Traders moved $458 million through Polymarket in the week of 2026-08-03, across 19.4 million individual trades. Algorithmic wallets accounted for 72% of weekly counterparty events (both maker and taker sides counted); everyone else split the remaining 28%. The most-traded market was "Will the U.S. invade Iran before 2027?", pulling in $2 million of volume.
Cumulative since November 2022:
+$140M to algorithmic wallets
·
-$82M from active retail.
Cumulative P&L by trader type, Polymarket since November 2022
Bots have captured +$140M; active retail traders are down -$82M.
Zero-sum market: bot gains and retail losses balance to within fees and the smaller wallet types.
Associate Professor of Finance · Knauss School of Business, University of San Diego
I study why markets behave irrationally and how human psychology shapes financial decisions.
My research on prediction markets uses 671M on-chain Polymarket records (222M resolved trades) to measure who trades, who
profits, and where private information appears. The dashboard makes that work continuously visible.
Open to collaboration.
I share the full trade-level dataset with academic researchers and welcome coauthorship proposals
from PhD students and faculty working on prediction-market questions. I also provide referee-style
comments on prediction-market papers in progress. Reach out by email.
Consulting & expert testimony.
Available for litigation, regulatory, and compliance engagements involving prediction-market
microstructure and informed-trading detection. Schedule a 15-minute intro call →
Top 30 Polymarket markets, week of 2026-W32
Rectangle area = weekly USD volume. Color = market category. Hover for full market name.
Color = category
Geopolitics Politics Other Macro / Finance Crypto
Detecting informed trading
A per-event test separates episodic private information from sustained skill. Zero false positives on a real-data placebo. See all six surveillance indices →
Six live DV-PMI Surveillance Indices cover the full market integrity picture: informed trading, insider timing, adverse selection, resolution surprise, wash trading, and concentration risk.
How traders price uncertainty
Human traders systematically overestimate long shots and underestimate near-certainties — even in liquid prediction markets. See the calibration data →
Bloomberg Businessweek's Big Take quotes Della Vedova on where insider trading concentrates
Denitsa Tsekova and Leonardo Nicoletti analyze 34,000 Polymarket trades flagged as potential insider activity: roughly $200 million in flagged volume from January through June 2026, driven by bets on military action in Iran and Venezuela. Winnings were heavily concentrated, with the top 1% of profitable wallets capturing more than half the money; 57% of those wallets were created less than 24 hours before trading. The piece quotes our assessment of where private-information trading concentrates: elections, committee decisions, award shows, ballot measures, and government votes, the markets where a small group knows the outcome first. Their category-level findings match that ordering; financial and economic markets showed the least flagged activity.
The Atlantic quotes Della Vedova on 'outsider trading' after the Paris hair-dryer incident
Kaitlyn Tiffany's July 20 piece examines a new class of prediction-market conduct: winning a bet by changing the world instead of forecasting it. In April a Météo-France sensor near Charles de Gaulle jumped 4 degrees Celsius in twelve minutes; a wallet positioned above the 21-degree strike against an 18-degree consensus collected roughly $20,000, and Météo-France filed a criminal complaint. Polymarket changed its data source but let the winnings stand; France ordered internet providers to block the site on July 16. The article quotes our recommendations: screen for excess accuracy (a wallet betting one song hits No. 1 while betting no on every rival shows confidence the market's prices cannot explain, whether it comes from an inside line or rigged streams), limit bet sizes where results are cheap to rig, and never resolve a market from a single number at a single source. The trader who rigs the sensor never out-forecasts anyone; the same trade-record footprint our insider-timing and resolution-surprise indices flag gives them away.
Bloomberg: Polymarket bets on Iran deal fuel fresh insider-trading fears
Bloomberg cited Josh's research on informed trading in a feature on suspicious Polymarket activity tied to the US-Iran conflict. The article identifies $45 million in flagged trades on Iran-related markets and profiles wallets that bet on a peace deal at 6% odds before the announcement. Quote: "In the political markets, there's constantly information being created. In a market in which you have a small amount of decision makers, information is likely to be able to be leaked out."
On CNBC: bots are outperforming traders in prediction markets
Walked through the headline finding from Paper 1 and how the dashboard makes the bot-vs-retail transfer continuously visible. Watch the full segment on CNBC.com.
One email per week with the updated indices, the “this week on Polymarket” narrative,
and any methodology changes. No pitches. No market recommendations. Unsubscribe in one click.
Your email is used only for the weekly update. Not shared. Not sold.
Joshua Della Vedova · Knauss School of Business, University of San Diego
Updated weekly · 2026-W32
