Forecasting newsletter for November 2024. Polymarket Total Value locked goes down from $500M to $100M.

The key question this month was whether prediction markets in general and Polymarket in particular would manage to retain the large volume they attracted during the US elections.

This is how total value locked (a) (TVL) looks for Polymarket:

So the exponential rise is pretty sharply cut off, but a fifth remains. I’m curious to see whether it will continue growing from that now higher base, or whether it will attrit.

There are a bunch of potential business models for Polymarket, like sponsors paying to get a market up, making money providing liquidity, advertisements, or some other cool innovative option I haven’t yet thought of. But the most straightforward business model is users paying transaction fees, whether directly or as a spread. As some rough math, if users pay fees of 0.5% to ¹ over TVL, with 100M TVL that’s a cashflow of 500K to 5M/year. With TVL of 500M to 50B that’s a cashflow of 6M to 1B a year.

Index

• Prediction markets and forecasting platforms

  • Polymarket

  • Kalshi

  • Cultivate Labs

  • Manifold Markets

  • Other platforms

• Research and articles

• Odds and ends

Prediction markets and forecasting platforms

Polymarket

Shayne Coplan was raided (a) by the FBI. Coplan frames this as being politically motivated (a).

Polymarket has a few dashboards (a).

French regulators noticed Polymarket after a Frenchman won a large bet on the US elections. Polymarket prophylactically blocked (a) French users.

Elon Musk agreed with some Polymarket odds (a) that the chopsticks would catch Space X’s returning booster.

Kalshi

Pirate Wires reports how Kalshi coordinated social media influencers to portray Shayne Coplan in a negative light after his FBI raid (a). Pirate Wires is variously associated with Polymarket and has plenty of conflicts of interest, but overall I find the reports plausible. This is why Kalshi doesn’t have the mandate of heaven.

Kalshi added USDC as a crypto payment option.

Cultivate Labs

Cultivate Labs started a forecasting tournament (a) for US middle and high schoolers. The Cultivate Labs newsletter mentions a sleek “Analyst Research Companion”, with interested readers instructed to get in touch via email.

Manifold Markets

Manifold looks back (a) at its election coverage (a).

Here (a) is a calibration calculator for Manifold.

Other platforms

brkt.gg (a) (presumably internet slang for “be reckt”, or “be wrecked”, meaning to get destroyed financially) is a crypto prediction market.

playmoney.dev (a) is a prediction market that got started after Manifold added real money options. I like its interface for being simple, not a Skinner box. Reminds me of earlier, more innocent times.

A commenter on Good Judgment Open laid out (a) his case for 72% Harris/28% Trump prediction.

Interactive Brokers reports half a billion traded in election market contracts (a).

DraftKings is aware (a) of Polymarket. To give a sense of their scale, they recently got a $500M (a) credit line from Morgan Stanley and others.

Research and articles

A post on LessWrong looks at how “The Solomonoff Prior is Malign” is a special case of a simpler argument (a).

Eric Neyman writes about Seven lessons I didn’t learn from election day (a). They were: Kamala Harris did not run a bad campaign; polls aren’t useless, they were pretty good, actually; Theo the French Whale did not have edge; we didn’t learn which campaign strategies worked; Donald Trump isn’t a good candidate; spending money on political campaigns isn’t useless; his opinion of the American people didn’t change

Zvi Mowshowitz writes about how The Online Sports Gambling Experiment Has Failed (a), i.e., online sports gambling has proved to be generally harmful and addictive.

The Super Model substack compares how different models would win/lose money if they traded on Polymarket every day: 538 vs. DDHQ (a), Silver vs. the 13 Keys (a), Manifold vs Metaculus (a)

Vitalik Buterin looks at prediction markets and “info finance” (a).

DeepMind announced a much better weather forecasting model (a), and released the code (a) andweights. They also made the code (a) (but not weights) for AlphaFold3 open source

Dan Schwarz writes about The Death and Life of (play money) Prediction Markets at Google (a)

Reuters looks at challenges of weather prediction in Africa

Odds and ends

My friend Eli has an AI 2025 Forecasting survey (a), asking readers to pre-register their expectations for 2025 for a few benchmarks.

The SCOR foundation, the philanthropic arm of a French insurance giant, funded (a) a project (a) to have long-term prediction markets for climate stats (a). I’m not sure how big of a deal this is. It could be a moderately large deal.

Here is a contract for a Dapp developer for a prediction market… on Upwork, for $15-$20/hour. Meanwhile a YC-backed sports prediction markets is paying $125k to $275k/year (a) for a senior backend engineer.

A 2017 market comparing the price of Bitcoin to that of a proposed hard fork (a) is an interesting case study. So is the stock price of United Healthcare going up slightly after its CEO was murdered—it provides a natural experiment for markets for firing CEOs proposed by Robin Hanson.

The next DC forecasting and prediction markets meetup (a) will happen on Dec 11

---

To place a man in a multi-stage rocket and project him into the controlling gravitational field of the moon where the passengers can make scientific observations, perhaps land alive, and then return to earth–all that constitutes a wild dream worthy of Jules Verne. I am bold enough to say that such a man-made voyage will never occur regardless of all future advances.

—Lee deForest (1873-1961) (American radio pioneer and inventor of the vacuum tube.) Feb 25, 1957

Superhuman machine intelligence is prima facie ridiculous

—Many otherwise smart people, 2015

— Sam Altman’s blog (a), 2015, found when looking for sources for the first quote.

---

This newsletter is sponsored by the Open Philanthropy Foundation.

1. Users would normally pay fees as a percentage of each trade, not of total value locked, but volume specifically would be much lower if not incentivized, so I feel that the degree of indirection is worth having.