Rabia Yoruk

XchangeWiser

A stock market product for individual investors. Forecasting and simulation showed each user the projected gains and losses of a decision.

Role
Data Scientist and ML Engineer
Where
OptiWisdom
When
2022 to 2023
Stack
  • Python
  • Custom forecasting models
  • Simulation
  • Docker
  • Kubernetes

1,000

individual investors, approximately

4 years

in production

XchangeWiser forecasting and simulation, simplified. Flow: Market data, then Data pipeline, then Forecast models, then Simulation, then Investor app. Market data prices Data pipeline clean, store Forecast models custom Simulation gains, losses Investor app about 1,000 users XchangeWiser forecasting and simulation, simplified. Flow: Market data, then Data pipeline, then Forecast models, then Simulation, then Investor app. Market data prices Data pipeline clean, store Forecast models custom Simulation gains, losses Investor app about 1,000 users
Fig. 1 XchangeWiser forecasting and simulation, simplified.

Problem

New investors often see one number: the price today. They do not see the range of what can happen next. XchangeWiser wanted to show that range before a user makes a decision.

What I built

  • Custom forecasting models for the assets in the product.
  • A simulation layer that turns forecasts into projected gains and losses for a user’s position.
  • The services that ran both in production, with retraining and deployment through CI/CD.

About 1,000 individual investors used it, and it ran for four years.

80 100 120 140 day 0day 20day 40day 60
  • Sample paths
  • Median
  • 10th to 90th percentile
Fig. 2 Projected value of a 100 unit position over 60 trading days. Illustrative, synthetic data.