VINDICT — The Weighted Hit Rate

01 · what counts as a prediction

Three things have to be true before we score anything.

If a claim has no clear date, no clear outcome, and no source to back it up, it never makes the board.

Clear, dated, and checkableA

It names a specific outcome and a date. You have to be able to prove it wrong — if you can't, we don't score it.

No vibes · no vague maybes

Settled by the public recordB

A call only counts as right or wrong once a trusted, independent source confirms what really happened — never on our word alone.

Settled by the record

Same rules for everyoneC

The same formula and the same cutoffs apply no matter who said it. Every source is shown in full.

Same math · sources shown

Most of what someone says never makes the board. Opinions, vibes, and vague maybes are left out. We only score clear, dated claims we can check against the public record.

02 · the formula

The whole score,
in one fraction.

Every settled call is either right (worth 1) or wrong (worth 0). Each one also carries a weight — how hard the call was to make.

Score = Σ_i w_i·o_i Σ_i w_i × 100

w_i = lead-time multiplier × contrarian multiplier

w_ihow hard the call was

o_i1 if right, 0 if wrong

Σadd up every settled call

03 · the two multipliers

What makes a
call hard.

A call's weight comes from two things we can check: how early it was made, and how much it went against what most people believed at the time.

Lead time× up to 3.0

How early they called it. We measure from the first time they said it on the record to the day it came true. The earlier the call, the more it's worth — but the bonus grows more slowly over time.

How far ahead	Multiplier
Under 1 week	1.0
1 week – 1 month	1.1
1 – 6 months	1.3
6 – 12 months	1.6
1 – 2 years	2.0
2 years or more	2.5 ↑ cap 3.0

The tiers are for reading — under the hood it's a smooth curve

1 + 0.5 · log₂(1 + months / 3), capped at 3

We count from the first time they said it on the record — so they get credit from when they first made the call, not the last.

Going against the crowd× 1.2 – 1.85

How much the call went against what most people believed at the time. Being right when almost everyone expected the opposite earns the most. Calling something everyone already saw coming earns the least.

What most people thought	Multiplier
Agreed with the crowd · ~80% expected it	1.2
A real coin-flip · ~50%	1.5
Went against the crowd · only ~15% expected it	1.85

p = the share of people who expected it at the time

1 + (1 − p)

We use the odds from betting and forecasting sites from back when the call was made. When none existed, we fall back to a set, written-down rating.

PolymarketMetaculusKalshiOur rating

weight=lead time×against the crowd

Multiply the two together and you get the call's weight. A bold, early call that went against the crowd can be worth about 5× a safe, last-minute one. And it works both ways: if a bold call turns out wrong, it pulls the score down just as hard.

04 · worked example

Two real calls.
One hit, one miss.

The same simple math, run on two of Nick Fuentes's predictions — so you can check it yourself.

Nick Fuentes2 settled calls · example

Standard numbers — check the math yourself

	Call 1 — Hit	Call 2 — Miss
The claim	“Trump is the Iran War candidate.”	“Headed for a catastrophic loss” in 2024.
Made	10 Oct 2024	9 Aug 2024
Settled by	US strikes on Iran — 21 Jun 2025	Trump wins — 5 Nov 2024
How early	~8.5 months ahead	~3 months ahead
Early bonus	1.6	1.3
What the crowd thought	Anti-war MAGA · went against it	A close call · near coin-flip
Against-crowd bonus	1.85	1.5
Weight (bonuses ×)	2.96	1.95
Right or wrong	Right · 1	Wrong · 0
Points earned	2.96	0

Score = 2.962.96 + 1.95 × 100 = 2.964.91 × 100 = 60

Simple hit rate · 50

Weighted · 60 ↑

the hit was the harder call

A simple hit rate would call this 50 — one right out of two. The weighting lifts it to 60, because the call they got right was the earlier, bolder, harder one. This is just an example: a real score uses someone's full record, and the numbers above are the standard ones we publish, so anyone can check the math.

05 · the guardrails

The rules that
keep it honest.

Five simple rules that decide what gets scored, what counts as settled, and what we leave blank.

01

Two sources, or it doesn't count We need a dated source for the claim and a separate trusted source for what happened. One source on its own isn't enough to put a call on the board.

02

No credit after the fact A call is judged on the exact thing it predicted. If something related happens later, that's a new call — we never use it to rescue a wrong one.

03

A track record first, then a score With fewer than 2 settled calls, we show only the calls themselves — the dated, sourced receipts — and no overall score yet.

04

We show how sure we are Every score comes with a range that gets tighter as the record grows. A thin record looks thin — we don't make a shaky number look exact.

05

Blank stays blank If there are no sourced calls in a category, we leave it empty — never padded, never guessed, never filled in just to look complete.

Methodology

Three things have to be true before we score anything.

The whole score,
in one fraction.

What makes a
call hard.

Two real calls.
One hit, one miss.

The rules that
keep it honest.

One formula.
Everyone on it.

Methodology

Three things have to be true before we score anything.

The whole score,in one fraction.

What makes acall hard.

Two real calls.One hit, one miss.

The rules thatkeep it honest.

One formula.Everyone on it.

The whole score,
in one fraction.

What makes a
call hard.

Two real calls.
One hit, one miss.

The rules that
keep it honest.

One formula.
Everyone on it.