How It Works

The basic idea

If you've used platforms like Polymarket or Kalshi, you're already familiar with the core concept. Those platforms let people trade on yes/no questions: "Will Team X win the championship?" or "Will inflation exceed 3% this quarter?" The price of a contract reflects the crowd's collective estimate of the probability. If a "Yes" contract trades at $0.65, the market thinks there's a 65% chance it happens. If you're right, you get $1. If you're wrong, you get nothing.

Dark Markets works the same way, but instead of asking about elections or sports, our markets ask about prices on darknet marketplaces:

"Will the price of cocaine exceed $62/gram on darknet marketplaces tomorrow?"

You can buy "Yes" or "No" just like on Polymarket or Kalshi.

From yes/no questions to a price estimate

A single yes/no question tells you something useful, but it only gives you one data point. To build a complete picture of what the market thinks the price will be, Dark Markets runs 10 binary markets simultaneously, each set at a different price threshold.

Here's how that looks in practice. Say yesterday's verified price for cocaine was $60/gram. Today, 10 markets open, spaced from 5% below to 5% above that anchor:

Reading down that table, you can see the crowd's thinking. There's a 92% chance the price will be above $57 (almost certain), a 53% chance it'll be above $59.67, and only a 6% chance it'll top $63. The point where the probability crosses 50% gives you the market's best estimate of the actual price. In this example, that's roughly $59.80.

Plot that crossover point day after day and you get a continuous price chart, which is the core intelligence product that our institutional subscribers use.

The 10 probabilities also tell you something a single price number can't: how confident the market is. If the probabilities drop sharply from 90% to 10% across just a few thresholds, the market has strong consensus. If they decline gradually across the full range, there's high uncertainty. That confidence signal is valuable intelligence in its own right.

Why use a market at all?

A reasonable question. Why not just scrape darknet marketplace listings and publish averages?

The short answer is that passive observation has real limitations. Listings can be fake. Prices can be stale. Marketplaces go offline. Vendors post decoy listings to mislead competitors. A single data source can give you a badly distorted picture.

A prediction market addresses this because participants have a financial incentive to be accurate. If someone knows the scraped data is wrong, for instance because a major marketplace just exit-scammed and prices are about to spike, they can trade on that knowledge and profit. The market price incorporates information from many sources, including sources that no scraper can reach: private channels, insider knowledge, offline intelligence networks.

This is the same principle that makes stock markets useful. The price of a stock reflects not just publicly available financial statements, but the private analysis, expertise, and sometimes non-public information of all the people trading it.

Step by step: how a daily cycle works

1. Markets open

Each day, 10 new binary markets open for every tracked substance and region. The thresholds are calculated automatically from the previous day's settlement price, spaced evenly from -5% to +5%.

2. Participants trade

Anyone with a verified account can buy "Yes" or "No" contracts in any of the 10 markets. If you believe the price will be higher than $61.00 tomorrow, you buy "Yes" on Market 7. If you think it won't reach $59.00, you buy "No" on Market 4.

Each market uses an automated market maker (AMM), a piece of software that always offers prices to buy and sell at. You can trade at any time without waiting for another person to take the other side. The AMM adjusts its prices as trades come in: more "Yes" buying pushes the probability up, more "No" buying pushes it down.

3. Information flows into prices

As people trade across the 10 markets, the probabilities shift to reflect the crowd's evolving estimate. Someone monitoring Telegram channels who knows about a supply disruption will buy "Yes" on the higher-threshold markets. An academic researcher who has access to listing data showing stable supply will buy "No" on those same markets. A law enforcement analyst who knows a major operation is about to go public might take positions accordingly.

No single participant needs the full picture. The market synthesises partial information from many sources into a set of probabilities that, taken together, produce a price estimate.

4. Ground truth is established

At the end of each 24-hour period, we need to determine what actually happened. What was the real price?

This is handled by the oracle network: a group of 10 independent nodes, operated by academic institutions and vetted research partners, that each independently collect data from 8-10 darknet marketplaces. They take 6 snapshots per day to capture intraday variation.

Each node submits its price observation using a commit-reveal protocol (they commit to their answer before seeing what anyone else submitted, preventing copying or collusion). After all submissions are in, a consensus mechanism filters out outliers and produces a single settlement price.

This design means no single entity controls what the "true" price is. An attacker would need to compromise at least a third of all nodes to influence the result, and even then, statistical filtering would flag anomalous submissions.

5. Markets settle

The oracle's consensus price is compared against each of the 10 thresholds. If the consensus price is $60.50, then Markets 1-6 (thresholds $57.00 through $60.33) all resolve "Yes" and Markets 7-10 (thresholds $61.00 through $63.00) all resolve "No." Holders of the correct side receive $1 per contract. Holders of the wrong side receive nothing.

Settlement is handled automatically by smart contracts, with no manual process or discretion involved.

6. New markets open

Immediately after settlement, 10 new markets open, now anchored to the new settlement price of $60.50. The cycle repeats, producing a continuous stream of daily price intelligence.