The last 25% of a market represents a disproportionate share of returns for winning-side positions.
This is where leverage is most useful.
Some PM leverage approaches rely on a global time-based full deleveraging rule to reduce jump risk exposure of their credit capital.
This approach is usually here to compensate for the lack of dynamic risk monitoring across position lifecycles.
Pure time-based deleveraging remove the most valuable opportunity surface for traders.
With Dimes’ J-factor:
~ 85%+ of winning-side positions carry leverage to resolution
~ 60% of winning-side returns have been achieved on the levered portion of the last 25% of a market’s duration
Two days after launching on @arespro, Dimes has already enabled leverage across 300+ markets spanning sports and crypto. And this is just the beginning.
Levered @Polymarket is on for the World Cup.
Why margin never came onchain
Most CeFi venues run spot margin on 100+ assets.
A decade into DeFi, spot margin is still unsolved for 95% of assets. Why?
Background
A margin desk at a CeFi venue runs 4 things simultaneously:
1/ a risk model that reprices collateral continuously against depth, vol, etc;
2/ a liquidation path with privileged execution inside the venue;
3/ hedging inventory across the venue's own derivatives and spot books that can net residual exposure internally;
4/ and a balance sheet that absorbs shortfalls when the other three misprice the tail.
Each one compensates for imperfections in the others.
A slower risk model is survivable if liquidation execution is fast and hedging is cheap. Thin hedging is survivable if the balance sheet is deep. And so forth.
Onchain, until recently, all four were structurally constrained at the same time for spot margin.
The risk modeling side was the most visible gap. Onchain lending protocols operate on a narrow per-asset parameter space (LTV, liquidation threshold, caps, liquidation bonus) set within a governance-ratified framework, against asset universes of typically 5 to 15 blue chips. Building a continuously-updated risk model across a wider asset base was tractable in principle, but required a real-time data pipeline and the will to stand up a broad live risk infrastructure.
The liquidation path was constrained by execution non-determinism. Onchain liquidations mainly run through third-party keeper bots competing in the public mempool, which means inclusion probability degrades under exactly the network conditions where liquidations matter most. Protocols either compensated by internalizing sequencing (Hyperliquid, dYdX v4) or by widening maintenance margins, which lowered effective leverage and pushed the product away from the regime where margin is economically useful. Recent work such as BAM on Solana and Ethereum's based-rollup preconfirmation stack begin to solve this by offering non-venue protocols guaranteed transaction inclusion, removing the dependency on third-party keeper participation at exactly the moment it matters most.
The hedging venue question is the one most people underestimate. CeFi margin books net internally because the venue hosts both the collateral market and the hedging market, and the balance sheet clears both. Onchain, a risk engine has to hedge into external venues, and until recently those venues either weren't deep enough, weren't programmatically addressable, or both.
Credit to sit behind the book is a consequence of the first three and will improve as the industry matures. Institutional lenders underwrite risk layers they can verify and size against, which requires the other three pieces to first produce an auditable layer with live operating data.
Closing thoughts
In new markets, credit usually arrives last because it requires every other piece of the stack to be legible and reliable enough to underwrite. Onchain followed that pattern. Settlement worked from day one, spot DEXs nearly a decade ago, perps once execution caught up. We believe at @dimes_fi that horizontal margin credit is next.
But once onchain spot margin works, it won't be a port of CeFi margin. A CeFi margin account is locked inside the venue that originates it. Onchain, the risk engine, liquidation path, hedging surface, and credit capital are all separately addressable, meaning margin sits horizontally across venues rather than inside any one of them. Any surface that captures trading intent, specialized terminals, wallets, agents, CLIs, can integrate the same margin layer through a simple API, an unlock that will materially expand onchain trading front-ends' volume and close one of the largest remaining feature gaps with CeFi venues.
We call this Headless Credit.
Composabilty is and will remain DeFi’s superpower.
A few weeks ago, the @BlockworksAdv team published a great piece covering the various strategies available to offer leverage in prediction markets: lending pools, prime brokerage, perps, and CFD-inspired strategies - the model introduced by Dimes.
The merits of the CFD-like approach:
1/ Enables dynamic and sophisticated risk engines
2/ Enables netting across positions, unlocking superior capital efficiency and better cost for traders
3/ Eliminates cross-collateral contagion risk inherent to pooled lending constructs
4/ Provides a flat fee structure, giving traders cost of capital guarantees
https://t.co/XireVNSk3Q
Mexico to win at kickoff: 69c → 1 = 1.45x potential return. Biggest danger: the draw.
South Africa NOT to win: 89c → 1 = 1.12x. Cashes on a Mexico win or a draw.
At 4x leverage: 89c → 1 = 1.49x.
Higher payout. Draw danger gone.
Imagine making 50% in 2 hrs by betting that South Africa will not win.
The edge is here for those who know where to look.
Exclusively powered by @dimes_fi. Now live on @arespro
Most leverage frameworks in prediction markets borrow their risk logic from continuous markets: static LTVs, fixed health thresholds, linear maintenance margin, time-gated deleveragings.
But PMs have frequent, sudden jumps, low liquidity, information asymmetry, noisy market making, and importantly, they resolve.
One-dimensional margin frameworks either leave leverage on the table or underprices risk. An asymmetry that often flips multiple times within a single market's lifecycle.
J-factor is how @dimes_fi deals with this set-up.
J-factor is a continuous composite risk score computed per position to price next-tick crash risk. It was trained on hundreds of thousands of market-hours and understand things like:
- Near-resolution reflexivity, and how it varies by market type
- The behavioral signatures of informed versus uninformed flow
- How liquidity behaves across different market types and setups
From a user's point of view, J-factor does three important things:
1/ it lets the (predicted) winning side carry as much leverage as is manageable all the way to resolution, magnifying traders’ potential returns
2/ proactively delevers positions when they get too risky, making liquidations the exception rather than the norm
3/ generalizes across hundreds of markets, including short-duration ones
The result is a system that lets traders’ upside run, protects capital providers’ downside, and maximizes leverage availability across categories.
Two days after launching on @arespro, Dimes has already enabled leverage across 300+ markets spanning sports and crypto. And this is just the beginning.
Levered @Polymarket is on for the World Cup.
Why Dimes chose institutional credit partners over a public retail vault to fund margin capacity
1️⃣ Guaranteed liquidity for front-ends: partners need predictable capacity their users can rely on. Retail vault liquidity is highly reflexive and expands / contracts with DeFi flows, creating uncertainty when reliability matters most. With Dimes, front-ends know liquidity is here to stay.
2️⃣ Higher operating standards: institutional capital brings deeper scrutiny across our risk models, reporting, controls, and incident processes. Dimes is the only levered prediction markets providers that has withstood the scrutiny of various sophisticated partners across onchain private credit and liquid funds for direct margining operations.
3️⃣ Controlled scaling: capacity grows with live performance, front-end onboarding, and risk-system maturity.
4️⃣ Multi-product extensibility: the same credit relationships can extend across new surfaces, from single-position leverage to portfolio margining, allowing us to expand capabilities alongside our front-end partners.
5️⃣ Best-practice capture: working closely with institutional credit partners allows continuous knowledge transfer on risk management.
Our prediction market leverage architecture is intentionally narrow and specialized:
- We do not create our own markets
- We do not host an orderbook
- We do not compete with PM venues and front-ends
- We do not make favored LPs your counterparty
- We do not attempt to own user acquisition
We sit in the middle: front-ends on one side, PM venues on the other, and supply the liquidity + risk engine that makes safely offering leveraged PM exposure accessible to onchain trading apps.
Sometimes, less is more.
Docs linked below👇
Prediction markets are becoming one of the most important ways people trade information.
Today, Ares is bringing 10x leverage trading for @Polymarket!
Same markets. Bigger conviction sizing.
Powered by @Polymarket, in partnership with @dimes_fi
Trade now: https://t.co/uhdQzw6XAu
Leverage is provided by Dimes and integrated natively into the Ares trading experience.
Trading with leverage magnifies potential losses and increases execution costs. Do not trade more than you can afford to lose. NFA.