While we prepare a thorough root cause analysis of the events that occurred on 9th April 2023, here’s our immediate plan for a robust future for Buffer Finance. 

It’s a long one, grab a cuppa ☕ and let’s get started!

To ensure clarity, we will outline the action plan as the identified issues, and their corresponding solutions.* *

Problem 1: Oracle failure, bad price feeds

Decentralized (price-sensitive) derivatives platforms like Buffer require a specific pull-based Oracle system that necessitates real-time Oracle price updates on-chain. At the time of launch, low latency oracles that could achieve this were still under development. 

While Pyth now offers a solution, Chainlink’s low-latency solution is still under development. 

Hence, we developed an on-demand Oracle system that fetches price feeds from some off-chain oracle, signs it, and stores it along with the signature for the keeper to use for opening and closing the trades.

Although robust, the system’s dependency on second-level data aggregation introduced a single point of failure to the dApp, especially during times of low liquidity/volume.

Solution: Using Pyth Network to Power Decentralized Options Trading

In a recent announcement, we shared our plans to shift 3rd party oracles like Pyth and Chainlink’s low latency oracle solution to power decentralized options trading on Buffer. 

After closely working with Pyth Network, we are able to bring a key infrastructure solution and partner to help Buffer with its critical Oracle-related needs much sooner than expected!

Why Pyth?

Pyth's new low latency data feed offers more granular real-time market data, enabling us to expand existing crypto and forex markets and introduce new ones. Currently, our partial integration with Pyth Network already gives us access to price feeds for a large number of trading pairs across both crypto and non-crypto markets. 

We plan to undertake a full contract integration in V3 while exploring fallback solutions using Chainlink's low-latency oracle.

Forex is back

On April 4th, we halted forex trading on dApp due to bad price feeds. 

Once integration with Pyth Network is complete, we will bring back forex trading on 2 pairs initially with more reliable and resilient price feeds. 

List of all pairs supported on the dApp:

Problem 2: Smaller Trade Sizes, BLP Health

On 4th April, the minimum trade size on aBLP and uBLP pool was raised to 25. This measure was taken to reduce the effectiveness of trading bots using micro position sizing strategies and restore BLP health. 

Note: this decision was not made lightly, as it affected the larger trader base that preferred smaller trade sizes.

Solution: A New 2-Tier Pool System

We are bringing back smaller trade sizes on a new POL-based pool that will allow us to create risk-isolated trading environments and make specific adjustments without affecting every single user at once. 

A 2 tier pool system will also enable us to utilize the POL-based pool as a pilot to evaluate new products on Buffer (especially v3 ) and build confidence before shifting them to the BLP pools. 

Pool Configurations

• POL (protocol-owned liquidity) based 

• Supported timeframes: all timeframes between 3 minutes and 4 hours

• Min trade size: 5 USDC

• Max trade size: 100 USDC

Initially, the pool will support USDC-based trading only. As the pool scales and we have enough data to backtest, a similar ARB-based pool can be launched. 

Fees distribution 

The pool will follow a similar fees distribution to current BLP pools:

• 40% to BFR stakers - higher APRs

• 60% to the treasury - to support pool scalability and act as a last-standing risk mitigation measure

💡 Note: Minimum timeframe for aBLP and uBLP will also start at 3 mins but the minimum trade size will be kept at 25 USD/ARB. We will gradually bring back smaller trade sizes on BLP as the OI increases and the pool recovers. 

Taking Risk Management to the Next Level

Additional interim risk management strategies:

• We will continue running circuit breakers for aBLP and uBLP pools which will automatically halt trading if the pools incur a loss of over 5K USD/ARB from the balance at any point. 

• New dashboards offering real-time open trades and all historical trades.

• We have optimized our internal communication so that the development team can be notified without delay at any time should intervention be needed.

V3 - Our Bluesky Goal

V3 is truly our blue sky goal, with no centralization, no spoke model architecture, and all mesh networks for a more robust, trustless, decentralized future. In light of recent events, we have accelerated Buffer V3 development. As such, details and a roadmap for V3 will be released shortly.

A Note to you!

At Buffer, we don’t defy our people, we believe in accepting and discussing flaws, and that’s the only way to grow. The Buffer community has always been our 🌟 north star in good times and (especially) in challenging ones. 

We have carefully analyzed and implemented your feedback inputs while avoiding subjective feedback loops. At the same time, we never assumed what worked on a small scale would still work on a larger scale. We don’t believe in getting caught up in competitor narratives or implementing models just because they worked for others. It has been inspiring to face all these challenges, and each one has created a more robust ecosystem.

Last but not least, as always, we look forward to hearing your takes on the action plan.