Papers, wiki, Option Blackboard, encyclopedia, and cards.
Stock portfolio optimization is the process of continuous reallocation of funds to a selection of stocks. This is a particularly well-suited problem for reinforcement learning, as daily rewards are compounding and objective functions may include more than just profit, e.g., risk and sustainability. We developed a novel utility function with the Sharpe ratio representing risk and the environmental, social, and governa…
This paper investigates how similarity in the informational representation of market states among Artificial Intelligence (AI) trading agents can generate systemic instability in financial markets. We construct a structural multi-agent market model calibrated using high-frequency microstructural moments. AI agents are modeled through a two-layer decision architecture consisting of a nonlinear representation layer and…
We show that AI-driven investment strategies are inherently self-defeating at scale. As AI adoption rises, three mutually reinforcing channels -- signal crowding, performative signal erosion, and Red Queen competition -- compress excess returns. We derive the alpha half-life $h(φ) = \ln 2/[θ+ δ(φ)]$, where $θ$ is the natural mean-reversion rate and $δ(φ) = Nφρa/λ(φ)$ is the AI-accelerated decay component, which is co…
Performance attribution analysis, defined as the process of explaining the drivers of the excess performance of an investment portfolio against a benchmark, stands as a significant feature of portfolio management and plays a crucial role in the investment decision-making process, particularly within the fund management industry. Rooted in a solid financial and mathematical framework, the importance and methodologies …
The goal of cryptocurrencies is decentralization. In principle, all currencies have equal status. Unlike traditional stock markets, there is no default currency of denomination (fiat), thus the trading pairs can be set freely. However, it is impractical to set up a trading market between every two currencies. In order to control management costs and ensure sufficient liquidity, we must give priority to covering those…
We propose a unified multi-tasking framework to represent the complex and uncertain causal process of financial market dynamics, and then to predict the movement of any type of index with an application on the monthly direction of the S&P500 index. our solution is based on three main pillars: (i) the use of transfer learning to share knowledge and feature (representation, learning) between all financial markets, incr…
We present SAiFE_gym, a Python module that provides a collection of simulation environments for studying trading problems in Constant Product Markets (CPMs) with Concentrated Liquidity (CL). These markets give Liquidity Providers (LPs) granular control over how their capital is allocated and enable them to adjust their range of liquidity provision dynamically based on market conditions, which in turn, dictates how th…
Credit risk models increasingly need to combine predictive accuracy with transparent explanations and auditable fairness constraints. Logistic regression remains attractive because its coefficients are easy to interpret, but it can miss nonlinear structure. Flexible models can improve prediction, but their explanations are often post-hoc and may not describe the decision rule itself. We introduce $\texttt{findr}$, sh…
Large language models (LLMs) are increasingly used in investment decision-making, yet prior work shows that they exhibit systematic, model-specific investment preferences. We study whether a model's overall investment stance can be calibrated to a specified direction and strength. We introduce an investment-bias dial, an inference-time intervention on a single neuron that continuously adjusts a model-level decision p…
Financial institutions are delegating consequential decisions to agentic AI systems that decompose goals, coordinate models and tools, and act with little oversight. Yet agentic AI governance in FinTech is under-investigated. We argue the binding governance constraint is not capability but verifiability. We define the Verifiability Gap as the shortfall between the verification delegated authority demands and the expl…
We audit whether candle-based machine-learning models can turn predictions of cryptocurrency extrema or short-horizon outcomes into positive Binance Spot paper policies after assumed costs. Numerical results come from scripted fixed-seed model runs and deterministic simulators; human-supervised AI agents supported the July 20 evidence-integrity revision through literature retrieval, separately tasked critique, artifa…
Large Language Models (LLMs) have emerged as powerful tools for processing the heterogeneous information environments of modern financial markets. This paper presents a systematic, comparative evaluation of five prominent LLMs: GPT-4 Turbo, Claude 3 Opus, Gemini 1.5 Pro, Llama 3 70B, and the domain-specialized FinGPT, with respect to their capacity for technical market analysis. The evaluation spans four structured t…
We propose an AI economist agent for economic and financial scenario analysis. Scenario design often requires analysts to assess emerging risks with limited historical precedent, combine information from many sources, and translate qualitative mechanisms into internally consistent quantitative paths. Large language models (LLMs) can search and synthesize this information, but fluent narratives alone do not establish …
Investment management is a high-stakes domain in which agentic AI systems must do more than generate plausible text. They must retrieve point-in-time data, assemble correct computational inputs, invoke specialized methods, and produce auditable structured outputs. We introduce FinSkillBench, an evaluation suite designed to measure whether language model agents can effectively use financial domain skills to solve inve…
Deep reinforcement learning (DRL) frameworks for portfolio optimization have shown promise for their ability to learn allocation rules dynamically from market data. However, these models fail to account for fat-tailed returns, which characterize actual market behavior with more frequent extreme events. Furthermore, historical data is treated homogeneously, without accounting for temporal importance, leading models to…
This study aims to determine whether the application of Deep Reinforcement Learning (DRL) as a specialized execution overlay can enhance pair trading in highly volatile cryptocurrency markets. Although classical implementations of the strategy have proven successful in traditional equities, they frequently exhibit rigidity and suffer from severe divergence risks when applied to high-variance environments. To address …
AI losses that arise through an insured organization's generative or agentic AI system require state reconstruction, not merely event reconstruction, because the relevant state changes as the system reasons, retrieves, calls tools, and acts. The relevant question is not only what loss occurred, but what the system was allowed to do, what it actually did, and whether that reconstructed loss can support insurance claim…
Ship finance is a data-intensive and document-heavy segment of asset-based lending, requiring the integration of financial, technical, contractual, and regulatory information from heterogeneous and largely unstructured sources. Increasing environmental regulation and ESG reporting requirements are adding further complexity to underwriting and loan-origination processes. Recent advances in artificial intelligence (AI)…
An activation function is the pointwise nonlinearity between linear layers. Without it, stacked layers collapse to a single linear map.
Adam is an adaptive first-order optimizer that keeps exponential moving averages of the gradient and its square, giving per-parameter step sizes.
An agent workflow is a structured loop that plans, calls tools or models, observes results, and repeats until a stop condition — a pipeline with memory, contracts, and failure handling rather than a single completion.
Attention builds a weighted average of values, with weights from a compatibility function of queries and keys. It lets a model focus on relevant parts of a context instead of a single fixed vector.
An autoencoder learns to reconstruct its input through a bottleneck, producing a compressed latent that can be used for denoising, retrieval, or as a generative seed.
Backpropagation computes gradients of a scalar loss with respect to every weight by applying the chain rule backwards through the computational graph.
Batch normalization re-centers and re-scales layer inputs using mini-batch statistics, then learns a scale and shift, reducing internal covariate shift and allowing higher learning rates.
Beam search is a heuristic decoder that keeps the k best partial sequences at each step instead of greedily taking only the top token — the classic seq2seq inference method.
BERT is a bidirectional Transformer encoder trained with masked language modeling and next-sentence prediction, then fine-tuned on downstream NLP tasks.
BPE grows a vocabulary by repeatedly merging the most frequent adjacent pairs, starting from characters or bytes, until a target vocab size is reached.
Chain-of-thought prompting asks the model to emit intermediate reasoning steps before the answer, which reliably lifts arithmetic, symbolic, and multi-hop tasks.
CLIP jointly trains an image encoder and a text encoder so matched image–caption pairs are close in a shared space, enabling zero-shot visual classification by text prompts.
Contrastive learning pulls representations of related pairs together and pushes unrelated pairs apart. It is the pretraining idea behind SimCLR, CLIP, and many embedding models.
A CNN shares a local kernel across spatial (or temporal) positions, building translation-equivariant features. It is the inductive bias that cracked modern computer vision.
Cross-entropy measures how well a predicted distribution q matches a target distribution p. For one-hot labels it reduces to −log q(y), the usual classification and language-model loss.
DQN approximates Q(s, a) with a deep net, using experience replay and a frozen target network so the TD target does not chase itself every step.
A diffusion model learns to reverse a gradual noising process. Sampling starts from noise and iteratively denoises toward the data distribution.
Dropout randomly zeroes hidden units during training so the net cannot rely on any single co-adaptation, then scales weights at test time (or uses inverted dropout).
Early stopping treats training time as a capacity knob: halt when a validation metric stops improving so the model does not wander into overfit.
An embedding is a learned dense vector for an object (token, sentence, image, user) such that geometry supports retrieval, clustering, or as input to a downstream model.
Fine-tuning continues training a pretrained model on a narrower distribution so the same weights specialize — classification heads, instruction following, or a desk domain.
FlashAttention computes exact attention with tiling that keeps softmax stats in SRAM, cutting HBM traffic and unlocking longer contexts at the same FLOP count.
GRU is a lighter gated RNN with reset and update gates, often matching LSTM quality at lower cost on medium-length sequences.
A GAN trains a generator and a discriminator against each other: the generator maps noise to fake samples, the discriminator learns real vs fake, and the equilibrium is a generator whose samples match the data distribution.
GPT is a decoder-only Transformer trained to predict the next token. Scale plus this objective produced in-context learning and the current foundation-model product line.
Gradient descent updates parameters against the gradient of a loss: θ ← θ − η ∇_θ L. Stochastic and mini-batch variants make the method tractable on large datasets.
A GNN updates each node from its neighbors. Message passing lets the model use relational structure — markets, molecules, citation graphs — instead of forcing a grid.
Hallucination is fluent generation that is not supported by the source or the world — a likelihood-trained model completing a pattern, not a database lookup.
In-context learning is when a frozen language model improves at a task from examples placed in the prompt, without weight updates.
Knowledge distillation trains a smaller student to match a teacher’s output distribution (soft labels), transferring behavior without copying every weight.
Layer normalization standardizes activations across features for each example, not across the batch — the stabilizer that made Transformers trainable.
A learning-rate schedule is the planned path of η_t — warmup, cosine, step decay — that often matters more than the architecture headline on a given run.
LSTM is a gated RNN whose cell state can carry information across many steps, with input, forget, and output gates trained by gradient descent.
LoRA fine-tunes a frozen model by learning low-rank adapters on selected weight matrices, cutting trainable parameters and storage versus full fine-tunes.
MoE routes each token (or example) to a sparse subset of specialist feed-forward experts, raising parameter count without paying dense FLOPs on every token.
A neural network is a layered function approximator: units compute a weighted sum, apply a nonlinearity, and pass the result forward so the whole stack can learn a mapping from inputs to outputs.
An activation function is the pointwise nonlinearity between linear layers. Without it, stacked layers collapse to a single linear map.
Adam is an adaptive first-order optimizer that keeps exponential moving averages of the gradient and its square, giving per-parameter step sizes.
An agent workflow is a structured loop that plans, calls tools or models, observes results, and repeats until a stop condition — a pipeline with memory, contracts, and failure handling rather than a single completion.
Grains and oilseeds — corn, soy, wheat, and their products — where weather, USDA prints, and harvest calendars sit on top of generic trend.
A CTA whose signals come from ML (trees, nets, representations) rather than a hand-written MA — still a futures risk engine underneath.
Alpha is return not explained by the risk factors you chose — a residual, not a personality.
Animal spirits is Keynes’s name for the non-model confidence that makes people invest or refuse to — the residual when rates and cash flows are not enough to explain the tape.
The 1997–98 Asian crisis was a sequence of peg breaks, bank runs, and sudden stops starting in Thailand — short-dollar corporate debt plus weak bank regulation meeting a reversal of carry.
An ABS is a bond paid from a pool of receivables — cards, auto, equipment — sliced into tranches with a waterfall.
Enter on a trend signal, then trail a stop at k × ATR behind the favorable extreme — Wilder volatility as the exit engine.
Attention builds a weighted average of values, with weights from a compatibility function of queries and keys. It lets a model focus on relevant parts of a context instead of a single fixed vector.
An autoencoder learns to reconstruct its input through a bottleneck, producing a compressed latent that can be used for denoising, retrieval, or as a generative seed.
Backpropagation computes gradients of a scalar loss with respect to every weight by applying the chain rule backwards through the computational graph.
Balance Sheet Constraint Dealer — Dealer SLR/balance-sheet limits reducing intermediation.
Bank Term Funding Program Legacy — Crisis facility allowing par advances against securities.
BTFP usage tracks how much funding banks obtain through the Bank Term Funding Program, offering insight into balance-sheet stress and demand for official liquidity backstops.
Bankruptcy is a court process that stays creditors and restructures or liquidates claims when a firm cannot meet its obligations as they come due.
Barings Bank was wiped out in 1995 by Nick Leeson’s hidden Nikkei futures losses in Singapore — a rogue-trader plus failed control story, not a macro crisis.
Batch normalization re-centers and re-scales layer inputs using mini-batch statistics, then learns a scale and shift, reducing internal covariate shift and allowing higher learning rates.
Beam search is a heuristic decoder that keeps the k best partial sequences at each step instead of greedily taking only the top token — the classic seq2seq inference method.
A bear market is a sustained decline in a broad index — the folk threshold is −20% from a peak, which is a headline, not a model.
A bear steepener is a curve move where long yields rise more than front yields (or fronts fall less) as the market prices more term premium, more deficit, or less faith in long-run restraint — and duration loses.
BERT is a bidirectional Transformer encoder trained with masked language modeling and next-sentence prediction, then fine-tuned on downstream NLP tasks.
The same BAB recipe on country indexes or international stocks — low-beta vs high-beta outside the US single-name tape.
A funding-stress monitor can warn before equity volatility reprices.
A liquidity premium monitor can separate default risk from liquidity-driven spread widening.
A card for handling fast markets, wide spreads, liquidity air pockets, and broken execution assumptions.
A card for tracking cross-currency basis, repo dislocations, and balance-sheet stress as early warning signals.