AI Quant Base

Understand the inputs

A common form is RSI = 100 − 100/(1 + RS), where RS is average gain divided by average loss. Wilder-style smoothing is widely used. Different initialization or smoothing choices can lead to different values.

Handle boundary cases

State how zero losses, zero gains and a completely flat sequence are handled. Confirm that the indicator has enough observations before using it. Platform implementations should be checked rather than assumed identical.

A threshold is not a forecast

A high or low RSI describes the selected sequence under that calculation. It does not guarantee a reversal. If a threshold becomes a trading condition, specify timing, exits and validation separately.

Sources & references

References support specific definitions or methodological context. Examples and research checklists are explanatory; no performance result is claimed.

Educational guide. No backtest or live trading result is reported on this page.

Continue exploring

Strategy research templates

Mean reversion: defining the reference

A mean-reversion hypothesis expects a deviation from a defined reference to narrow. The reference, horizon and exit rule determine what is actually being tested.

Read guide
Backtesting & risk

Out-of-sample testing and strategy changes

Out-of-sample testing evaluates a frozen research specification on observations that were not used to choose it. The split and the change history matter.

Read guide
Search the library

Search products, workflows, guides and tools.