使用時間序列方法提升對預言機操作價格攻擊的抵抗力

Abstract

自從 Aave 在 2018 年引入閃電貸功能後，盡管提高了 DeFi 上資產的流動性，但也吸引了許多駭客利用此功能進行價格操縱並套利。到了 2024 年，這類攻擊仍屢見不鮮。本文設計了 ATR_AWMA(Average True Range_Adaptive Weighted Moving Average) 演算法，旨在為項目方在報價時提供額外參考和保護。我們對該算法進行了模擬回測，使用了過去（2020 年）的數據，因為當時發生過一起重大的操作價格攻擊，和近期數據，因為近期也同樣發生了重大的操作價格攻擊，並和前人的論文結果做比較，取得了顯著成果。結果表明，ATR_AWMA 能穩定預測趨勢，防止價格被不當操縱。

Since Aave introduced flash loans in 2018, while increasing asset liquidity in DeFi, it has also attracted hackers exploiting this feature for price manipulation and arbitrage. By 2024, such attacks remain prevalent. This paper presents the ATR_AWMA(Average True Range_Adaptive Weighted Moving Average) algorithm, designed to provide additional reference and protection for projects during price quotation. We conducted backtesting using both historical (2020), because a significant price manipulation attack occurred at that time and recent data, because there is also a significant price manipulation attack occurred recently, comparing our results with previous studies. The findings indicate that ATR_AWMA can reliably predict trends and prevent improper price manipulation caomparing to some traditional methods