Online Markets

Dated Oct 4, 2021; last modified on Mon, 05 Sep 2022

WWW ‘21: The Web Conference 2021

REST: Relational Event-Driven Stock Trend Forecasting

REST, an event-driven stock trend forecasting framework, that overcomes two limitations of existing event-driven models. Models the stock context, and learns the effect of event information on the stocks under different contexts. Constructs a stock graph and designs a new propagation layer to propagate the effect of event information from related stocks.

The value of stock trend forecasting is not unanimous, e.g. Malkiel contends that forecasting is a fool’s game , while Simon’s RenTech is all about the math . But it seems like the question is an empirical one, and therefore, a good answer should exist. Why are there opposing camps?

Exploring the Scale-Free Nature of Stock Markets

Most existing neural methods treat stocks as independent of each other. However, financial literature shows stock markets and inter-stock correlations show scale-free network characteristics. The authors modeled the scale-free nature of inter-stock relations through temporal hyperbolic graph learning on Riemannian manifolds that can represent the spatial correlations between stocks more accurately. Stock selection then became a learning to rank problem, and the authors outperformed current forecasters.

Scale-free networks have the property that degree distribution follows a power law. Most notably is the relative commonness of “hubs”, vertices whose degree greatly exceeds the average.

The phrase “inter-stock relations through temporal hyperbolic graph learning on Riemann manifolds” is quite the mouthful. The key here seems to be, “How do we model a group of items whose relational distances vary over time?”

Hyperbolic learning is a shift from the more traditional learning that occurs in Euclidean spaces. Hyperbolic space is apt for tree data, while spherical space is apt for cyclic data. AFAICT, Riemannian manifolds are topological structures that are convenient when working with hyperbolic space. also mentions “Riemannian metric”.

is a bit dense. Digging deeper is not worth it as I don’t have a problem to solve/relate my learnings to. Maybe knowing that they exist is enough for now.

Detecting and Quantifying Wash Trading on Decentralized Cryptocurrency Exchanges

show how to identify wash trading activity in IDEX and EtherDelta, two popular decentralized exchanges on the Ethereum blockchain. They identified a lower bound accounts and trading structures engaged in wash trading. Found that 30% of all traded tokens haven been subject to wash trading at some point, with an emphasis on EtherDelta where 10% of tokens have almost exclusively been wash traded.

In wash trading, an investor simultaneously sells and buys the same financial instruments. Objectives: to artificially increase trading volume to suggest that the instrument is in demand; to generate commission fees to compensate brokers for something that cannot be openly paid for.

Towards Understanding and Demystifying Bitcoin Mixing Services

aim to understand Bitcoin mixing services as they have been used to facilitate criminal activities. Swapping and obfuscating are the most popular mixing strategies. The authors propose a transaction based analysis that successfully reveal the mixing mechanisms for 4 representative mixers. They further propose a method that identifies 92% of the mixing transactions that used obfuscation.

A cryptocurrency tumbler or cryptocurrency mixing service is a service offered to fix potentially identifiable/tainted cryptocurrency funds with others, so as to obscure the trail back to the fund’s original source.

Towards Understanding Cryptocurrency Derivatives

Crypto trading has evolved from a collection of spot markets (fiar for cryptocurrency) to a hybrid ecosystem features complex and popular derivatives. BitMEX is a market leader with +3B USD of volume per day, and allow users up to 100x leverage. Authors analyzed the evolution of BitMEX, the diverse ensemble of amateur and professional traders, and how it has led to dramatic price movements in the underlying spot markets.

References

  1. REST: Relational Event-Driven Stock Trend Forecasting. Xu, Wentao; Liu, Weiqing; Xu, Chang; Bian, Jiang; Yin, Jian; Liu, Tie-Yan. The Web Conference, 2021. doi.org . 2021. ISBN: 9781450383127 .
  2. Exploring the Scale-Free Nature of Stock Markets: Hyperbolic Graph Learning for Algorithmic Trading. Sawhney, Ramit; Agarwal, Shivam; Wadhwa, Arnav; Shah, Rajiv. The Web Conference, 2021. doi.org . 2021. ISBN: 9781450383127 .
  3. Scale-free Network. en.wikipedia.org . Accessed Oct 4, 2021.
  4. Into the Wild: Machine Learning In Non-Euclidean Spaces. Fred Sala; Ines Chami; Adva Wolf; Albert Gu; Beliz Gunel; Chris Ré. dawn.cs.stanford.edu . Oct 10, 2019. Accessed Oct 4, 2021.
  5. Riemannian Manifold. en.wikipedia.org . Accessed Oct 4, 2021.
  6. Detecting and Quantifying Wash Trading on Decentralized Cryptocurrency Exchanges. Victor, Friedhelm; Weintraud, Andrea Marie. The Web Conference, 2021. doi.org . 2021. ISBN: 9781450383127 .
  7. Wash Trade. en.wikipedia.org . Accessed Oct 4, 2021.
  8. Towards Understanding and Demystifying Bitcoin Mixing Services. Wu, Lei; Hu, Yufeng; Zhou, Yajin; Wang, Haoyu; Luo, Xiapu; Wang, Zhi; Zhang, Fan; Ren, Kui. The Web Conference, 2021. doi.org . 2021. ISBN: 9781450383127 .
  9. Cryptocurrency Tumbler. en.wikipedia.org . Accessed Oct 4, 2021.
  10. Towards Understanding Cryptocurrency Derivatives: A Case Study of BitMEX. Soska, Kyle; Dong, Jin-Dong; Khodaverdian, Alex; Zetlin-Jones, Ariel; Routledge, Bryan; Christin, Nicolas. The Web Conference, 2021. doi.org . Apr 19, 2021. ISBN: 9781450383127 .