Binary markets

Solana Binary Prediction Markets

Overview

A binary market is a market with exactly two outcomes — YES or NO — that resolves to one of them at a known future time. Each side is represented as a token (YES token / NO token) that pays out $1 at resolution if the side wins, $0 if it loses. While a market is live, the YES token trades between $0 and $1, and its price is the market's implied probability of YES:

• YES at $0.62 → market thinks YES is 62% likely
• NO at $0.38 → market thinks NO is 38% likely
• YES + NO always sum to ~$1 (minus spread/fees)

Onchain binary markets are useful because they aggregate forecasts into a single, tradeable, oracle-friendly probability that any other smart contract can read. Polymarket popularized them on EVM; in 2025–2026 the action moved to Solana for three reasons (per Helius CEO @mert listing "onchain binary markets, especially for tail assets" as a top-5 Solana priority):

1. Cheap, fast settlement — sub-cent fees and ~400ms finality make low-stakes long-tail markets economically viable, which is impossible on Ethereum L1.
2. Composability — outcome tokens are SPL tokens, so any AMM, lending market, or vault can use them.
3. Capital efficiency — Solana's perp infra (Drift) lets you express a binary view with leverage and cross-collateral, not just spot YES/NO.

Current major Solana venues (verified May 2026)

Venue	Model	Best for	SDK
Drift BET (app.drift.trade/bet)	Perp-style YES/NO with CLOB (price clamped 0–1)	Liquid crypto / macro markets, leverage, cross-collateral	@drift-labs/sdk
DFlow → Kalshi (pond.dflow.net)	Tokenized Kalshi markets (real SPL outcome tokens) routed via DFlow JIT router	Thousands of long-tail real-world markets (sports, politics, weather, econ)	REST API (no SDK needed)
Hxro Parimutuel	Pooled parimutuel — winners split the pool	Short-duration price events (BTC up/down in 5 min), gaming	@hxronetwork/parimutuelsdk

Skip these — they came up in research but are not viable targets:

• Monaco Protocol / BetDEX — SDK repo archived Nov 2025
• Limitless — runs on Base, not Solana
• Polymarket — has not launched a native Solana deployment as of May 2026; the closest thing is Kalshi-on-Solana via DFlow

When to use this skill

Trigger this skill when the user wants to:

• List active YES/NO markets on Solana
• Place a bet / trade on a binary outcome
• Build a bot that prices, makes markets, or arbitrages prediction markets
• Create a new market for a tail-asset event (e.g., "Will memecoin X be >$0.10 on June 30?")
• Pull probabilities from onchain markets to use as an oracle in another contract
• Decide which venue fits their use case

Prerequisites

• Node 20+ (use node --version to check)
• @solana/web3.js — npm i @solana/web3.js@^1.95
• Wallet — a Keypair loaded from a base58 secret or JSON file. Never commit secrets; read from process.env.SOLANA_SECRET or a local ~/.config/solana/id.json.
• USDC on Solana at mint EPjFWdd5AufqSSqeM2qN1xzybapC8GZsdLPi7Ut5GtT (note: outcome tokens on DFlow settle against this mint). Bridge from Ethereum via Wormhole/CCTP or buy on a Solana DEX — do not use the Wormhole-wrapped USDCet, the venues here want canonical Circle USDC.
• RPC endpoint — Helius, QuickNode, or Triton. Public mainnet RPC will be rate-limited.
• For DFlow: wallet must complete Proof KYC before placing real-money Kalshi trades.
• Per-venue SDKs (install only what you need):
  • Drift: npm i @drift-labs/sdk @coral-xyz/anchor
  • Hxro: npm i @hxronetwork/parimutuelsdk
  • DFlow: no SDK needed — REST + @solana/web3.js for signing

Workflow

Core concepts

Outcome tokens. On DFlow/Kalshi, each market mints two SPL tokens: a YES mint and a NO mint. You buy them with USDC at the current market price. At resolution, the winning token redeems 1:1 for USDC; the losing token is worthless. Because they are normal SPL tokens, you can transfer, LP, or borrow against them like any other token.

AMM vs CLOB.

• CLOB (central limit order book) — Drift BET and DFlow use orderbooks. Tighter spreads, but thin books on tail markets mean you need limit orders, not market orders.
• AMM — Some venues use bonding curves (LMSR or constant-product). Prices move continuously with size. Easier liquidity bootstrapping but worse for big tickets.
• Parimutuel — Hxro pools all bets and pays winners pro-rata at expiry. No mid-market trading; you only get the final payout ratio.

Oracle resolution. Markets resolve when an oracle reports the outcome:

• Drift BET — Drift's keeper network resolves to 0 or 1 based on the market's stated source (e.g., a price reference, a publicly attestable event)
• DFlow/Kalshi — Kalshi is a CFTC-regulated exchange and resolves through its own regulated process; you inherit that resolution onchain
• Hxro — settles against the oracle price feed declared at market creation (typically a Pyth price)

Pricing as probability. The YES token's USDC price is the probability. To convert:

• p_yes = yes_price (in USDC, since YES pays $1)
• p_no = 1 - p_yes
• Edge over a "true" model probability p*: edge = p* - p_yes (buy YES if positive)

Per-venue: Drift BET

Listing markets. Drift represents prediction markets as perp markets where contractType == Prediction. Use DriftClient.getPerpMarketAccounts() and filter:

const drift = new DriftClient({ connection, wallet, env: 'mainnet-beta' });
await drift.subscribe();
const allPerps = drift.getPerpMarketAccounts();
const predictionMarkets = allPerps.filter(m => 'prediction' in m.contractType);

The oracle price on a prediction market is the implied probability of YES, clamped to [0, 1]. Long = bet YES, Short = bet NO. See scripts/list-markets.ts for a runnable example.

Placing a bet. Use DriftClient.placePerpOrder with OrderType.LIMIT and a price between 0 and 1. Market orders work but get filled against the book and can be expensive on thin markets — prefer limits. See scripts/place-bet.ts.

Resolution. When the market expires, your perp PnL = (resolution - entry_price) * size for longs. Close the position with a closing order, or let it auto-settle.

Per-venue: DFlow + Kalshi

DFlow exposes a REST API; no SDK is needed. Two base URLs (dev — no key required for testing):

• Metadata: https://dev-prediction-markets-api.dflow.net
• Trade: https://dev-quote-api.dflow.net

Production requires an x-api-key header (contact DFlow).

Listing markets (browse by category):

GET /api/v1/events?seriesTickers=KXEPLGAME&status=active&withNestedMarkets=true

Each market in the response has yesMint, noMint, yesBid, yesAsk, noBid, noAsk.

Buying outcome tokens — call /order with inputMint=USDC, outputMint=<yesMint or noMint>, amount in USDC base units (6 decimals). DFlow returns a base64 transaction; deserialize, sign, and send.

Redeeming. After resolution (result: "yes" or "no", redemptionStatus: "open"), call /order again with inputMint=<winning outcome mint>, outputMint=USDC. Same endpoint handles both buy and redeem — DFlow detects the input.

Tail-asset coverage. Kalshi tokenized "thousands" of markets through DFlow in late 2025; this is currently the broadest long-tail surface area on Solana (sports, politics, weather, econ, entertainment).

Per-venue: Hxro Parimutuel

Best for short-duration price-direction markets:

import { ParimutuelWeb3, MAINNET_CONFIG } from '@hxronetwork/parimutuelsdk';
const pari = new ParimutuelWeb3(MAINNET_CONFIG, connection);
const contests = await pari.getContests(/* market pubkey */);

You stake into LONG or SHORT pools before lockout, the pool settles against the Pyth price at expiry, and winners split the losing side pro-rata to their stake. No mid-market trading — your only choices are stake size and side.

Creating markets for tail assets

• Drift BET — market creation is permissioned to Drift governance; submit a proposal in the Drift Discord/forum with a precise resolution source. Best for assets/events with a clean machine-readable settlement.
• DFlow/Kalshi — market creation goes through Kalshi (regulated). You cannot self-list; you can request a market via Kalshi's market request flow, and once listed it appears in DFlow.
• Hxro — markets are templates parameterized by oracle feed and duration; new templated markets can be deployed permissionlessly if the underlying Pyth feed exists.

For genuinely permissionless tail-asset market creation, none of the major Solana venues are fully open today (May 2026) — this is a real gap and an opportunity. The closest path is Hxro on a new Pyth-listed asset.

Common pitfalls

1. Confusing YES token price with probability when the AMM is skewed. On AMM venues, the displayed "price" can include a curve-dependent skew; the marginal price (what you'd pay for one more token) differs from the average price you actually get for a large order. Always size against the orderbook depth or AMM curve, not the headline number.

2. Slippage on thin markets. Tail markets have $hundreds, not $millions, of depth. A $500 market order can move the price 5–20 cents. Use limit orders at your target probability and walk the book gradually. Set slippageBps explicitly on DFlow's /order (don't trust auto on illiquid markets).

3. Oracle resolution disputes / ambiguity. Read the resolution source before trading. "Will SOL hit $300 by Dec 31" — at which price source, at which timestamp, with what tick? Drift and Kalshi both publish exact resolution criteria; markets where the source is ambiguous have historically taken weeks to resolve and sometimes get voided.

4. USDC bridging confusion. Solana has multiple USDC variants. Canonical Circle USDC is EPjFWdd5AufqSSqeM2qN1xzybapC8GZsdLPi7Ut5GtT. Wormhole-wrapped USDCet (A9mUU4qviSctJVPJdBJWkb28deg915LYJKrzQ19ji3FM) is not accepted by Drift BET or DFlow — you must swap to canonical USDC first (Jupiter handles this in one tx).

5. Market expiry timing. "Expires Dec 31" usually means UTC midnight, not your local time. Drift BET shows expiryTs (unix seconds); DFlow shows closeTime and settleTime separately — the market stops trading at closeTime but pays out at settleTime, which can be hours or days later. Don't park capital in a position you can't redeem promptly.

6. Fee + funding drag (Drift only). Drift prediction markets are perps, so they accrue funding payments between longs and shorts. On a 6-month market, funding can eat 5–10% of your edge. Use the funding rate as a probability check: persistently positive funding means longs are paying to be long → market expects mean-reversion toward NO.

7. Outcome token redemption is not automatic. On DFlow, winning YES/NO tokens sit in your wallet after resolution — you must call the redemption endpoint to convert them to USDC. They won't auto-settle.

References

• scripts/list-markets.ts — fetch and print active Drift BET prediction markets with current YES prices
• scripts/place-bet.ts — place a YES or NO bet on a Drift BET market by symbol or market index
• references/venues-comparison.md — side-by-side venue comparison table (fees, oracles, market creation, tail-asset support, SDK maturity)
• Drift BET docs
• Drift TypeScript SDK
• DFlow Prediction Markets API
• QuickNode guide: Trading Kalshi markets via DFlow on Solana
• Hxro Parimutuel SDK quickstart