Analysis
Why Prediction Markets Are Stuck at a Local Maxima
Polymarket, Kalshi, and the illusion of critical mass in on-chain prediction markets
T
Tim.0xAnalysis
Traditional betting platforms will continue to surpass on-chain prediction markets in absolute scale.
The real question is why, how, and for how long.
Polymarket and Kalshi have recorded over $41B and $25B in notional volume respectively — figures that dwarf most other prediction market platforms.
Data source: https://dune.com/datadashboards/prediction-markets
Since the U.S. elections, Polymarket — while indirectly creating room for Kalshi’s growth — has attracted outsized attention both within and outside Crypto Twitter. That attention signals early success, but it is success at the lowest peak.
Despite achieving product–market fit (PMF), this moment is neither the ceiling nor the destination for prediction markets.
0xsmac frames prediction markets as having reached a local maxima due to early success — a phrase I will deliberately overuse throughout this article to underline that this is not an isolated observation, but a confirmation of Smac’s thesis.
“Success is very intoxicating. It is very difficult to handle all the fame and adulation…
You start to believe that everybody wants you, that everybody is thinking of you all the time.”
— Two Kites Dancing in a Hurricane, 0xsmac
When we talk about unicorn outcomes, PMF cannot be separated from timing, market discovery, and the attainment of critical mass. The last of these is the most deceptive.
Apparent critical mass at a local maxima is often misread as terminal PMF. History is littered with examples — Nokia being the canonical one — where dominance masked fragility.
This is where Polymarket and Kalshi appear to stand today.
Critical mass is not binary. It can be real, or it can be illusory. Distinguishing between the two determines whether a market compounds or stalls.
False critical mass most commonly appears when:
The first point is obvious. The second deserves scrutiny.
Using notional volume as a success metric is deeply misleading.
Not only can it be gamed — it is mechanically easy to inflate.
By atomization, I mean repeated buy–sell sequences executed back and forth to manufacture the appearance of activity without introducing real liquidity or risk transfer.
Just as Jez highlighted here, a more practical way to confirm this came from DataDash:
A transaction shows a Polymarket user spending $1 to purchase 510 shares.
Polymarket reports 510 shares traded, which will be recorded as $510 in notional volume — despite only $1 of real value changing hands.
This demonstrates how notional volume can dramatically overstate true activity.
Polymarket and Kalshi are currently overvalued due to deceptive metrics and a misinterpreted sense of critical mass.
While Polymarket incentivizes tighter spreads through trader incentives
(see rewards), spread discipline alone cannot manufacture depth.
Liquidity is structural, not cosmetic.
When a $2 trade can materially move a market, that is not price discovery — it is fragility.
If a token with a $2B FDV can collapse from a $150k short, the valuation was never real.
The same dynamic applies here.
Market making itself has evolved.
The shift from “3.0” to “4.0” reflects a move away from exploiting price dislocations and toward strategies centered on funding rates, open interest, and positioning
(context).
High open interest may provide an entry point for market makers in prediction markets — but whether this becomes sustainably profitable remains an open question.
Kalshi even offers a revealing case study. Its disclosures indirectly highlight how internal market making can be unprofitable, despite early-mover advantages
(Yahoo Finance).
That should not be the case — and it’s a red flag.
Prediction markets do not yet compete directly with sportsbooks — not because they can’t, but because they don’t need to.
That separation may only hold until local maxima is fully explored.
The most important missing primitive is parlays.
Kalshi has introduced them. Polymarket has as well — but with structural flaws.
Polymarket’s reliance on UMA for resolution has repeatedly surfaced as a bottleneck, most notably in the
Zelenskyy suit market.
Polymarket parlays are conditional rather than stacked, unlike traditional sportsbooks where outcomes compound cleanly. This dramatically increases resolution ambiguity and dispute risk.
Real-world events are messy.
Polymarket continues to struggle with non-exhaustive outcomes, as seen in the Time 2025 Person of the Year market, where “Architects of AI” won — but “Other” resolved.
This outcome ambiguity will persist until markets become structurally exhaustive.
One promising direction is synthetic outcome modeling.
Tulip outlines a compelling framework for how Polymarket could evolve beyond this local maxima
(thread).
Prediction markets are here to stay — but their current peak is not their destination.
Integration into superapps offers distribution, but does not remove structural ceilings.
Trade wisely.
Local maxima are always lower than they appear.