Who called it right?
Tech analysts forecast markets for a living. We track the calls they make in public, check each one against what actually happened and rank them on how often they get it right
- 0
- Analysts ranked
- 2
- Predictions scored
- 11
- Still to fall due
- –
- Average accuracy
Five years ▾
At least 33% of MSPs will be using generative AI to build their own tools and automation processes by 2025.Robin Ody · made Feb 2024 · No survey isolates generative AI used to build in-house tools, but 2025 surveys show internal AI use near-universal:…
Anthropic is going to IPO itself into a potential multi-trillion dollar value in the upcoming days.Jay McBain · made Jun 2026 · Anthropic had only filed confidentially (1 June 2026).
How we scoreShow
1. Find the call
Public predictions from the last five years in VETTDD’s categories: reports, press releases, keynotes, interviews. Each one is quoted and linked.
2. Wait for the deadline
A prediction is only scored once its date has passed. Until then it’s listed as pending.
3. Check what happened
Against independent evidence where it exists: company results, filings, other firms’ data. Where none exists, we score on the analyst’s firm’s own figures and put a caveat on the verdict. Each verdict links its sources.
4. Weigh it
Calls looking more than two years ahead count one and a half. Calls made under six months out, or when the outcome was close to certain, count half. Trend calls with nothing to measure count half again, so nobody climbs the table by being vague or stating the obvious.
Growth-rate forecasts are scored on how many percentage points they missed by: within 1 point is Right, 2.5 Mostly right, 5 Half right, 10 Mostly wrong; further out, or growth called when the market fell, is Wrong. Other number forecasts are scored on how far off they were: within 10% of the actual figure is Right, within 25% Mostly right; further out, Half right if it moved the right way, otherwise Wrong. Small samples: until an analyst has many verdicts, their accuracy is pulled towards the table average, as if they had 5 more average calls, so a handful of lucky calls can’t top the table. No independent evidence: the verdict is scored on the analyst’s firm’s own figures and carries a caveat. Analysts can reply to any verdict.