Settings → Genetic Options

The Genetic Options configuration tab becomes available only when Genetic Evolution is selected in Settings → What to Build → Build Mode. This section allows you to configure how genetic evolution algorithms are applied to evolve trading strategies over multiple generations.

1. Genetic Evolution Configuration

Max # of Generations

• Defines how many generations the population will evolve.
• Recommended range: 5–100
• Note: Too many generations often yield diminishing returns. It's often more effective to restart evolution after stagnation.

Population Size

• Number of strategies per island.
• Recommended range: 10–100+
• If using multiple islands, total population = island count × population size.

Crossover and Mutation Probability

• Set the likelihood of genetic operations:
  • Crossover: This value says how much parents (strategies from previous generation) are used to interfere with one another. With a high value more blocks are exchanged between those strategies to create candidates for a new generation
  • Mutation: it is the likelihood of changing a strategy’s rule (block) to make a new strategy. With a low mutation setting, only minor changes are being made to strategies during the evolution process
• Higher mutation rates can help explore more diverse strategy solutions.

2. Island Options

Islands represent isolated sub-populations that evolve independently, enhancing diversity and exploration.

Islands

• Number of separate evolutionary sub-groups.
• Recommended range: 1–10
• More than 10 islands may slow progress due to excessive population size.

Migrate Every Xth Generation

• Frequency of strategy migration between islands.
• Recommended value: Every 10 generations
• Helps “unstick” populations trapped in local optima while preserving diversity.

Population Migration Rate

• Number or percentage of strategies to migrate.
  • For small populations (e.g., size 10): use 10–20%
  • For larger populations (e.g., size 100): use 1–5%

3. Initial Population Generation

Genetic evolution can begin from:

• A randomly generated population
• A user-defined set of strategies loaded from a databank

Use Strategies from Initial Population Databank

• Enable this option to load strategies from the Initial Population databank.
• If there are fewer strategies than needed, the remainder will be generated randomly.
• Note: These are not filtered using the Initial Population filter.

Generated Decimation Coefficient

• Controls oversampling of candidate strategies for initial selection.
  • Example: Decimation = 3 means generating 3x more strategies than needed, then selecting the best.
• Improves initial quality but increases build time significantly.

Filter Generated Initial Population

• Optional filter to set minimum requirements (e.g., number of trades) for initial population strategies.
• Recommended to keep filters minimal to allow evolution to refine weaker strategies.

4. Evolution Management

Start Again When Finished

• Automatically restarts evolution after reaching the final generation.
• Suitable for long, unattended runs.

Restart Evolution Fitness If…

• Triggers a restart if no fitness improvement is detected across generations.
• Helps avoid wasted computation on stagnant populations.

5. Fresh Blood Mechanisms

Used to introduce new genetic material and maintain diversity.

Detect Same Strategies and Replace

• Detects duplicate strategies in each generation and replaces them with randomly generated ones.

Replace X% of the Weakest Strategies

• Periodically removes the lowest-performing strategies and replaces them with new ones.
• Promotes stronger candidates and avoids population stagnation.

6. Show Last Generation Databank

• Displays a special databank showing the current generation’s strategies.
• Available only for the first island.