I'm looking to implement some sort of speed run mode in my game.

Let's say I want to assign each level three 'winning' bars, (bronze, silver and gold for example), which are independent of the player, and are based only on the difficulty of the level itself.

I can think of two methods:

  • Having a heuristic which outputs these three bars, given the level's settings.
  • Spending some time for testing the game over several players, and collecting statistics (mainly time to completion), in order to determine the bars.

Are both methods considered acceptable? Or is one of them is somewhat "banned"?

I would also like to consider a more limited situation, where there's not a real option to perform a wide-scale statistics collection. Would that completely throw the second method off the windows? Or is small-scaled (for example, having 5-10 players playing a level for a few times) is usually enough in common indie game developer scales?

I guess the answer here should depend on the nature of the game itself, but I wonder if this certain type of test ('winning bars' for speed runs) is considered easy to handle.

  • \$\begingroup\$ Truthfully there's no substitute for large scale data collection. You could always try to build an AI to attempt to run the course with different settings and adjust until it looks human. Or you could get each person to run the courses multiple times to make up the numbers (not very scientific because order effects will mean people improve, but your playerbase will improve anyway so it's somewhat representitive). Ideally you could make the demo version used by your playtesters connect to some central server after each runthrough to post results so you can analyse them quicker. \$\endgroup\$
    – Pharap
    May 13, 2017 at 15:32

1 Answer 1


In my personal experience, it depends on the game and its complexity. But it's not uncommon to use player/designer statistics and base values off them.

Often heuristics don't give accurate values for all levels and you might waste a lot of time trying put together an accurate algorithm for all of your level types.

What you might find though, is what ever scores/times you think are achievable, players will always beat your estimates so take that into consideration.


You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .