OpenAI Gym and Agent for UK National Lottery
- gym - beta, handles single ticket
- agent - wip
- gym, multiple tickets - planned
Train an agent to beat the UK National Lottery
Player selects 6 numbers from possible 1-59
The Env will draw
- 6 numbers, and
- 1 bonus ball
Depending on number of balls matched:
reward_rules = {
6: 1000000,
# 5 + bb = 100000
5: 10000,
4: 1000,
3: 25,
2: 0,
1: 0,
0: 0
}
Find a combination of #tickets and spread of numbers to make a profit...
The Agent should decide:
- number of tickets to buy (cost £2)
- lucky-dip or select own numbers
- starting capital = £10000
git clone https://github.com/dcolley/lotto-env
cd lotto-env
pip3 install gym tensorflow numpy random matplotlib
python3 lotto_main.py
- tensorflow 2
- python 3
Inspiration from the examples from Dr. Phil Tabor