10. Crypto and prediction markets

Political polls (POLL)

POLL lists US political polls by type with date, subject, pollster, sample and result, and opens each one with its margin of error and the running average.

  • 3 min
  • 3 questions
  • Lesson 4 of 4

Why you would care

What do voters say, and how much should you trust one poll? Elections and approval ratings move policy expectations: taxes, regulation, trade, defense spending. Markets price those expectations, and prediction markets (previous lesson) trade them directly. A single poll is noisy; the skill is reading averages and knowing which polls to weigh.

The idea from scratch

A poll asks a sample of people a question and generalizes to the whole population.

What to read:

Sample type
Who was asked: all adults, registered voters (Reg), or likely voters (Likely, the closest to an election result)
Sample size
How many people answered; bigger is more precise
Margin of error
The range the true value probably lies in (often about ±3 points for 1,000 people)
Pollster
Who ran it; methods and track records differ

Polling averages combine many polls to reduce noise; a weighted average gives more weight to recent, larger or better-rated polls.

  1. Question to a sample
  2. Result with margin of error
  3. Average of many polls
  4. Trend over time
  5. Market and policy expectations

See it in Gloom

Gloom screenshot: POLL, Approval tab
POLL, Approval tab: recent approval polls with date, subject, sample type and result; pollster names blurred here. Historical example.

open ittype POLL. Tabs: Approval, Favorability, Generic, Senate, Governor, House. / searches subjects and pollsters; Enter opens a poll with its sample, margin of error, the weighted average and a trend chart; o opens the source. POLL is an official plugin: typing it offers to install it.

  1. Sep 17 JD Vance ... Adults Disapprove 63 · Approve 36: the newest poll.
  2. SAMPLE: Adults, Reg (registered voters), Likely (likely voters).
  3. RESULT: the two main answers; red when the leading answer is negative for the subject.
  4. Several polls on the same day (Aug 24) with different samples give different numbers: same question, different populations.
  5. Decimals (53.3 · 44.4): pollsters that publish weighted figures.

Practice and recap

Try it3 tasks
  1. Two polls on Aug 24 for the same subject: adults 65/33, registered 63/37. Why might they differ? (Different populations, plus normal sampling noise.)
  2. Approve 40 with a ±4 margin: plausible range? (36 to 44.)
  3. Why do likely-voter polls matter most before an election? (They best match who will actually vote.)
Common mistakes4 mistakes
  • Treating a one-poll move inside the margin of error as a change.
  • Comparing polls of different populations.
  • Ignoring the pollster's method and track record.
  • Treating polls as forecasts of the outcome on the day.
Check yourself3 questions
  1. What is the margin of error?
  2. What is the difference between "Reg" and "Likely"?
  3. Why use a polling average?
Answers
  1. The range within which the true value probably lies, given the sample size.
  2. Registered voters vs voters screened as likely to vote.
  3. To reduce the noise of any single poll.
Words in this lesson6 words
poll
A survey of a sample, generalized to a population.
sample
The people actually asked.
margin of error
The likely range around a poll's result.
registered / likely voters
People registered to vote / screened as likely to vote.
polling average
A combination of many polls.
generic ballot
A poll asking which party people would vote for, without candidates.

Educational material about reading market data, not investment advice.