FTEC Info Session

Overall Introduction

Cutting-edge Technologies:

  1. Novel Qiant. Tech
  2. AI Tech
  3. Blockchain Tech
  4. Big Data Tech

FinTech

image-20260825141107634

FinTech vs. Traditional Finance

  • Traditional finance
  • Tech (math, CS, statistics)
  • FinTech (Robo-advising, block chain, AI finance, etc)

UG Programm

Not economy degree, is a BSc

  • Key Feature: Technology Empowers Finance
  • Integration : Finance, Math, CS
  • Tri-Curriculum : Core of Finance, Foundations of Mathematics, Computer Technology

Fundamental courses :

  • Calculus

  • Multivariable Calculus

  • ICS

Major required :

  • Blockchain, Cryptocurrencies and Smart Contracts
  • Academic Orientation for FTEC Students
    • Module
    • Combine in-class and out-class field trip

Activities

  1. Field Trip
  2. Competitions
    1. Math Module
    2. 量化交易
    3. BlockTrain + Finance
  3. Research (X programm)
  4. FinTech Club

Operating Financial Markets on Blockchains （区块链）

Blockchains are decentralized digital trust platforms

  • decentralized : not rely on central, but by distribute computer

Trust : Human success is based on flexible cooperation in large numbers, This require trust.

  • Tribal Trust
  • Distributed Trust
  • Institutional Trust

Blockchain

  • Consensus Layer

    A public append-only data structure

    • Persistence : once added, data can never be removed
    • Consensus : all honest participants have the same data
    • Liveness : honest participants can add new transcript
    • Open : anyone can add data

  • Compute layer (blockchain computer)

    DApp logic is encoded in a program (called a smart contract) that runs on blockchain

    • Transparency : no single trusted third party

  • App layer

    Example : DeFi is just operating financial markets on blockchain computer

    DeFi is Atomic

    • Option to add - all or none logic of execution for transactions interacting with multiple instruments

    Use Case : Flash Loan Arbitrage | Anyone can take a very large capital loan (with no collateral), perform arbitrage, earn money and return capital, in a single transaction

Stablecoins : Bridging DeFi and TradFi

USDC (by Circle), USDT (by Tether)

Centralized Stablecoins

Note : Circle is a US listed company

LLMs in FinTech : Revolutionizing Finance with AI

LLMs: Large Language Models trained on massive texts to understand and generate language

Training Recipe :

  1. Pre-training : Learning to generate, and trained on large amount of unlabeled data
  2. Supervised Fine-Tuning : Learning to follow instructions , and trained on labeled data
  3. RL Training : Align with human preference, and optimized for correctness & quality
  4. Continual Learning : Specializing general LLMs for task-specific adaptation
    • Prompt-Based (RAG) Approach
    • Parameter-Based (LoRA) Approach

Trilemma in Large Language Models

image-20260825144941055

From General LLM to Financial LLM

  1. Chatbot
  2. Reasoning
  3. Agent
  4. Innovator
  5. Organization

LLM of application :

  1. Multimodal Financial Big Data (Data Layer)
  2. Financial Knowledge, Toolchain (Model Layer)
  3. Chatbot, Copilot, Agent

Example 1 :Daily Market Report

Example 2 : Financial Event Analysis

AI in Quantitative Trading (QT)

Usage of mathematical models and data-driven techniques for automatic financial decision making

  • Financial Asserts : Stock, Commodity, Bond, Crypto
  • Financial Exchanges

The real situation :

  • Direct application of fancy ML methods always fail in finance scenaros
  • Overfitting on noise & garbage in garbage out always happen
  • Simple linear model and human intuition still play a vital role

Low signal-to-noise ratio:

  • Predictive signal is weak, unstable, and regime-dependent
  • Easy to overfit on noise rather than signal

image-20260825151105876