The Roosevelt Institute Study, Explained — what it found, what it missed, and what Kalshi said back.
Retail losses (RI estimate)
$583.5M
Study period
5 years
Trades analyzed
400M+
Sports share
64%
All figures below are from the Roosevelt Institute's analysis of public Kalshi trading data. Source: rooseveltinstitute.org
$583.5M
Estimated retail losses
July 2021–May 2026
400M+
Trades analyzed
on Kalshi platform
$32B+
Total volume studied
Kalshi lifetime
$371.6M
Sports share of losses
64% of total
$75.20
Mean bet size
average trade
6.3%
Profit concentration
of orders = 63.2% of gains
Report details
The study analyzed more than 400 million trades totaling over $32 billion in volume on Kalshi. It divided traders into two structural categories:
Using this framework, the study found that market takers (retail) lost $583.5 million in aggregate over the study period, with most of that wealth flowing to market makers. Sports contracts alone accounted for $371.6 million of those retail losses — the category that Kalshi markets most aggressively to consumers.
Retail traders are net losers
Market takers — those accepting bets rather than posting them — lost $583.5 million in aggregate over the study period. The losses are especially concentrated in sports contracts, which accounted for roughly two-thirds of total retail losses.
Wins are skewed toward large bets
When retail traders did make money, it was disproportionately from a small number of large wagers. 63.2% of all taker revenue flowed to just 6.3% of matched orders — those with a dollar value of at least $200.
Professional traders capture most gains
Market makers — those posting bet offers — systematically capture the difference. The study frames this as sophisticated traders exploiting an information and automation advantage over casual users who bet based on intuition or public information.
Data transparency concerns arose mid-research
Public Kalshi data on Dune was cut off in May 2026 — a few days after Sportico published analysis showing retail parlay losses of $117M in early 2026. The Roosevelt Institute completed its analysis using data it had already pulled before access was restricted.
Every trading data study has scope limitations. These are worth knowing before drawing conclusions:
Kalshi disputes both the figure and the framing.
On market structure: Kalshi argues it is a financial exchange — not a casino — that matches buyers and sellers without taking the opposite side of any trade. The company says the premise that there is a "hidden house" misunderstands how order-driven exchange markets work.
On the methodology: Kalshi contends the study made a classification error: it used maker/taker order types as proxies for user sophistication. Maker and taker designations describe how an order was executed, not who placed it. According to Kalshi, this caused the study to count some institutional market makers as ordinary users and some casual app users as professionals, distorting the $583.5M figure.
On skill gaps: Kalshi's position is that a difference in trading skill among users does not imply a problem with market structure. Skilled traders outperform on any competitive market — this is not unique to prediction markets.
Kalshi's full official response is available from the company directly. Confirm the latest statement at kalshi.com.
The Roosevelt Institute study did not emerge in isolation. Multiple independent analyses have reached similar conclusions about profit concentration in prediction markets:
The specific $583.5 million figure may be contested. The pattern it reflects — that most ordinary participants lose money while a small number of sophisticated traders capture most gains — is not.
Understanding who you are trading against matters. On Kalshi and Polymarket, you are not competing against a house with fixed odds. You are competing against other market participants — some of whom are professional trading firms with automated systems, proprietary data feeds, and dedicated market-making infrastructure.
This does not mean retail traders cannot win. It means that in any given market, the odds of who is on the other side of your trade are not uniform. In liquid, actively traded markets — especially sports — you are more likely to face a sophisticated counterparty.
The Roosevelt Institute study suggests that thin-margin, high-volume sports betting is where ordinary users fare worst. Niche markets with less professional participation, or markets where you have genuine informational edge, are where retail can realistically compete. The data supports what experienced traders have long argued: selectivity matters more than activity.
The Roosevelt Institute is a nonprofit think tank founded in 1948 and based in New York City. It focuses on corporate power, financial regulation, and progressive economic policy. The lead author of this report, Brad Lipton, is the organization's Director of Corporate Power and Financial Regulation. The report is part of a four-part series titled "The Hidden House: Prediction Markets and How They're Shaping Society."
The full four-part series — The Hidden House — goes beyond retail losses to examine prediction markets' effects on media, finance, sports, and regulatory frameworks. Only Part 1 is analyzed on this page.
What's well-supported
What's disputed
The Roosevelt Institute estimated that retail traders (defined as market takers — those who accept bet offers) lost $583.5 million on Kalshi from its launch in July 2021 through May 2026, across more than 400 million trades totaling over $32 billion in volume. Sports contracts accounted for the largest share of those losses, at $371.6 million.
The figure comes from the Roosevelt Institute's analysis of public Kalshi trading data and is the organization's estimate. Kalshi disputes the methodology, arguing the study misclassified users — labeling institutional market makers as ordinary users and some casual app traders as professionals. Whether the $583.5M figure is accurate depends on how you define 'retail trader,' which is itself the crux of the debate.
Kalshi disputed the study's findings. The company argued there is no 'house' on its platform — it operates as an exchange that matches buyers and sellers rather than taking the opposite side of trades. Kalshi also claimed the study rests on a methodological error that conflates maker/taker order types with user sophistication, leading to the misclassification of who counts as 'retail.'
The evidence strongly suggests that most retail participants lose money over time. The Roosevelt Institute's Kalshi analysis, a Wall Street Journal investigation of Kalshi and Polymarket, and multiple academic studies of Polymarket data all reach similar conclusions: the vast majority of trading gains are captured by a small number of sophisticated users (the top 0.1–1%), while most participants experience net losses.
Can Retail Traders Win?
4 conditions where retail has real edge — and where you're outgunned
How Concentrated Is Prediction-Market Profit?
The data on who captures most of the gains — and by how much
Why Do Prediction Market Traders Lose Money?
The psychological and structural reasons most participants end up negative
Who Makes Money on Prediction Markets?
Market makers, informed traders, and what role retail actually plays
4 common questions answered
Continue exploring
