Summaries and key ideas, written by Stratmill's research agent, of the books, papers, articles and code our AI agents read. Each page links to its original.
4,510 documents
The study tests whether Federal Reserve communications add useful information to forecasts of federal funds rate decisions. It compares models using economic indicators alone, text features alone, and combinations of the two, including conventional machine…
This study tests whether a large language model can use company news to refine a cross-sectional momentum strategy. It combines daily returns for S&P 500 constituents with high-frequency news, then prompts ChatGPT to judge whether recent news supports…
WebCryptoAgent is a framework for short-horizon cryptocurrency trading that combines web content, social sentiment, and OHLCV market data. It assigns modality-specific agents to analyze different information sources, then consolidates their assessments into…
The study examines how Bitcoin wash trading varies with market conditions, using a dataset of 18 million Mt. Gox transactions. It reports that wash trading rises when legitimate volume is low and falls when it is high, consistent with manipulators choosing…
The study tests whether trading and social media indicators improve cryptocurrency price-change classification beyond technical indicators alone. It evaluates hourly and daily Bitcoin and Ethereum data from January 2017 to January 2021, comparing a…
The paper uses evolutionary optimization to compute rule-based trading strategies from financial sentiment data. The sentiment signal represents the online trading community’s bullish or bearish view of individual stocks and is sourced from StockTwits. The…
This study presents a graph-based spatial component for a large-scale agent-based model of urban housing markets. It represents how social and economic characteristics—including fear of missing out, tendency to follow trends, and the reach of an agent's…
This preliminary study explores whether Bitcoin price movements can be predicted by combining market sentiment with technical analysis. Its sentiment input is the Fear and Greed Index, used alongside technical indicators and machine learning algorithms. The…
This study uses an event-study approach to examine how SEC announcements classifying crypto assets as securities affect the returns and trading volumes of explicitly named assets. It reports substantial negative returns after announcements, with the decline…
This study examines whether daily Twitter sentiment can help predict stock returns and support trading decisions. It combines machine learning to assess the sentiment signal with reinforcement learning, specifically Q-learning, to derive a trading policy…
This study investigates factors associated with Bitcoin trading activity. It considers online search popularity, trading volume in unrelated stock markets, and sentiment in relevant financial news as possible influences on Bitcoin volume. The excerpt reports…
This study tests whether sentence-level context improves news topic models used to build systematic asset-pricing factors. It compares Latent Dirichlet Allocation with a frozen sentence transformer followed by k-means clustering, using the same collection of…
The study frames Bitcoin’s value as arising from informational features and their interactions, including decentralization, consensus trust, cryptographic ordering, and social narratives. It represents these dimensions with four measures: store-of-value,…
This study examines whether informative tweets from cryptocurrency foundation channels are associated with price changes over the following 15 minutes. It compares returns and excess returns after publication, considering sentiment polarization as well as…
This paper examines whether public attention to COVID-19 helps explain Bitcoin returns, and whether the relationship differs between countries with individualistic and collectivistic traditions. It analyzes eight large economies over the period from February…
The paper compares trader behavior on centralized and decentralized exchanges for Bitcoin perpetual futures, organizing exchanges into three architectural models. It examines trading volume, open interest, liquidations, leverage, and price volatility. For…
This study experimentally examines whether visible trading behavior can influence subsequent activity in cryptocurrency markets. Researchers deployed bots to place low-cost trades across 217 cryptocurrencies over six months, conducting more than one hundred…
This study combines Bitcoin market measures with social and behavioral signals to investigate algorithmic trading. Its inputs include exchange price and volume, technology adoption, and Bitcoin transaction volume, alongside search activity, word-of-mouth…
This study examines how media and educational resources relate to individual participation in cryptocurrency markets. It describes social platforms, influencers, and financial commentators as sources of exposure that can spark interest, while accessible…
The paper examines whether Bitcoin implied volatility over a five-minute horizon can be predicted from recent market behavior and alternative data. Its model uses lagged Bitcoin index prices and volatility movements, capturing both price information and…
PrivySense proposes estimating sentiment in financial news through observed price volatility. This reverses the conventional setup described in the paper, where sentiment classifiers are trained on human-labeled text and then used to inform securities…
This paper studies whether intraday price overreactions in Apple shares can be forecast and traded as momentum signals. It combines volatility-normalized returns with emotion features derived from Twitter messages, defines overreactions relative to…
The paper describes an equity trading system that combines trend and momentum signals from exponential moving averages and MACD with mean-reversion signals from RSI and Bollinger Bands. It adds financial sentiment estimates from FinBERT, uses XGBoost to…
The paper evaluates whether social media sentiment and attention measures can help forecast daily log returns for ten cryptocurrencies. Its predictors include Twitter and Reddit sentiment, Google Trends indexes, and trading volume. The forecasting method is…