Mining Strategies

The AutoMiner supports five tile selection strategies and multiple progressive sizing modes. Each strategy approaches the mining game differently — some optimize for expected value, others for risk management or unpredictability. All strategies work with both ORE and GODL.

Kelly Criterion

The Kelly Criterion is a mathematically optimal strategy for sizing bets based on your edge and bankroll. Originally developed for information theory and applied to gambling and investing, it maximizes the expected logarithm of wealth — which translates to the fastest long-term growth rate without risking ruin.

How It Works

The strategy evaluates each tile's expected value based on the current deployment distribution, then calculates the optimal fraction of your bankroll to deploy using the Kelly formula:

f* = (bp - q) / b

Where b is the odds (net payout ratio), p is the probability of winning (1/25 for each tile, adjusted by deployment distribution), and q is the probability of losing (1 - p). Smelted uses a fractional Kelly approach, typically deploying a fraction of the full Kelly amount to reduce volatility.

When to Use It

• You are mining over many rounds and want to maximize long-term growth.
• You want automatic risk management — Kelly naturally scales down bets when your edge is thin.
• You prefer a mathematically principled approach over intuition.

Pros and Cons

• Pro: Optimal long-term growth rate with built-in bankroll protection.
• Pro: Automatically adjusts bet size based on edge — bets less when conditions are unfavorable.
• Con: Can be conservative in rounds with very even deployment distribution.
• Con: Requires many rounds to realize theoretical edge; short sessions may underperform.

Best EV

The Best EV (Expected Value) strategy selects the tile with the highest expected value in the current round. Expected value is calculated for each tile based on how much SOL is already deployed there relative to the total pool and the potential payout.

How It Works

For each of the 25 tiles, the strategy computes:

EV = (probability of winning) x (reward if win) - (cost to deploy)

Tiles with less SOL deployed relative to others will have a higher reward-to-cost ratio. The strategy picks the tile with the maximum EV and deploys the configured amount. Unlike Kelly, Best EV does not adjust the bet size — it deploys your full per-round budget to the best tile.

When to Use It

• You want to consistently pick the most favorable tile each round.
• You are comfortable deploying a fixed amount and want to maximize the per-round expected return.
• Round deployments are uneven, creating clear EV differences between tiles.

Pros and Cons

• Pro: Simple and intuitive — always picks the mathematically best tile.
• Pro: Performs well when deployment distribution is skewed.
• Con: Does not manage bet sizing; deploys full budget regardless of edge quality.
• Con: Other miners using the same strategy may converge on the same tile, reducing edge.

Low Pool

The Low Pool strategy targets the tile with the least total SOL deployed. The logic is straightforward: tiles with less competition mean that if you win, you keep a larger share of the reward pool.

How It Works

The strategy scans the current deployment state of all 25 tiles and selects the one with the lowest total SOL. If multiple tiles are tied at the minimum, one is chosen at random. Your configured per-round budget is deployed to this tile.

When to Use It

• You want to minimize competition for reward share on the winning tile.
• You notice that most miners cluster on a few popular tiles, leaving others underserved.
• You prefer a contrarian approach to tile selection.

Pros and Cons

• Pro: Maximizes your share of the pot when you win — less competition means bigger payouts.
• Pro: Simple to understand and predictable in behavior.
• Con: Does not account for why a tile is underfunded — it may simply be that experienced miners are avoiding it.
• Con: Can be predictable if other miners use the same strategy.

Random

The Random strategy selects a tile using uniform random distribution. Every tile has an equal probability of being chosen, regardless of current deployment state.

How It Works

Each round, the strategy generates a random number between 0 and 24 and deploys to that tile. No analysis of the round state is performed. The configured per-round budget is deployed in full.

When to Use It

• You want maximum unpredictability so other miners cannot anticipate your deployments.
• You believe the deployment distribution across tiles is efficient (no clear edge).
• You want a baseline to compare against other strategies in backtesting.

Pros and Cons

• Pro: Completely unpredictable — no other miner can game your selections.
• Pro: Zero analysis overhead; fastest execution.
• Con: Ignores all available information about tile deployment distribution.
• Con: Will often deploy to suboptimal tiles when clear EV differences exist.

Even/Odd

The Even/Odd strategy deploys to either all even-numbered or all odd-numbered tiles. This provides broad coverage with a simple, deterministic selection rule.

How It Works

Tiles are numbered 1-25. Selecting "Odd" deploys to tiles 1, 3, 5, 7, 9, 11, 13, 15, 17, 19, 21, 23, 25 (13 tiles). Selecting "Even" deploys to tiles 2, 4, 6, 8, 10, 12, 14, 16, 18, 20, 22, 24 (12 tiles). Your per-round budget is split evenly across all selected tiles.

When to Use It

• You want broad coverage — roughly half the board each round.
• You prefer a simple, predictable deployment pattern.
• You want to pair it with a progressive sizing mode for recovery mechanics.

Pros and Cons

• Pro: Covers roughly half the tiles, giving a ~48-52% chance of winning each round.
• Pro: Simple and deterministic — no analysis required.
• Con: Spreads your SOL thin across many tiles, reducing per-tile payout share.
• Con: Does not consider tile EV or deployment distribution.

Sizing Modes

Sizing modes adjust your bet size based on win/loss history. They can be combined with any tile selection strategy except Kelly Criterion (which has its own bankroll-based sizing). All sizing modes use a base bet per tile that you set, and the AutoMiner stops if the effective bet exceeds your wallet balance.

Martingale

Doubles your bet after each loss. Resets to base bet on a win. Recovers all losses with a single win, but exponential growth can drain your bankroll fast on losing streaks.

effective_bet = base_bet × 2^consecutive_losses

Anti-Martingale

The reverse: doubles your bet after each win. Resets to base bet on a loss. Rides winning streaks while capping downside to the base bet. Lower risk than Martingale since losses always revert to base.

effective_bet = base_bet × 2^consecutive_wins

D'Alembert

Increases bet by one unit after a loss, decreases by one unit after a win (minimum: base bet). Linear progression — gentler than Martingale but slower to recover. Works well when wins and losses roughly equalize.

effective_bet = base_bet × (1 + unit_offset)

Fibonacci

Uses the Fibonacci sequence (1, 1, 2, 3, 5, 8, 13...) as the bet multiplier. Advances one position on loss, moves back two positions on win. Grows slower than Martingale but recovers without needing to win back every single round.

effective_bet = base_bet × fibonacci(position)

Pair any sizing mode with any tile selection strategy: Best EV + Martingale picks optimal tiles with aggressive sizing, Even/Odd + Fibonacci covers half the board with moderate recovery mechanics, etc.

Choosing a Strategy

The right strategy depends on your goals, risk tolerance, and how many rounds you plan to mine. Here is a comparison to help you decide: