Contract developers, clearing robots, market makers, and institutional users face the same problem every day: how to obtain data the fastest, most accurately, and most interpretable, while keeping costs the lowest? Traditional forecasting oracles seem convenient but bury hidden dangers such as 'used but not new', 'update delays', and 'out-of-control slippage'. The Pull Model, adopted here, emerges as a 'data interaction disruptor', prompting a rethink of all game rules from computing strategies to matching mechanisms, and then to risk management. This article will discuss how Pyth returns 'data control' to application developers and pioneers a new core paradigm for on-chain financial systems.
I. "Push has become a thing of the past": The interactive revolution brought by the pulling model.
In the past, the development logic using oracles was usually illustrated as follows:
Oracles charge based on heartbeat (updated every second or every 15 seconds on average)
Regardless of business scenarios, prices go on-chain.
Developers passively use this data.
The biggest problem with this model is the "mismatch" between costs and actual needs: a single data write consumes gas fees even if there are no transactions and no one places orders. More seriously, in high-volatility markets, old prices may become sources of systemic risk—this means a lot of bad debt hazards for liquidation protocols; for matching protocols, it means serious slippage disputes; for institutional users, it means misjudging the generated strategy losses.
Pyth's pulling model completely solves this problem:
Cost visibility: All data on-chain is uniquely triggered by a specific transaction. Whoever calls the data pays the fee; if there’s no call, there’s no data writing cost.
Timing precision: You define when to get new data rather than allowing the oracle to push it at set times. Mandate pulling new prices before matching, opening accounts, and liquidations to reduce the risk of "settling with old prices."
Transparency and governance: Prices can be accompanied by confidence intervals, age, EMA, and other risk control indicators, allowing all transactions to be based on "decision consistency" data.
This means that developers can precisely control the "writing volume" of data like turning on a faucet, completely freeing themselves from the traditional "heartbeat cycle tax."
II. The three lifelines of risk control: how confidence intervals, age, and EMA are written into contracts.
Under the pulling model, Pyth provides developers and protocols with the three most valuable risk control parameters, which they can use to build rich and flexible risk control strategies:
Confidence Interval: Reflects market pricing discrepancies; the larger the interval, the higher the market volatility and noise. Developers can use it to restrict high-risk transactions, reject high-noise market making, and provide more reasonable risk price adjustments for liquidations and margins.
Price Age: Marks the risk trend of price data decomposing from now. Age thresholds can be pre-written into smart contracts, prohibiting the consumption of outdated data to ensure each transaction uses the "most up-to-date" price possible.
Exponential Moving Average (EMA): Buffers during severe market fluctuations. EMA does not replace the marked price but can provide a more conservative and stable anchor point for decisions in critical scenarios such as liquidations and forced liquidations.
By merging these parameters into contracts, contract developers can easily achieve a more explainable and flexible risk control system. For example:
When the confidence interval exceeds the threshold, the protocol can refuse new positions.
When the price age is too long, users must synchronize updates before trading when initiating redemption.
Use EMA as a reference for liquidation trigger prices to avoid systemic liquidation triggered by momentary price spikes.
This risk control design not only gives the protocol a stronger self-rescue capability against extreme volatility but also becomes key for developers to attract long-term stable users and build a good reputation.
III. The preferred weapon of the "multi-chain era": cost, consistency, and resilience growth.
Today's multi-chain parallel situation in DeFi brings expansion dividends but also gives rise to huge governance and operational challenges. For oracle systems, ensuring price consistency and reducing system costs in a multi-chain environment often becomes a nightmare for developers and liquidation teams.
Pyth's pulling model was born in response to the multi-chain era:
On-demand writing, significantly throttled: data writing is initiated only when there is a demand for calls on each chain, reducing redundant costs and coordination pressure caused by cross-chain heartbeats.
Consistency and resilience: The price source of multi-chain deployments uniformly comes from Pythnet, encapsulating different chains to achieve "final result consistency." Developers can set uniform parameters such as age and confidence for risk control across chains, and automatically trigger pause/read-only mode when cross-chain pricing inconsistencies occur.
In conjunction with Pyth's unique "delayed consistency panel" (monitoring the median and p99 latency from chain pulling to on-chain), developers can monitor and make transparent the performance of data consistency across chains in real-time, achieving efficient governance of "bad chain degradation, good chain priority."
IV. Case analysis: Pyth's innovative gameplay in DEXs and lending protocols.
The following are two typical cases where developers utilize Pyth's pulling model and risk control parameters to reconstruct system experiences.
Case 1: Implementation of perpetual DEX.
The new generation of perpetual contract protocols can achieve more efficient matching through the design of "mandatory pulling + updating at the moment of transaction," avoiding data latency and on-chain mining arbitrage.
When users place orders, the transaction first pulls the latest price, then writes the order; if the update fails, the transaction rolls back.
Market making/RFQ displays quote "toxicity" through confidence intervals, attracting liquidity market making.
In times of extreme volatility, use EMA as a conservative anchor point for margin and liquidation price synthesis.
Market sentiment can be quantified (confidence/prive real-time dashboard), significantly reducing matching delays.
This design significantly reduces platform transaction slippage, greatly increases the enthusiasm of extreme market makers, and notably decreases user disputes and operational interventions.
Case 2: Institutional-level lending market.
Based on the complex risk control needs of advanced industry users, the lending market can utilize Pyth's multi-layer risk control indicators for refined management:
The trigger price still uses the marked price, but dynamically adjusts the discount/feeding ratio based on the confidence interval.
When the price age exceeds the threshold, new positions are restricted, or queuing to unlock is processed only when the new price is reached.
Short selling and liquidation paths support batch multi-asset updates; if a single liquidation fails, the entire transaction rolls back.
The interest rate model adapts in real-time based on EMA references.
These transformations reduce bad debts, make liquidation efficient and transparent, and allow for explainable user risk management, suitable for compliance-oriented institutions.
V. Enter governance:$PYTH Coexistence growth mechanism of economy and developer ecosystem.
#PythRoadmap 's long-term evolution relies on $PYTH
The full design of the token economic mechanism:
Data supply incentives: Data sources and publishers need to stake$PYTH , receiving incentives based on data quality, response speed, and coverage breadth; this design improves the stability of data supply and prevents centralization risks.
Governance participation drive: DAO communities participate in category updates, fee rate formulation, and risk strategy adjustments through staking and voting, providing ecological subsidies to protocols with deep usage and ecosystem co-construction, forming positive feedback incentives.
Revenue-sharing mechanism: Pyth's institutional subscriptions (Pro) and blockchain protocol aggregation fee income are directly used for buybacks, burns, and governance pool distributions, giving $PYTH tokens continuous value support capability.
This innovation in governance and economic models allows developers, protocols, and institutional users to feel the benefits of deeply coordinated governance rights and revenue rights from multiple dimensions, strengthening the stickiness and contribution of all parties within the ecosystem.
VI. Conclusion: Pyth opens a new era of decentralized data for funders.
Just by looking back at the past five years, one can understand that data has become the "oxygen" of financial markets—whoever controls the data, controls pricing, and thus owns market pricing rights. The "pulling model" and risk three-parameter system that @Pyth Network insists on and continuously optimizes not only allows developers to regain control over data and risk management capabilities but also accelerates the bridging of the data gap between on-chain and off-chain.
As DeFi develops towards cross-chain, contract complexity, and deep interoperability between on-chain and off-chain, $PYTH has the opportunity to grow into the core framework connecting global digital assets and institutional financial data. The next generation of decentralized financial systems will be depicted in this image—a wealth diverter infusing "instant, controllable, transparent, low-cost equal data water" into all market participants seeking stability and innovation.
#PythRoadmap 's story may have just opened a new chapter.