Evolution
A Brief History of Financial AI
From statistical arbitrage to financial emergence
Quant Era1982–2011
1982
Renaissance Technologies[Wikipedia]
Jim Simons founds quantitative trading pioneer1988
Medallion Fund[Wikipedia]
Launches with 66% avg annual returns before fees1988
D.E. Shaw[Wikipedia]
David Shaw founds computational finance pioneer2001
Two Sigma[Wikipedia]
Overdeck and Siegel found ML-native hedge fundDeep Learning2012–2018
2012
Upstart founded[Wikipedia]
Ex-Googlers pioneer deep learning for credit decisions2015
PayPal ML fraud[Harvard]
Fraud rate drops below 0.3% using neural networks2016
Deep learning fraud research[Springer]
Fu et al. apply CNNs to 260M bank transactions2018
S&P acquires Kensho[TechCrunch]
$550M for AI analytics—largest AI deal at the timeNLP Era2019–2022
2019
FinBERT released[arXiv]
First finance-specific language model (Araci)2020
GPT-3 sentiment trading[ScienceDirect]
LLM-based strategies outperform traditional NLP2022
NLP adoption accelerates[LSE]
Hedge funds deploy transformer models at scaleFoundation Models2023–2024
2023
BloombergGPT[Bloomberg]
50B params trained on 363B financial tokens2023
Morgan Stanley + OpenAI[Morgan Stanley]
GPT-4 deployed to 16,000 wealth advisors2023
FinGPT open source[arXiv]
AI4Finance releases open financial LLM2024
JPMorgan LLM Suite[CNBC]
AI assistant deployed to 60,000+ employeesEmergence2025–Now
2025
Labs target finance
OpenAI and Anthropic invest in financial pre-training & build RLHF pipelines. Financial capabilities start to accelerate.2026
The race is on
AI systems are making consequential decisions across lending, trading, and risk management.The firms building now will shape capital flows for decades.