Finforge builds Sophon-3, a system of models that keep learning. It screens the market, forecasts, and constructs portfolios using only information available at the time. The research is published, limitations included, and the system runs live today.
Finforge exists to give retail investors the kind of Al research that has always been locked inside institutions.
We publish our methods, our evidence standards, open for anyone to read, question, and verify.
Research paper
Why markets demand Al that keeps adapting. Sophon-3 turns continually-learning models into a live, tradable portfolio.
Research paper
Most strategies die leaving the lab. Our live trading and backtests are the same system, run twice - with identical, auditable results.
Research paper
How our live agents are configured and locked in before they trade real money — so results can't be cherry-picked afterward.
See how each project is built and how it performs. Every agent includes a Sophon-3 model node tree, a growth chart for benchmark index comparison and a table with YoY metrics. There are more projects on our Agents page.
Medium Risk
Nasdaq 100
Last updated
loading...
Alpaca account
Paper account ****8UEM
Live return [1]
22.1
Benchmark return [1]
11.9
Excess return [1]
10.1
Max drawdown [1]
-18.0
Medium Risk
Nasdaq 100
Last updated
loading...
Alpaca account
Paper account ****KYND
Live return [1]
22.1
Benchmark return [1]
11.9
Excess return [1]
10.1
Max drawdown [1]
-14.5
Low Risk
Nasdaq 100
Last updated
loading...
Alpaca account
Paper account ****L76H
Live return [1]
18.8
Benchmark return [1]
11.9
Excess return [1]
6.8
Max drawdown [1]
-4.5
High Risk
S&P 500
Last updated
loading...
Alpaca account
Paper account ****1456
Live return [1]
17.3
Benchmark return [1]
9.4
Excess return [1]
7.9
Max drawdown [1]
-34.4
