The internet has a returns problem.
Every day it produces thousands of investment claims — strategies, signals, systems, "proven" methods. Almost none of them come with evidence. We're building the place that asks for it.
The loudest claims travel furthest
Social feeds reward confidence, not accuracy. A bold promise spreads; a careful caveat doesn't. So the promises keep getting bolder.
Archetypes, not quotes — but odds are you've scrolled past all six this week.
This isn't cynicism. It's measurement.
People have been counting for decades — funds, traders, forecasters, influencers. The numbers are remarkably consistent:
Professional US large-cap funds that trailed the S&P 500 over the 15 years to 2024. Nine in ten — with research teams and Bloomberg terminals.
Day traders who lost money among those who persisted 300+ sessions in Brazil's futures market. Only 1.1% out-earned minimum wage.
Average accuracy of 68 named gurus, across 6,582 public market calls tracked over eight years. Counted per forecast instead of per guru it's 46.9%. A coin manages 50.
What happens to a trading edge once it's published. Across 97 academic strategies, returns dropped by more than half. Markets eat public edges.
Financial influencers who are measurably anti-skilled — following their advice costs about 2.3% a month. The kicker: the anti-skilled ones have more followers than the skilled ones. The algorithm can't tell the difference. Neither can your feed.
If nine in ten professionals can't beat a simple index fund over 15 years, the stranger with the rented Lamborghini probably can't either.
Why the nonsense keeps winning
Not because everyone is lying. Because the system that surfaces claims never checks them — and three quiet forces do the rest.
Survivorship bias
You only hear from winners. Start 1,000 accounts and luck alone makes a handful look brilliant. They post screenshots; the rest go quiet. The screenshot is real. The odds aren't.
Backtest alchemy
Test enough strategies on the past and one fits it perfectly — by accident. That's the one that gets packaged and sold. Out of sample, published edges lose over half their returns.4
Upside-down incentives
When the product is a course, confidence pays better than accuracy. If a strategy truly printed money quietly, it wouldn't need a checkout page.
So we do the unsexy part: we check.
One claim at a time, in public, with our work shown. The method is old-fashioned on purpose:
Catch the claim
Write it down exactly as made. "This strategy returns 1% a day." No strawmen, no paraphrasing it into something easier to knock over.
Ask what would prove it
A real track record. Fees and slippage included. Out-of-sample. Risk-adjusted. All trades — not the highlight reel.
Answer it in public
We publish what we find, whichever way it points. When a claim holds up, we'll be the first to say so. Extraordinary evidence is rare — not impossible.
What you'll never see here
Be there when the checking starts
The first checks are in the works. Leave your email and they'll land in your inbox — evidence, receipts and all.
No spam, no sharing — unsubscribe anytime.
Sources — checked, obviously
- S&P Dow Jones Indices, SPIVA U.S. Scorecard, Year-End 2024 — 89.5% of US large-cap funds underperformed the S&P 500 over the 15-year period ending Dec 2024.
- Chague, De-Losso & Giovannetti (2020), "Day Trading for a Living?" — of Brazilians who day-traded equity futures 300+ sessions (2013–2015 cohort), 97% lost money; 1.1% earned more than minimum wage.
- CXO Advisory Group, "Guru Grades" — 6,582 public US stock-market forecasts by 68 experts, 2005–2012. The source reports two figures with different denominators, and they are not interchangeable: "Terminal accuracy is 46.9%" across all forecasts, and "If we average by guru rather than across all forecasts, terminal accuracy is 47.4%". We show the per-guru figure, because the claim being tested is about the forecasters.
- McLean & Pontiff (2016), "Does Academic Research Destroy Stock Return Predictability?", The Journal of Finance — across 97 published predictors, returns were 26% lower out-of-sample and 58% lower post-publication.
- Kakhbod, Kazempour, Livdan & Schürhoff (2023), "Finfluencers", Swiss Finance Institute — 56% of finfluencers are "antiskilled" (−2.3% monthly abnormal returns when followed); antiskilled finfluencers have more followers than skilled ones.
How we checked. Each of the five figures above was taken from the source named beside it when this page was written, in July 2026, with the population, period and measure kept exactly as the source states them — that's why every line says which funds, which traders, which window. Correction, 8 August 2026: this page previously described 47.4% as the average across all 6,582 forecasts. It is the average across the 68 gurus; per forecast the source reports 46.9%. Corrected, with both figures now named. What these five have not yet had is the line-by-line re-verification we now run on everything we publish, where a second, independent reader opens each source and checks every number against it. That round is queued for this page; when it's done, this note will say so and carry the date. Educational, never advice. Spot something wrong in the meantime? Tell us — we correct in public.