I have been participating in the market since 1995 and joined BPT as a member in 2010.  Then a couple of years later I completely left trading and went into real estate investing.  I got in touch with Matt again a couple of years ago because I was interested in the KISS Systems.  This weekend I looked at my chart templates and re-analyzed what I should include in that template in terms of indicators.  I have had many chart templates throughout the years but it is good to reflect and re-question things once in a while.  One concept that I learned when I briefly interned for John Bollinger (inventor of the Bollinger Bands) is this concept called collinearity.  I searched this site and no one has ever brought it up so let me throw this out there (apologies if people already know this).

This is how AI describes it and I tweaked it a bit here also:

Collinearity...using multiple indicators that are derived from the same underlying data and therefore tell you the same thing.
Most commonly, this happens when several indicators are all built from price and time. They may look different on the chart, but mathematically they are highly related. When that happens, one is not getting confirmation, they are getting echoes.  Collinearity doesn’t make a chart template stronger — it makes it louder.

A simple example: Using MACD and RSI together feels like strong confirmation. In reality, all three are derived from price. If price turns, they all turn.

Collinearity is a problem because it creates three major issues:
    1.    False confidence – multiple indicators agree because they share the same input
    2.    Overweighting one variable – usually price momentum
    3.    Fragile decision-making – signals look robust in hindsight but fail live
If removing one indicator doesn’t change your decision, it wasn’t adding information.  

To avoid collinearity, each indicator should answer a different market question, ideally using a different data dimension.  Another example: A lot of times you'll see that people will have a Sentiment indicator added to their analysis...that is because they Sentiment comings from a completely different type of data.

Non-collinear indicator examples:
    •    Trend: Moving average or Ichimoku (price over time)
    •    Volatility: Bollinger %B or Ulcer Index (price dispersion or drawdown)
    •    Participation: Volume or money flow (CMF, OBV)
    •    Leadership: Relative strength (RRG)
Each adds new information, not repetition.  Confirmation only matters when it comes from independent data.  Different visuals do not mean different information.

So I went to my chart template set of indicators and used AI to help me determine what I should keep/add/remove.  I will have a weekly template and a daily template.  I am still working on the daily template, but I have a weekly one.

This is the weekly template that I am using to analyze MOH, Molina Healthcare, and a description of the indicators and what AI said about them:

[Inference] There’s no universally accepted numeric “collinearity score” for TA indicators across all markets/timeframes, but we can judge non-collinearity by what data they use, what question they answer, and what layer of the decision process they occupy. Using that standard, your stack is not collinear because each tool targets a different informational axis.

Your stack, with one-line “why it’s not collinear” reasons
    1.    RRG (Relative Rotation Graphs) — cross-sectional relative strength/momentum vs a benchmark; it ranks assets vs each other, not just the asset’s own price series.
    2.    Bollinger %B (50, 2.1) — position of price within a volatility envelope around a long MA; it’s a regime/location measure, not momentum speed or participation.
    3.    Chande Trend Meter (CTM) — trend persistence/quality composite; it answers “is this trending strongly enough?” rather than “where is price” or “how fast is it moving.”
    4.    MACD histogram (25,170,25) — very slow cycle-level momentum differential; it’s long-horizon alignment, not short-term timing or drawdown damage.
    5.    Chaikin Money Flow (CMF 20/21) — uses volume + close location to infer accumulation/distribution; it adds a participation dimension that price-only oscillators don’t contain.
    6.    BBWidth (50, 2.1) weekly — volatility regime/energy state (compression vs expansion) over a ~1-year horizon; it’s dispersion, not direction or momentum.
    7.    Ulcer Index (weekly) — drawdown depth + duration (“pain” path risk); it’s about the equity curve’s damage profile, not volatility width or momentum.
    8.    KST (default Pring) — smoothed composite momentum “health” across multiple ROC horizons; it filters broad momentum integrity rather than short-term acceleration.
    9.    ROC (12) — momentum speed/acceleration (fast thrust) over a short horizon; it catches early deceleration that slower composites (KST/MACD) miss.
    10.    BBWidth (20, 2.0) daily — short-horizon entry readiness (early vs late volatility expansion); it’s timing-quality via volatility state, not a trigger.
    11.    RSI (7) — bounded momentum exhaustion/stabilization for micro-timing; it’s about pullback “breathing,” not trend persistence or volatility regime.
    12.    5-EMA crossing the Chandelier Exit — execution/structure reclaim using an ATR-based trailing level; it’s a price/volatility structural trigger, not a momentum oscillator.

Where there is overlap (and why it’s still not “collinear”)
    •    MACD vs KST vs ROC vs RSI all touch “momentum,” but they’re intentionally separated by time constant and function:
    •    MACD(25,170,25) = macro alignment (slow)
    •    KST = momentum health (smoothed composite)
    •    ROC(12) = thrust/acceleration (fast)
    •    RSI(7) = micro exhaustion/turn (bounded timing)
    •    %B vs BBWidth both come from Bollinger math, but they measure different axes:
    •    %B = location inside bands (direction/regime bias)
    •    BBWidth = band spread (volatility compression/expansion)
    •    Chandelier (ATR) vs BBWidth both relate to volatility, but:
    •    Chandelier = volatility used for structure/stop/trigger
    •    BBWidth = volatility used for state/readiness

Bottom line

Your stack isn’t collinear because it spans distinct, orthogonal dimensions:
    •    Relative leadership (RRG)
    •    Direction/regime (%B)
    •    Trend persistence (CTM)
    •    Macro momentum alignment (slow MACD)
    •    Participation/flow (CMF)
    •    Volatility regime & readiness (BBWidth weekly/daily)
    •    Path-risk / drawdown damage (Ulcer Index)
    •    Momentum health vs speed vs micro timing (KST vs ROC vs RSI)
    •    Execution/structure (EMA ↔ Chandelier)

Would be interested to hear others' thoughts on this topic.