How to get useful results
Measure several catch intervals in the same maze, then use an average instead of your best run. For upgrades, compare the income before and after one controlled change. Small, repeatable tests produce better decisions than guessed rates.