CCI has no ceiling
Most oscillators are bounded. RSI and stochastic live between 0 and 100, Williams %R between −100 and 0. CCI is the exception, and it is the only property of it you really need to hold on to:
CCI is unbounded. There is no maximum reading. ±100 is a convention, not a limit, and readings past ±300 happen.
Everything practical about building a CCI strategy follows from that.
Why a CCI threshold does not port between instruments
A bounded oscillator is self-normalising: RSI at 30 means "recent losses outweigh recent gains by a fixed ratio" on any instrument, because the formula divides one by the other. A CCI reading has no such ceiling, so how often price reaches ±100 depends on how volatile the instrument is relative to its own recent average.
The consequence is concrete:
| RSI at 30 | CCI at −100 | |
|---|---|---|
| on a quiet FX pair | reached a few times a month | reached often |
| on a volatile index | reached a few times a month | reached constantly |
| threshold portable between the two? | largely yes | no |
So a CCI strategy tuned on one symbol and applied to another is not a strategy applied to new data — it is a different strategy, because the threshold now selects a different fraction of bars. This is the sharpest single example of the general point that a strategy does not port between instruments.
What to do instead of a fixed threshold
Two options, both of which restore portability:
Re-derive the threshold per instrument. Plot CCI on the symbol you intend to trade, look at what fraction of bars exceed your candidate level, and pick the level that gives you the trade frequency you want. Then record it next to the instrument name, because it is not transferable.
Or make the condition relative. "CCI above its own n-period average" is dimensionless in the same way an ATR ratio is, and it means the same thing everywhere — the construction is described in the ATR regime filter.
Building the mean-reversion version
| Field | Value | Note |
|---|---|---|
| Indicator | CCI | |
| Length | 20 | conventional |
| Oversold | −100 | re-derive this per instrument |
| Overbought | +100 | |
| Exit | return to 0, or an ATR stop | CCI gives no exit of its own |
Mean reversion is the honest use of CCI, and it carries the mean-reversion hazard: the strategy is short volatility. It wins often and small, and loses rarely and large, when a move that looked extended keeps going. So the drawdown figure matters more here than the win rate, and a high percentage of winners is expected rather than impressive — which numbers can carry a decision covers why that pairing misleads.
And the usual caveat
An extreme CCI reading says price is far from its recent average. It does not say the move is over. In a trend, "far from the average" is the normal state, and a mean-reversion entry against a trend is the classic way to be right four times and lose more on the fifth than the first four made.
Tactix AI — Studio vs Guide
Tactix AI is AlfaTactix’s product assistant brand (open Tactix AI).
- Tactix Studio turns a one-sentence strategy description into a draft across Timeframe, Signals, Filters, and Risk in the visual Strategy Builder. You review and edit every field before Code Generator writes MQL5 or Pine Script.
- Tactix Guide explains the step you are on — what to fill, what a control means, or how to phrase a rule — without dumping untested source code.
That is form-first automation: the LLM never replaces Code Generator, and you keep plan limits and real-time validation.