill mixed together. Educational posture: treat public highlight reels as marketing objects until a denominator appears.
Educational takeaway: if you cannot see the misses, you are not looking at a track record—you are looking at a trophy shelf.
A base rate is the background frequency of an outcome in a relevant reference class. Before celebrating that someone “called five winners,” ask what fraction of similar calls succeed by chance in that class—and how large the attempt set was. “Five winners” without “out of how many attempts?” is incomplete journalism.
Reference-class literacy matters. Calling the direction of a broad index over one year is a different base-rate problem from calling twenty microcap names over one month. Copying a manager’s top holding after a lagged holdings snapshot (such as a Form 13F quarter-end picture) is a different problem from scoring a dated, direction-and-horizon forecast. Do not borrow a base rate from the wrong class.
Minimum sample-size humility: tiny samples swing wildly. A 80% hit rate on five calls is a different statistical object from a 55% hit rate on two hundred calls under the same definitions. Education does not invent a universal magic N for every domain—but it does refuse to treat N=3 as a career proof.
Ask aloud: “Out of how many forecasts or picks did these winners come?” “How many peers tried the same style and disappeared?” “What fraction of the starting list still exists?” If nobody can answer, your confidence should fall—not rise.
Hit rate counts how often a binary call was correct under a definition. Magnitude asks how large the wins and losses were. A high hit rate with tiny wins and rare huge losses can still destroy a portfolio story; a modest hit rate with asymmetric wins can look worse on a “percent correct” poster while being a different object entirely. Write which scoreboard you are using.
Other misleading boards: (1) accuracy without a horizon; (2) accuracy without a benchmark; (3) “I was early” excuses without a pre-committed window; (4) counting easy calls that almost everyone got right; (5) restarting the scoreboard after a bad streak. Calibration literacy adds another layer—stated confidence should match hit rates in buckets—see the forecast-calibration guide.
Cherry-picking windows is a cousin of survivorship. Starting the chart the day before a lucky streak, or ending it before a drawdown, creates a custom museum. Demand the window rule before the pretty equity curve.
Banned conclusion: “High hit rate means copy this person.” Allowed conclusion: “Under definition D, window W, and sample N, hit rate was H and magnitude summary was M; missing losers and unresolved cases are listed; this is not a trade tip.”
Big-fund holdings snapshots (often associated with Form 13F for many U.S. institutional managers) are lagged quarter-end pictures with amendments, share-class traps, and coverage limits. They are not live blotters of genius trades. Building a gallery of famous names’ current winners without the funds that closed, the positions that were trimmed, or the lag clock is a survivorship-shaped collage.
Insider-trade headlines create similar galleries: a cluster of buys without routine sales, plan sales, or tiny purchases relative to holdings. Falsification habits—asking what would weaken the story—belong beside any “insiders know” caption. See the falsify-smart-money-thesis guide.
Comparing managers fairly requires keeping failed names and closed books in view when the question is skill across a peer set. See the comparing-managers guide’s checklist spirit: side-by-side research that refuses silent deletion.
Educational posture: ownership disclosures are evidence objects with clocks. They become tip culture when losers and lag are cropped out of the frame.
InsightMeter’s educational stance is that public predictions should be scored under explicit horizons, universes, and benchmarks—not remembered as vibes. Methodology language emphasizes sample size, survivorship handling, and definition risk. The product does not claim to erase bias; it aims to make rules visible so readers can see what entered the score.
Practical anti-survivorship habits in this style of scoring include: fixing definitions before looking; keeping unresolved or delisted outcomes from silently vanishing when methodology says so; separating backtest filters from live scoring (see the backtest-versus-live guide); and documenting what would count as a miss. Soft Free/Pro sample lengths are documentation depth—not secret alpha.
Scorecards still need reader humility. A leaderboard number without the rules is another trophy shelf. Pair this article with reading-forecast-scorecards and evaluating-forecast-track-records so denominators stay attached to percentages.
Soft product note: explore Free teasers to practice vocabulary; Pro (optional) unlocks longer history and CSV on stated research windows—never because a scoreboard guarantees future returns.
Illustrative only — not real people and not a recommendation. Suppose a fictional newsletter “North Peak Letters” posts a collage: “8 of our last 10 stock picks won.” A careful student asks for the denominator. The publisher’s archive (fictional) shows 40 picks in the same window under the same written rules; the collage selected the prettiest ten. Separately, 6 of the 40 names delisted or stopped being mentioned mid-window.
Careful educational sentences: (1) “The collage shows 8/10; the archive shows a larger attempt set—out of how many matters.” (2) “Delisted or dropped names must be handled under a pre-written rule, not silence.” (3) “Hit rate without magnitude and without horizon is incomplete.” Banned sentences: “8/10—copy the next pick.” “Smart money newsletter proven.”
Binder fields: claimed scoreboard; full attempt N; shown n; horizon; benchmark; treatment of unresolved/delisted; magnitude note if any; posture “track-record literacy, no recommendation.”
Optional pairing: if the same week a social thread shows a famous fund’s holding from a lagged snapshot, keep that ownership row separate from the newsletter scoreboard—two different evidence objects, two different clocks.
Use this checklist before trusting any “winning track record” collage.
Checklists are educational process tools. They do not create profitable trades.
Most track-record errors are missing denominators wearing confidence.
Fixing them makes media reading slower and much clearer.
Short answers for common track-record questions. Educational only—not investment advice.
Q: If someone shows a long list of winners, isn’t that enough?
A: Not for education. Ask for the attempt set, the rules, the misses, and the window. Winners without denominators are marketing.
Q: What is a “good enough” sample size?
A: There is no single tip-sheet number for every domain. The educational habit is to distrust tiny N as career proof and to write N beside every percentage.
Q: Does InsightMeter remove survivorship bias completely?
A: No product can claim omniscience. InsightMeter-style scoring aims to make definitions and survivorship handling explicit in methodology and scorecards so readers can see the rules—not to promise bias-free alpha.
Q: Are big-fund holdings proof of skill?
A: Not by themselves. Lagged snapshots, peer survivorship, and missing losers complicate skill stories. Treat them as ownership evidence with clocks—not genius certificates.
Q: Should I fade every public track record because of bias?
A: This site’s educational answer is not a fade tip either. Ask better questions, demand denominators, and refuse both blind copying and blind contrarianism dressed as certainty.
Even careful anti-survivorship worksheets cannot recover every quiet failure or private book. Public data are incomplete by nature.
This article is educational process guidance—not a ranking of gurus, not a recommendation to follow or fade anyone, and not a trading system.
A clean scoreboard is not a buy recommendation; a messy scoreboard is not a personalized short recommendation.
Nothing here recommends buying or selling any security based on track-record literacy habits.
Assign the North Peak Letters fiction. Require attempt N, shown n, horizon, and delist treatment before any sentence that uses the word “proven.” Copy-trading recommendations fail the drill.
Have students rewrite a social collage caption to include “out of how many?” in the first line.
Pair with calibration habits so confidence adjectives face sample-size pressure.
Use scorecards and methodology pages to read definitions beside percentages. Screens accelerate documentation; they do not excuse missing denominators.
Soft CTA: after one clean checklist on a public collage, you may create a free account (utm_campaign=checklist30s) to practice vocabulary—Pro longer history + CSV remain optional.
Keep ownership snapshots, forecast scorecards, and tip-thread collages in separate notebook sections—even when they mention the same ticker.
Which market forecasters actually get it right? · Does a forecaster’s stated confidence mean anything? · Backtest vs live forecast scoring pitfalls · How to falsify a “smart money” thesis · How to read a forecast scorecard fairly
Treat winning track records as incomplete until denominators, horizons, benchmarks, and missing-loser rules are written down. Survivorship bias and cherry-picked windows inflate skill stories; base rates and sample-size humility shrink false certainty. InsightMeter-style scoring tries to keep definitions visible—not to promise bias-free profits. Soft Free teasers versus Pro longer history + CSV are documentation choices—not performance promises. Continue with forecast track records, calibration, backtest-versus-live, falsifying smart-money theses, scorecards, methodology, and glossary. Nothing here recommends buying or selling any security.
Learn by topic: When insiders buy or sell · Official company news filings · Private fundraising notices · What big funds own