// THE POLICY ENGINE
Controller Basics
The controller is your policy engine. It watches the gap between market price and redemption price and applies pressure to close it. You do not tune it with equations. You tune it with intent, then read the charts to see if the intent landed.
Think in terms of response style rather than math.
The Two Gains
Kp, the immediate reaction
Kp answers the current gap right now. Higher Kp reacts harder the moment price drifts.
- Too little feels sluggish. The gap lingers because you are barely pushing.
- Too much overshoots. You slam the gap shut and bounce past it, creating choppy back-and-forth.
Ki, the accumulated correction
Ki responds to pressure that does not go away. It builds a correction over time so persistent drift gets addressed even when no single moment looks dramatic.
- Too little leaves slow, grinding drift unresolved.
- Too much keeps pushing after conditions have already changed, arriving late and overcorrecting.
In the Arena, a gain switches on as soon as you set it above zero. A gain of zero simply means that term is not participating.
The Bounds
The rate bounds and integral bounds cap how much authority the controller is allowed to use.
| Setting | Effect |
|---|---|
| Wide bounds | More room to act during stress, but mistakes are louder when they happen. |
| Tight bounds | Calmer and more predictable, but may not be enough for a tougher run. |
Bounds are guardrails. They do not make the controller smarter, they limit how far a good or bad decision can travel.
A Tuning Habit That Works
- Change one control.
- Run it.
- Watch the peg gap and the controller output specifically.
- Decide if the change helped before touching anything else.
Most overtuning comes from changing three things at once and crediting the wrong one. Single steps build real intuition fast.
Aggressive settings can look impressive in a calm stretch and fall apart under stress. The strongest controllers are usually the calmest ones that still have enough authority when it counts.
Next: how reserves back all of this up, in Liquidity.