# Create ROI tools for analysis

I want to create some ROI tools on my game for personal analysis.

Let's say that I have 3 different level of special powers with a recursive cost and certain bonus. So, I want to see when certain level will start have income.

1st example: 1000 coins for 10 uses. Bonus: +10% on farming speed. 2000 coins for 10 uses. Bonus: +20% on farming speed. Etc

In this example, I know exactly what the bonus will do. I have the price of the farming per unit and I can calculate the income per use for each power level. So, I can tweak the numbers using the ROI tool to make the effect better.

2nd example: 1000 coins for 3 power ups on a weapon. Bonus: +1 attack power, +5% chance to destroy the weapon. 2000 coins for 3 power ups on a weapon. Bonus: +2 attacking power, 10% chance to destroy the weapon.

3rd example: 1000 coins per use. Bonus: 30% heal, 5% chance to get injured. 2000 coins per use. Bonus: 50% heal, 10% chance to get injured.

As you can see, the first example is simple to calculate the income vs cost. However, the second and third example don't have an exact effect and both of them have a probability to either good or bad effect. How do you work on this type of ROI? How do you calculate the pros of each level?

There are two ways to approach a problem like this, empirically and anecdotally.

Empirically, you could use tools like Excel or Machinations to model each. What you are looking at are essentially feedback loops. The more you use an item the more likely it is to have a negative effect, but also the more cumulative effect it will have. If you can model the total beneficial output against the detrimental output over time/uses, and adjust for cost, you should see some point where a higher level is more effective.

For example, in an RTS like StarCraft you could graph the number of gatherers (or resources spent on gathers) against the amount of resources gathered per period of time. You then also model that against how many resources you'll have a certain period, if you build only a specific number of gatherers.

Believe it or not, you're already using some empirical analysis. The fact that you are doubling both the reward and the penalty for each level, as well as the cost, is an example of simple empirical balancing. If you just continue with this and actually look at how it changes per each hit. There's a lot of possible outcomes, it's about modeling them all and seeing what's there.

Anecdotally, you could place the bonuses in the game as they are and track what players use. As players aim to maximize their effectiveness in a game, they'll naturally tend to a certain set of behavior. If more are skipping on the higher levels, maybe they bonus is too little or the negative too harsh. Then you can adjust, and continue watching results as players play.

It can be daunting, but I'd say start simple and if you find some specific concept you want to model, search for or ask another question about that specific concept.

Side note, those current numbers seem like they'd lead to an uninteresting choice for the player. Level 2 is literally 2 times the effects. If you introduce some sort of scaling or easing. Consider maybe the base benefit of the next level is 3 times as strong, but the chance of the weapon breaking increases at a faster rate. When you get your basics modeled, maybe consider something like that as a next step.