Limitations and Trust Boundaries

A pane this configurable can produce readings that look crisp and mean less than they appear to. This page is the longest explicit list in the pack of ways that happens — not disclaimers, not "consult a professional," but the specific interpretation mistakes careful readers make and how to avoid falling for them twice.

Most of these are not bugs. They are honest consequences of the choices the indicator makes to get comparability across timeframes and instruments. Knowing them is how you earn the right to trust a reading instead of just being impressed by it.

1. The alignment trap

The most dangerous misread in the pack. If you remember one thing from this page, remember this one.

All MACD Slots Bullish fires when every enabled slot whose values are present is on the bullish side of its own signal. That sounds like consensus. It is not consensus when the slots share ingredients.

Consider a reader who enables slots 1 through 8 with the same source, the same MACD MA family, the same fast and slow lengths, the same signal length, and only varies the timeframe. Each slot is computing the same convergence story on a different cadence. Those slots do not independently agree when they align; they rediscover the same persistent underlying momentum at eight sampling rates. The alignment alert fires often during steady trends and goes quiet during chop, and both behaviors are about the underlying signal's persistence, not about eight independent observers voting the same way.

Said plainly: ten slots from the same ingredients produce one vote echoed ten times. The alignment alert is information about how correlated the slots are, not evidence that multiple things agree.

How to keep this honest. If you want alignment to carry weight, differentiate the slots deliberately — different MA families, different sources (close vs. hlc3 vs. ohlc4), different fast/slow length pairs, different timeframes, different tickers. The less each added slot shares with its neighbors, the more meaningful their agreement becomes. Conversely, if you only differ by timeframe, treat All MACD Slots Bullish as a nicely phrased "the trend is persistent right now," not as multi-layer confirmation.

How to verify. Configure slots 1–4 identical except for timeframe. Watch the alignment alert across a strong trend and across a choppy session. Observe that it fires freely during the trend and sits quiet during chop — and that neither behavior tells you anything the 5m slot alone did not already.

2. ATR collapse can pin a slot

The normalization divides each slot's raw MACD, signal, and histogram by that slot's own ATR. When ATR collapses — overnight on a thinly traded instrument, during a holiday session, in low-volume minutes between news prints — the denominator is small, so even small MACD values normalize to large numbers. Large numbers push the sigmoid toward its asymptotes, and the slot line can pin at 0 or 100 on what is actually a tiny price move.

Pinning is not a signal. It is a statement that the slot's MACD is large relative to an unusually small ATR. When the session normalizes, the ATR catches up and the slot un-pins, often without any dramatic change in price.

How to verify. Watch a slot pin during overnight on an instrument with light after-hours volume. Cross-check by opening a second pane with raw ATR on the slot's timeframe and confirming the collapse. The underlying move in the price chart should be modest.

How to handle. When a slot is pinned and ATR looks unusually small for the session, discount the slot's reading until liquidity returns. Do not fire orders off boundary pinning during those sessions.

3. Cross-asset slots reflect their own symbol, not the chart

A slot with Optional Ticker = "SPY" reflects SPY's normalized MACD on that slot's timeframe, computed in SPY's own ATR. The line is drawn in the chart's pane because that is where the indicator lives. The evidence behind the line is SPY.

That means:

• A cross-asset slot is context, not confirmation of the chart symbol.

• If that cross-asset slot is given meaningful weight in the blend, the blend is a mixture of two instruments wearing one number. Blended MACD Is Bullish on a configuration where SPY carries 30% of the weight is partly a statement about SPY. It is not a pure claim about the chart instrument.

• If the cross-asset market is closed while the chart market is open (crypto vs. equities on a weekend), the cross-asset slot's value can freeze or step strangely because its underlying bars are not updating. That is not repaint; it is session reality.

How to verify. Set a slot to Optional Ticker = "DXY" while charting a stock. Observe that the slot's line often moves contrary to the chart's action, because DXY is doing its own thing.

How to handle. Use cross-asset slots deliberately, name why the context matters, and prefer weight 0 (observer mode) unless you have a reason to let the cross-asset signal steer the blend.

4. 70 and 30 are not RSI thresholds

The reference lines at 30 and 70 look like RSI brackets. They are not. This pane is a stretch read on top of MACD/signal/histogram, not a measure of overbought-ness. The indicator does not fire alerts on crossing those levels — by design.

A slot fast sitting at 90 on a violent instrument during a real move is not equivalent to a slot fast sitting at 90 on a quiet instrument on a flat day. Both are saying "this slot's MACD is several ATR-units above zero," but the word "several" is scaled by the slot's own ATR and by ATR Sensitivity. The slot's color is the layer that tells you whether MACD is above or below its signal. That is not oversold-vs-overbought mechanics. Treating 70/30 crossings as reversal triggers is a documented misread, not a clever use of the pane.

How to handle. Use 70/30 as orientation brackets. If you want mean-reversion mechanics, use an oscillator designed for that. If you want to know whether MACD is stretched far from its zero line in ATR units, read the distance from 50 and remember the sigmoid is saturating — the marginal information above 70 or below 30 decreases as the slot approaches the asymptote. If you want the MACD-vs-signal posture, read the line color.

5. Master smoothing is a visual filter, not a correctness improvement

When master smoothing is enabled, the blended fast, slow, and histogram are each run through one additional MA pass before being plotted and before the blend alerts read them. The pane gets visually calmer. The underlying read does not get more accurate; it gets slower.

That has two practical consequences:

• The blend alerts (Blended MACD Is Bullish / Bearish) now fire on the smoothed values and therefore fire later than they would on the raw blend. They are lagged alerts by construction.

• Turning on master smoothing does not make the slots smarter. They are unaffected. Reading the slot lines is still the better way to see the real-time state of each contributing context.

How to handle. Use master smoothing when you want a regime-style read of the blend. Do not treat a smoothed blend alert as an early entry. Do not treat the visual calm as stronger evidence.

6. Weight-zero slots count toward alignment

This one surprises readers often enough that it belongs in its own section.

Setting a slot's weight to 0 removes it from the blend (the weighted-average skips zero-weight slots). It does not remove it from the alignment count. All MACD Slots Bullish inspects every enabled slot whose values are not na, and weight is not part of that test.

If you add an observer slot at weight 0 that tends to disagree with your three default slots, expect the alignment alert to go silent more often. That is the observer doing exactly what you added it to do. If you add an observer at weight 0 that tends to agree with your three defaults (same ingredients, different cadence), the alignment alert will fire more freely — and, per section 1, that is correlation not consensus.

How to handle. Decide up front whether observer slots should participate in alignment. If not, disable them rather than zero-weighting them. The difference matters.

7. Mild blend with split slots

The blend is a weighted average. When contributing slots disagree — say two slots above 60 and two slots below 40 — the blend sits near 50 and its histogram is small. A reader new to the pane sometimes reads that as "the blend is saying the market is quiet." It is not. It is saying the contributing slots are split.

The slot lines are the place to see the split. The blend is the summary, not the evidence. If the blend is mild and the slot lines are arranged as a fan across the 50 line, the correct read is "the contexts disagree," not "the market is calm." Acting on the blend alone in that state is acting on an average that is hiding its dispersion.

How to handle. Whenever the blend sits near 50, look at the slot lines before concluding anything about posture. The pane is designed to let you see both layers at once precisely because this state is common.

8. Histogram above-and-falling is common and subtle

A full-green-to-faded-green shift means the blended histogram is still above 50 but it is shrinking. The convergence story is still on one side of equilibrium, but the acceleration has turned. Readers sometimes ignore the faded state because it is still "green," which is how a decelerating move gets mistaken for a continuing one. Same caution in the bearish pair.

How to handle. When the histogram fades, treat it as a pace change, not a direction change. A full green that persists and a faded green that persists tell different stories. Both are useful; they are just not the same story.

9. ATR Sensitivity is expressive, not corrective

Raising ATR Sensitivity makes the pane more reactive — slots reach the boundaries faster, histogram columns get larger sooner. It feels sharper. It does not add evidence; it magnifies the evidence already in the raw MACD relative to ATR. A chart that looked indecisive at sensitivity 1.0 still has the same raw convergence story at sensitivity 3.0; the stretch is just more visually obvious (and the boundaries are reached sooner, which loses resolution at the extremes).

How to handle. Pick a sensitivity that gives you useful resolution across the boundary range you actually care about. If the pane pins often on your instrument, lower it. If the pane looks mushy, raise it. Do not read a sensitivity change as a change in the underlying read.

10. Reading the defaults as a recommendation

The default configuration — slots 1, 2, 3 at 5/15/60 with EMA 12/26/9 and equal weights — is a working starting point, chosen so the pane renders something useful on install. It is not a recommendation. Nobody on this end has run a study against your instrument, your session, your hold times, or your risk tolerances, and any manual that implied otherwise would be manufacturing authority it did not earn.

Where this bites: a reader installs the indicator, likes what the defaults look like on the first chart they load, and adopts them as the "supported" configuration — the one they will not question later. Then they carry those same defaults onto a fundamentally different instrument — an altcoin with wildly different volatility character, an overnight session with thin tape — and wonder why the pane feels wrong. The pane is not wrong. The defaults were never claimed to fit every context.

How to handle. Treat the defaults as a scaffold that lets you see the pane working. Then ask, in order: is 5/15/60 the right cadence for the timeframe I actually trade? Is EMA the smoother I want, or am I curious about ALMA or Jurik on this instrument? Is ATR Length = 14 giving me a yardstick that reflects the regime I am operating in, or is it averaging across a regime change I care about? The settings page walks each of those questions; this page is flagging that the question deserves to be asked in the first place.

If you have adopted a configuration and want to know whether it still fits, the cheapest sanity check is to load the pane on the same instrument two weeks later and see whether the slot lines are doing what you expected. If the pane now reads ambiguously where it once read clearly, the regime has probably shifted under your configuration, not vice versa. That is information, not a defect.

Where to go next

For the workflows where these boundaries most often bite, Workflows. For symptom-level troubleshooting on boundary pinning, runtime errors, and alert timing, Troubleshooting. For the mental model that makes the stretch-vs-overbought distinction intuitive, For the Geeks.