How It Works
We count every word Trump says. Markets bet on his words. We compare the two.
What is Trump-O-Meter?
Trump-O-Meter analyzes every public speech and social post by Donald Trump, counting every word he uses. Meanwhile, prediction markets like Polymarket and Kalshi let people bet on which words Trump will say next. We connect the two: his actual word frequency versus what markets think he'll say.
Where do speeches come from?
We aggregate transcripts from multiple publicly available sources, including official government publications and professional transcriptions. Coverage includes rallies, press conferences, bilateral meetings, interviews, and other public remarks. All transcripts come from established sources — never AI-generated.
How is word counting done?
Our proprietary speech analysis engine processes each transcript through multiple stages of natural language processing — tokenization, base-form reduction, and intelligent filtering. Words like "taxes" and "tax" are recognized as the same word, while filler words like "the", "and", "is" are removed. The result is a precise count of every meaningful word Trump uses.
You can explore the full vocabulary breakdown on the vocabulary page, where words are sized by frequency and colored by type (noun, verb, adjective, etc.).
A full example
Say Trump says: "The farmers are farming beautiful farms and they won't stop"
Here's what our engine produces:
| Word | Analysis | Base form | Result |
|---|---|---|---|
| The | common word — skipped | — | — |
| farmers | → base form | farmer | farmer: 1 (noun) |
| are | common word — skipped | — | — |
| farming | → base form | farm | farm: 1 (verb) |
| beautiful | → base form | beautiful | beautiful: 1 (adjective) |
| farms | → base form | farm | farm: 2 (noun) |
| and | common word — skipped | — | — |
| they | common word — skipped | — | — |
| won't | contraction — skipped | — | — |
| stop | → base form | stop | stop: 1 (verb) |
Final count: farm: 2, farmer: 1, beautiful: 1, stop: 1. Colors match what you see in the word cloud.
Market-grade precision: farm ≠ farmer
Our engine uses dictionary-aware reduction — not simple root-stripping. This is a critical distinction when real money is on the line:
- farm / farms / farming / farmed
- All recognized as farm — same dictionary entry, counted together.
- farmer / farmers
- Recognized as farmer — a different word entirely. Never conflated with "farm".
Prediction markets resolve on exact words. A market on "farm" won't accidentally trigger from "farmer" — our engine handles this correctly by design. The same precision applies to contractions, possessives, and other edge cases that simpler word-counters get wrong.
What do the numbers mean?
Each word page shows several statistics based on historical data:
- Weekly hit rate
- The percentage of recent weeks where Trump used this word at least once. A 73% hit rate means Trump said the word in roughly 3 out of every 4 weeks.
- Avg. mentions per week
- The average number of times this word appeared per week across the analyzed period.
- Trend
- How recent usage compares to the historical average. A trend above 1.0 means Trump is using the word more than usual lately.
- Last seen
- The most recent week where this word appeared in a speech or post.
- Peak week
- The single week with the highest count for this word.
What are prediction markets?
Prediction markets are platforms where people trade contracts based on future events. On platforms like Polymarket and Kalshi, you can bet on whether Trump will say specific words in his next speech.
For example, a market might ask: "Will Trump say 'tariff' in his next public remarks?" If the market price is 85 cents, that means traders collectively estimate an 85% chance he'll say it.
What is base rate and gap?
Base rate is how often Trump historically says a word, expressed as a probability. If Trump said "tariff" in 73% of recent weeks, the base rate is 73%.
Gap is the difference between what markets predict and what recent history suggests. If a market prices "tariff" at 90% but the base rate is 73%, there is a 17-point gap. A gap is a research signal, not a trading recommendation.
You can explore these on the market signals page.
Why are some words grouped together?
Prediction markets sometimes create a single bet covering multiple related words. For example, a market might bet on whether Trump says "NATO", "friend", or "ally" in a speech — any of those words would count.
We track each word individually so you can see which one he actually uses, but the market sidebar shows how they're grouped for betting purposes. You can compare words side by side using the comparison view.
What's the difference between Speech and Post?
Speech data comes from official transcripts of rallies, press conferences, bilateral meetings, and other public remarks.
Post data comes from Trump's Truth Social account (live, updated continuously) and an archived dataset of his Twitter posts from 2009 to January 2021 (read-only history).
Some prediction markets specify which source counts for resolution — the badge on each market card tells you whether it tracks speeches, posts, or both. You can filter the chart by source using the toggle on any word page.
How often is data updated?
New transcripts are ingested as they become available — typically within hours of a speech. Truth Social posts are tracked continuously. Prediction market data syncs every few minutes from Polymarket and Kalshi APIs.