The one-indicator hypothesis.
The map comes first.

Before a new trading week begins, the previous week's high, low and close are complete. Those three inputs define the next weekly pivot map, including the full 0.702 zones. The coming week's path remains unknown. Its reference areas can already be marked.

That timing difference is the foundation of the TradeLean one-indicator hypothesis. A moving average or oscillator is also available at the opening of the week, but its future readings depend on the prices that follow. The active pivot map can keep its coordinates while the market moves towards them. This creates time to prepare alerts, scenarios and invalidation conditions before the encounter.

The decision to test is what confirmation adds to that preparation. Does another indicator reduce failed setups, improve execution or sizing, or identify a condition that the map and price response miss? Its benefit must be assessed alongside the opportunities discarded and the distance consumed while waiting.

The investigation follows that sequence: the geometry of the 0.702 reference, the timing of the map, the price response and the contribution of extra indicators. An interactive catalogue compares 100 named tools by their information and timing. A separate exploratory screen examines 100 fixed parameter configurations. The inventories and methods are defined separately so each comparison can be inspected.

Preparation comes first. Confirmation should have a measurable job. More lines are not automatically more information, and more information is not automatically a better decision.

Watch the full walkthrough.

The published video puts the pivot map beside RSI, MACD and moving averages on the charts discussed below.

Leading in preparation.
Conditional in prediction.

A useful definition begins with the information clock. At the start of a new reference period, the previous high, low and close are already completed. The corresponding pivot levels can therefore be calculated before the new period develops. The future close, the path into a zone and the market’s response are not yet known.

That is the relevant sense in which the roadmap is leading: the address is specified before the encounter. It is not a claim that the calculation contains future prices. Both statements can be true at once. The reference can be genuinely available in advance, while the outcome at the reference remains uncertain.

Moving averages illustrate the distinction well. They summarize an evolving price history. An oscillator can identify a stretched or changing condition before a subsequent turn, even though its calculation uses observed prices. Ichimoku even plots certain calculated spans forward on the chart. Moving a plot into the future is not the same as obtaining future information. Their coordinates come from earlier observations, even though the display extends beyond the current candle. Ichimoku calculation reference ↗

The distinction is operational rather than promotional. A map answers where attention should be organized. A trigger answers when a particular rule allows action. A momentum measure answers something about the move that is developing. Judging all three by a single label conceals the different jobs they perform.

Two boundaries.
One fixed area of attention.

The roadmap begins with the high H, low L and close C of a completed reference period. Its range is D = H − L, with D greater than zero, and its central pivot is P = (H + L + C) / 3. For the active period, those completed inputs define both boundaries of each outer zone.

The whole area is available before the next encounter. Its boundaries do not need a new oscillator extreme or a moving-average crossover to exist. Within the active reference period, an unchanged source history gives an unchanged band. Price can approach it, enter it, leave it and return without requiring the band to follow those moves.

The range normalization makes the map portable in price units. A larger completed range produces a wider zone; a smaller range produces a narrower one. That is a defined scaling rule rather than a claim that every asset has the same response probability. The symmetric support and resistance constructions also let the two sides be evaluated on the same geometric terms.

A useful property of the centre.

Because C lies between L and H, P lies between L + D/3 and L + 2D/3. The central references P ± 0.702D therefore sit beyond the previous high or low by at least (0.702 − 2/3)D, approximately 0.035333D. This is an exact bound under the stated input assumptions.

The whole band needs its own description. Its 0.618 inner edge can extend inside the preceding range by up to approximately 0.048667D, while its 0.786 outer edge is always beyond that range under the same assumptions. The zone can straddle a previous extreme. A theorem about the centre should not be extended to every point of the area.

This is also why the empirical comparison uses complete shifted bands of equal width. Two nearby lines inside the target area do not constitute two competing zones. To evaluate whether this particular band is preferable, compare its encounter frequency, information overlap and subsequent responses with those of the other whole bands.

TradingView’s standard Fibonacci pivot implementation lists 0.382,0.618 and 1.000 extensions around its pivot. The selected 0.618–0.786 outer bands and 0.702 centres are the TradeLean framework examined here. Source sessions, adjustment conventions and the reference-period setting remain part of the definition. Standard pivot formulas and input conventions ↗

See what was available.
Then let price arrive.

The replay below uses archived market bars rather than an invented winning path. Choose an instrument and reveal completed bars one at a time. The pivot map stays fixed for the selected week. EMA9, EMA21 and SMA50 update with the close of each revealed bar. The readout reports RSI and the MACD histogram at that same information point.

All examples use the fixed calendar week of 14–20 September 2026. Choosing one calendar window for illustration does not make these five charts a representative performance study. It does make the comparison inspectable. You can see a move that reaches an area, a move that stops short and a reference that is simply not involved.

At the first revealed bar, the map is already known because the prior week has finished. The indicator readings are also available for that completed bar. What is not available is the future path. That is why the tool does not display unrevealed candles or let a final moving average leak backward into an earlier decision.

The right comparison is not that a moving average “does not exist” when the week opens. It exists. What does not yet exist is its future value after an unknown series of new closes. The pivot reference for the active week does not need those closes. Its future address within that period is already specified.

Watch how much of the interpretation is available without the overlays. The outer band is the same with the EMA switched on or off. A new moving-average condition may add information about the approach; it does not create the location. This distinction gives every extra line a specific burden of proof.

These are cached Binance, Gate and Yahoo observations from the existing research collection, depending on instrument. They are not a live feed. Equities follow observed exchange-session bars, while crypto trades around the clock. A displayed four-hour session block can therefore represent different elapsed trading time across the two groups. The full input provenance is included with the downloadable files.

Same setup.
What did the extra filter change?

The direct comparison uses 1,725 eligible price-response setups across the frozen 69-instrument availability universe. Each original opportunity is evaluated without an indicator gate and with six explicitly defined conditions. Each condition is tested as an immediate gate and as a wait-for-confirmation policy. Every branch keeps the original opportunity in the denominator.

This audit answers a narrower, testable question than “which indicator is best?” It asks whether a particular extra condition improves one defined response to the full 0.702 band. No first-touch profit rule is assumed. The first completed close inside the band is followed by an inward close within six bars, without an intervening outer-boundary touch. The response is known before the entry is scheduled at the next native open.

The comparison keeps the same price map.

The stop is the fixed outer boundary, the target is the central pivot and the observation endpoint is up to 12 native bars after the response, capped at the end of the active map. The baseline and delayed policy share that endpoint. A later confirmation cannot obtain a longer window merely because it arrived late. If the original stop or target is touched before confirmation, the waiting setup is cancelled.

One unit, R, is the distance between the original baseline next-open entry and its stop. Every policy uses that same original unit. Waiting does not receive a newly favourable risk denominator. A skipped setup scores zero; it is not removed. The result is a mean hypothetical score per original opportunity, not an account return, compounded equity curve or recommendation to take these trades.

Both-hit candles remain in the study. The main calculation assumes the stop was reached first, with an alternative target-first bound shown as a sensitivity check. Adverse stop gaps use the opening price. The cost scenarios are 0, 10 and 25 basis points of roundtrip notional. They are modelling assumptions shared across the comparison, not claims about the actual spread, fees, borrowing or financing of every asset.

Participation is part of the answer.

The primary 4H/weekly sample contains 862 opportunities across 68 instruments. At the 10 bp scenario, the unfiltered baseline averages -0.060R per opportunity. The table keeps that same baseline for every row. A less negative filtered average may simply reflect spending more time in cash; that must not be mistaken for a strong profitable trading system.

The EMA9/21 gate retains only 25 of 862 opportunities. It skips 474 baseline losers, but also 363 of the 370 baseline winners. That is strong evidence that this particular directional gate is highly restrictive for the defined reversal response. Its small loss after costs is not a demonstration that it discovered a superior active strategy.

The lighter RSI30/70 state gate retains 88.1%. It excludes 67 baseline losers and 36 baseline winners. Its pooled change is 0.018R per original opportunity. The corresponding equal-asset change has an interval spanning zero, so this audit does not establish either dependable improvement or equivalence. Here the condition is a state—RSI at least 30 for a long, or at most 70 for a short—not a requirement that a new threshold crossover occurred on that exact bar.

Waiting has a measurable price.

The MACD signal-state wait policy executes 353 of 862 opportunities. Of those, 297 are delayed entries; their median delay is 3 native bars, with a median 1.26% price move in the response direction before entry. That last percentage is conditional on the delayed entries, not every setup, and is not a net return.

It also skips 366 original losers and 143 original winners. The overall effect can be decomposed exactly into baseline losses avoided, baseline gains missed and the change on the cases that were still executed. Treating only the delayed successful charts as “the cost of confirmation” would miss the first two parts.

The cost assumption can reverse the ordering.

With no modelled costs, the pooled baseline averages 0.023R and RSI50 wait averages 0.017R. At 10 bp, those scores become -0.060R and 0.007R. At 25 bp, both pooled scores are negative: -0.183R and -0.010R. Taking fewer trades reduces modelled transaction costs, which is part of why the ordering changes.

The later chronological slice, from 2025 onward, contains 615 primary-grid opportunities. At 10 bp its baseline score is -0.041R and RSI50 wait is -0.001R. This is reused historical evidence, not an untouched prospective holdout. The model does not demonstrate a reliably profitable unfiltered strategy, nor does an uncertain near-zero filtered result establish a durable edge.

Read the coverage before generalising.

The daily/monthly and daily/weekly comparisons contain 532 and 303 setups respectively. The stricter weekly/quarterly, monthly/yearly and weekly/monthly alignments contain only 14, 9 and 5. Those higher-timeframe observations remain visible as small descriptive samples. They do not justify broad statements about every timeframe.

For the all-history primary comparison, 18 assets have at least 20 baseline opportunities and qualify for the equal-asset summary. Their paired uncertainty intervals resample both assets and shared calendar-quarter blocks. The table also retains pooled results from all 68 instruments in that configuration. These are different summaries with different populations. The later slice does not meet the same per-asset requirement; an unavailable interval is not evidence of zero effect.

The supported practical conclusion is specific. The pivot map supplies the address without an oscillator, and several familiar trend gates discard almost the entire defined response population. Confirmation can consume movement. It can also avoid losses and reduce costs. The complete paired record gives each of those effects a number; it does not support declaring all additional indicators useless.

Inspect the complete summary CSV, fixed protocol and code, all paired event rows and source hashes. The execution assumptions belong to this study. TradingView’s strategy documentation illustrates why bar timing, simulated fills and costs matter; it does not independently validate these results.

The map was ready.
Watch what the panels added.

A useful example needs more than a coloured line through a convenient low. The location must come from the preceding completed week. The price response needs an observable definition. Every added condition must use the same completed-bar clock. The following four cases keep those elements separate.

For a lower band, the price response is a completed close back above its inner boundary after a close inside the zone. For an upper band, it is a close back below the inner boundary. These are explicit price observations, not fills assumed at the wick. The comparisons then ask when six familiar indicator conditions first agree at or after that response.

These are four selected positive illustrations, chosen after the outcomes were visible. The scan retained 57 close/reclaim candidates across BTC, ETH, SPY and AAPL from 2025 onward. The complete candidate table, selection criteria and source hashes and per-bar values are downloadable. The examples explain a preparation advantage; they do not estimate a representative success rate or prove that every touch reverses.

The price differences above are signed changes between recorded closes, not executable entries, missed profits or net trade returns. A condition absent before the week ends is censored there; the observation window is not extended until it eventually agrees. Indicators are computed from earlier cached history, not restarted at the beginning of each chart.

Remove a panel only after defining its job.

For the task of marking the next fixed reference area, another oscillator is unnecessary: the completed-period calculation already provides the coordinates. For the observed local responses above, waiting for every chosen directional condition would have delayed or excluded the observation. That is a concrete reason to avoid stacking confirmations by habit.

The claim becomes different if the added measure is intended to reduce failures across many encounters, size exposure or manage execution. Four favourable charts cannot settle those tasks. Their value here is to make the map-first process inspectable, with the full comparison record retained alongside them.

Definitions: RSI, MACD and moving averages. The calculations and selected thresholds used in these examples are listed in the downloadable specification.

A signal can be accurate about the move—and late for the decision.

A moving average is a weighted description of prices that have already been observed. In a simple linear trend with slope b per bar, an n-period simple moving average sits b(n − 1)/2 behind the latest close. That is a property of averaging an arithmetic sequence, not an estimate of future return.

For the conventional EMA smoothing factor α = 2/(n + 1), the steady-state lag behind that same indefinitely established linear trend is b(1 − α)/α, again b(n − 1)/2. The assumptions matter. A sudden jump, a curved path and the initialization of an EMA behave differently. The formula does not say every moving-average signal arrives a fixed number of bars after every reversal.

This is a practical cost when the decision requires an early price reference. Waiting for confirmation can mean an entry farther from invalidation, a smaller remaining distance to the next zone or a setup that has already completed most of its movement. The value of reduced false signals has to compensate for that cost.

The opposite case matters too. Acting simply because a predetermined band has been reached can produce repeated premature entries in a persistent trend. A confirmation rule may be useful precisely because it refuses some of those entries. “Earlier” is an information-timing advantage; whether it is an execution advantage depends on what happens next and what the rule costs.

A fair measurement therefore records both clocks: the moment the map becomes available and the moment the action rule permits a decision. It also records the price paid for the delay. A chart that marks only the final winning signal hides the failed opportunities and the distance consumed while waiting.

The same reasoning applies to VWAP. Its evolving price-and-volume average provides information the HLC pivot does not contain, but the averaging process can lag the current move. TradingView explicitly describes that limitation. This is a reason to define its job, not a reason to assume its volume input has become irrelevant. VWAP calculation and lag ↗

The 0.702 zone is a band.
The evidence must cover the band.

In this framework, “the 0.702 zone” means the complete interval from 0.618 to 0.786 times the previous reference-period range, offset from the central pivot. The 0.702 number names its centre. Treating a neighbouring line inside that same band as a competing zone answers the wrong question.

The lower zone is [P − 0.786D, P − 0.618D]. The upper zone is [P + 0.618D, P + 0.786D]. Both have width 0.168D. The full-zone study therefore compares the band with adjacent bands of exactly that width, and follows several distinct observation clocks: arrival, the remainder of the reference period, consecutive closes inside the band and every individual close.

The strongest directly supported RSI result concerns Bitcoin and Ethereum on four-hour bars with weekly pivots. In qualifying completed weeks that reached the lower zone from inside the map, the archived OHLC observations contain a price inside the zone at which reconstructed RSI14 was below 30 in 71 of 84 Bitcoin weeks and 70 of 85 Ethereum weeks. That is 84.5% and 82.4% respectively. These are full-band, period-level observations; they are not percentages of winning trades or percentages of every individual zone entry.

A witness uses an actually recorded open, high, low or close that lies inside the band. RSI at that price is reconstructed from the preceding completed native-bar Wilder state. The archive records that price, although it does not store the live RSI print or a complete tick sequence. The upper number allows any price in the intersection between the candle’s range and the band. It is a compatibility bound, not a confidence interval, and does not assume every intervening price traded.

The first arrival bar gives a different answer. Bitcoin has a witnessed lower-band RSI<30 price in 60 of 84 first-arrival bars, with a compatible maximum of 71; Ethereum has 55 of 85, with a maximum of 73. Later observations within the same week account for part of the larger period-level overlap. Calling all these observations “the hit rate” would hide that distinction.

The upper band provides the symmetric test. Bitcoin’s observed-price RSI>70 witnesses occur in 68 of 89 qualifying weeks reaching the upper band; Ethereum’s occur in 74 of 86. This is evidence of a relationship between outward price location and momentum state on the matched map. Its strength is empirical and can be inspected rather than assumed.

Across the sampled crypto instruments, 1,074 of 1,434 qualifying lower-band weekly encounters contain an oversold witness somewhere within that week: 74.9%. Pooling all sampled asset classes gives 1,388 of 2,780, or 49.9%. Those are deliberately visible populations. The crypto result is not silently extended to equities, indices or every timeframe.

Consolidation explains much of the difference between a band encounter and a bar-by-bar state. Bitcoin has 70 oversold closes among 403 completed closes inside the lower band, compared with 523 oversold readings among 10,473 valid closes overall. Its completed close-based episodes inside the band include 51 oversold episodes out of 155. A price can arrive stretched, then remain in the same fixed zone while RSI recovers. The reference remains useful while the momentum state evolves.

The controls are complete neighbouring zones. In the pooled four-hour/weekly lower-side sample, oversold closes account for 12.1% inside the inner neighbouring 0.450–0.618 band, 20.8% inside the target 0.618–0.786 band and 29.3% inside the outer 0.786–0.954 band. The gradient supports a relationship with downside extension. It does not establish that the target band is uniquely optimal for every purpose. A farther band can concentrate more extreme readings while being reached less often and allowing more movement to occur before arrival.

This trade-off is central to a useful comparison. Extreme-state concentration, encounter frequency, preparation time and the subsequent response are different objectives. Optimizing only the first could favour waiting for a very distant band. A roadmap instead exposes the whole sequence of references and allows the decision rule to state which trade-off it accepts.

The study attempts 69 assets and six chart/pivot combinations. Arrival analysis excludes incomplete reference periods, missing native state and native candles that straddle a parent reset. Completed-close profiles are retained independently, and right-censored close episodes are reported separately. The higher-timeframe arrival samples are consequently small in several combinations. All input conventions and exclusions accompany the downloadable data. Wilder RSI calculation and default inputs ↗

Which indicators repeat the location?
Which describe something else?

A useful challenge to an indicator stack begins with a defined state. On a support-side encounter, price below a moving average is directional agreement; RSI below 30 is an extreme condition. These are different standards. A high directional-agreement rate cannot be relabelled an oversold rate or a profitable-entry rate.

The full-band calculation measures 85 fixed configurations from 17 directional or momentum families, using windows 9,14,20,50 and 200. It compares their completed-bar conditions inside the entire target band with their wider valid sample. Each feature keeps its own denominator so that a long warm-up or a missing volume feed does not silently remove observations from every other indicator.

The first three rows illustrate a concrete simplification. A lower outer-band location often already accompanies the price-side or slow-momentum condition those features display. They can describe a state that is highly associated with the map location. The shorter oscillator and volume-flow rows show why the same inference should not be extended to every function. Their state can change substantially while price remains in the band.

The MACD row uses the screen’s fixed fast 14/slow 28 construction. It is not a claim about every histogram, signal-line crossover or default 12/26/9 implementation. The stochastic row uses raw range position rather than every possible smoothed stochastic variant. Details at this level prevent a large feature count from becoming an exaggerated claim about complete trading systems.

Several apparently different panels also repeat one another by construction. With the same high–low window, raw Stochastic%K is 100 times Donchian close position, and Williams%R is Stochastic%K minus 100. A reading below 20, below 0.2 and below−80 is the same range-position condition under those definitions. Likewise, the MACD line and a PPO built from the same two positive averages have the same sign. These are mathematical relationships within the indicator stack; they do not imply that the full pivot input recovers every oscillator value.

ATR, standard deviation and efficiency ratio are excluded from this direction/extreme table because their natural jobs are different. ATR can inform movement scale, standard deviation describes dispersion and efficiency ratio describes path directness. Their contribution belongs in a volatility, sizing, exit or regime test. The broader exploratory screen retains them, with its narrower direction target explicitly stated.

The evidence therefore supports selective consolidation. A pivot map can be the primary display for price location and preparation. A redundant confirmation can be removed if the complete decision process survives that removal. A genuinely different data layer should be evaluated on the task it is intended to improve.

One hundred tools.
Different information clocks.

There is no single public, cross-platform census that establishes an uncontested “100 most popular indicators” ranking for every asset and trading community. This catalogue is an explicitly curated set of 100 familiar tools and indicator families. Its purpose is to compare what each tool observes, when its reference becomes available and whether that reference remains fixed.

The distinction immediately improves the hypothesis. Most smoothing and momentum tools update as price develops. Some alternatives also provide references known before the next encounter: previous-period highs and lows, classic pivots and frozen Fibonacci drawings are examples. Other tools use information outside an HLC pivot, including traded volume, displayed orders and open interest. These differences help separate information that changes the decision from information that simply repeats the current price condition.

The evidence audit marks every entry individually. Twenty-one catalogue names map to 20 computed feature families, four entries have archived component or task tests, and 75 are architecture reviews without a numerical head-to-head test of that implementation. The two ATR entries share one computed family. Open “Evidence coverage” on any card to inspect its status.

The catalogue is an architectural comparison, not a performance league table. A tool tagged “updates” is not declared unprofitable. A tool tagged “fixed after input closes” is not declared predictive. Several entries are packages, measures or drawing frameworks rather than one standardized algorithm. Proprietary or implementation-dependent tools are marked accordingly.

This prevents a common counting trick. SMA20 and SMA50 are different parameter choices within one family. Stochastic position, Williams %R and a normalized Donchian position can encode closely related range information. One hundred columns can be much less than one hundred independent observations about a market.

The underlying function names and categories can be checked in the TA-Lib reference directory. The platform’s Technical Ratings methodology provides another concrete example of how platforms combine families and parameter lengths. Neither source establishes the popularity ranking or the superiority claim proposed in this article.

One map can replace clutter.
It cannot retain every property of the path.

The most useful simplification is selective. A pivot roadmap retains the completed period’s range and closing position. It deliberately ignores much of the sequence of transactions inside that period. That is why its output is compact, stable and easy to organize across timeframes.

Two histories can finish with H = 110, L = 90 and C = 100 and therefore produce exactly the same pivot grid. Yet the last three closes can be 90, 110 and 100 in one history, or 90, 90 and 100 in another, with the high of 110 reached inside a candle in the second history. SMA3 is then 100 in the first case and 93.333… in the second.

The conclusion is mathematical: no universal function of only those three completed HLC inputs can recover every moving average for every admissible history. The order of gains and losses also affects RSI. Volume-weighted measures require volume. Order-book measures require orders. Choosing not to display those measures is a workflow decision, not a proof that their inputs are encoded in the pivot.

This does not weaken the case for a clean chart. A pilot does not need every measurement on one display at once. The useful question is whether the omitted measurement changes a particular decision often enough to justify the extra attention. A compact map can be the right primary interface while other information remains available for a different task.

In the TradeLean workflow, the map organizes location and scenarios. Price action supplies the current response. Risk controls constrain the consequence of being wrong. None of those jobs requires five moving averages to agree before the relevant zone is even noticed. Whether another measure improves the response can then be tested directly.

A broader computational check.
A narrower claim than “best indicator.”

For this review, a separate reproducible screen was run on 100 fixed configurations: 20 calculation families at lengths 9, 14, 20, 50 and 200. The families include common averages, RSI, rate of change, MACD/PPO, stochastic measures, CCI, Bollinger position, volatility measures and volume-based transforms. The exact definitions are included in the research download.

This count is transparent. It is not 100 distinct independent indicator families, and it is not the 100-tool catalogue above. Stochastic and Williams %R are affine transforms of the same range position under these definitions; Donchian position overlaps that information too. Parameter variants are correlated. A larger column count must not be mistaken for independent evidence.

The target is deliberately modest: the sign of the close-to-close move over the next three native bars. Models use information at the current completed bar. A fixed ridge linear-probability specification compares pivot features alone, one indicator feature alone and pivots plus that feature. Scaling is fitted on the first 60% of common usable observations; the last three training rows are purged and the final 40% form the retrospective test segment. Probabilities are clipped before Brier scoring.

The common-history rule is strict. Every configuration in a comparison uses the same eligible observations. Long warm-ups, including the nested 200-period smoothers, leave evaluable comparisons for 59 assets in four-hour/weekly and 62 in each of daily/monthly and daily/weekly. The common-case requirement also excludes the foreign-exchange feeds from this new screen because its volume-based features need usable volume. The remaining weekly and monthly configurations did not meet the common-history requirement. They were attempted and are displayed as unavailable, rather than silently removed from the scope.

The output is a set of descriptive prediction losses, not a simulated return series. Lower Brier loss means better probability accuracy for this one direction target. It does not assess the proper job of every volatility or volume tool. ATR can be useful for position sizing without being a good standalone forecast of whether the next three closes finish higher.

The useful contribution is transparency: the full 300 aggregate cells and asset-level results can be inspected. Small positive incremental differences and negative differences remain in the data. There is no unsupported claim of statistical significance, no retrospective declaration of a winner across all purposes and no conversion of a fitted probability into a guaranteed trade.

A genuine supremacy test would have to go further: preregister each tool’s intended use, freeze definitions and parameters, retain a new sample, control repeated comparisons and evaluate the complete decision process with executable costs. The current screen is a reproducible challenge to the hypothesis, not a certificate that a preferred tool has defeated every alternative.

Give each observation a job.

The practical workflow begins before a trigger. Select the instrument, venue, session and pivot reference period. Record the completed input high, low and close. Mark the relevant bands. Identify the level that would change the current interpretation. That work can be done while the market is still far from the area.

Next comes the approach. A fast move into the outer band is different from a slow grind, even if both end at the same central line. A first encounter differs from repeated tests. A lower-timeframe rally inside a larger-period support band differs from a rally into resistance on the larger map. None of those observations requires a new oscillator panel; all of them require the price path and the relevant timeframe context.

An extra indicator then receives a concrete question. Does it reduce a particular failure mode? Does it improve sizing? Does it identify an execution problem that price and pivots miss? Does the benefit survive the signals it discards? If the answer cannot be stated, the indicator is functioning as reassurance rather than a tested part of the process.

That is where Lean thinking belongs. Removing unnecessary confirmation reduces cognitive load and makes the remaining rules easier to follow. It does not authorize removing inconvenient observations from the research. A clean chart and a complete evidence record are compatible—and should be designed together.

The phrase “one indicator” is therefore best understood as a primary interface. It includes a market feed, a price chart, a timeframe convention, explicit conditions and risk management. It does not mean that one horizontal line replaces judgment, data quality, execution or the possibility of loss.

A price address is not a measurement of orders.

The pivot map tells us where a reference lies. It does not directly measure how much displayed depth, stop exposure or liquidation risk sits there. Those quantities require their own data sources. A bright cluster on a liquidation heatmap is not interchangeable with resting executable liquidity, and a historic volume profile is a record of trading rather than a live inventory of orders.

The earlier TradeLean liquidity investigation found locally elevated reported exposure in specific asset–timeframe–side combinations. Its complete major-asset matrix included 23 positive and 57 nonpositive mean comparisons. That evidence can support targeted monitoring of a zone; it cannot support a claim that every 0.702 band always contains exceptional liquidity.

This distinction protects the one-map argument from an unnecessary promise. The map can still organize a liquidity investigation even where no exceptional pocket is present. Its value is a stable coordinate system for asking the question. The heatmap supplies an observation about the current or archived risk at that address.

Explore the liquidity investigation and its interactive maps ↗

Likewise, ATR remains a volatility measure rather than a directional forecast. Turning it off the main chart does not make volatility cease to matter. Its information might belong in a sizing rule or risk dashboard instead. A good interface displays the right information at the right stage. ATR’s defined purpose ↗

A fixed formula still needs a causal implementation.

The mathematical claim assumes the previous period is genuinely complete. A script that reads an unconfirmed parent candle, leaks future higher-timeframe values or redraws a historical projection can violate that information boundary. “Uses pivots” is not by itself a complete non-repainting audit.

Source conventions matter. Regular and extended sessions can produce different highs and lows. Futures settlement prices can differ from the last intraday trade. Adjusted price histories can change after corporate actions. A feed revision can alter previously recorded inputs. For comparisons, save the venue, session, price-adjustment setting and timestamps alongside the grid.

The archived native-bar replay deliberately reveals completed bars only. It does not claim to know the intrabar order of a high and low. At a gap through a level, an OHLC candle can show that the nominal level lies between extremes without proving a trade occurred at that exact price. Those details are particularly important when converting an attractive diagram into an entry rule.

TradingView’s documentation distinguishes live-bar fluctuations, historical plotting changes and future-information leakage. The useful objective is not to attach a blanket moral label to every updating indicator. It is to state which values were final at the decision point and reproduce them. Repainting and information availability ↗

What the matched audit leaves open.

The matched audit above freezes one band-response rule, entry clock, stop, target and observation horizon for its paired comparisons. A stronger general claim still requires a new sample, additional predeclared decision rules and independent replication. A fast indicator and a slow indicator should not be allowed different favourable exit assumptions merely because their signals appear at different moments.

Measure advance preparation separately from execution: when was the reference known, when was the setup eligible and when could the decision actually be acted on? Then compare false signals, missed opportunities, distance consumed while waiting, adverse excursion, net outcome and time spent exposed. Include the unfiltered pivot process and price-action-only controls, not only extra indicators chosen after an attractive result.

The candidate list also needs to be frozen. Testing 100 variations and reporting the best of them creates a selection problem even when every calculation is individually correct. Sullivan, Timmermann and White studied this problem for technical trading rules. Bailey and co-authors formalized related risks of backtest overfitting. These are reasons to design the experiment carefully, not reasons to abandon systematic chart research. Technical-rule data snooping ↗ Backtest overfitting ↗

Equivalence requires its own test. If the claim is that removing an indicator does not materially damage a useful outcome, specify a non-inferiority margin before the new sample is observed. A wide confidence interval that includes zero does not establish that two approaches are interchangeable.

Finally, separate the market functions. Trend, momentum, volatility, liquidity and execution are not one target variable. Earlier research by Lo, Mamaysky and Wang found that some technical patterns contained incremental information in their historical stock sample. That is not validation of this pivot framework, but it is a reminder that a universal statement dismissing all technical measures has a much higher evidential burden than a claim about a leaner personal workflow. Foundations of Technical Analysis ↗

One primary map.
A clear reason for every extra line.

The foundation is established by the calculation itself: completed reference-period inputs produce a map available before the next period develops. The full outer bands can be marked before price approaches them. That preparation advantage does not depend on selecting a successful historical reversal.

The full-band record supplies a second, empirical observation. In the matched BTC and ETH four-hour/weekly examples, qualifying weeks reaching the lower zone frequently contain an oversold RSI state; many completed closes inside it also share the measured trend conditions. These results explain why several panels can describe an encounter that the map had already located. Their incremental usefulness still depends on the decision being made.

The matched decision audit adds the missing outcome comparison. Its primary setup rule has a negative pooled score after the hypothetical 10 bp cost. Additional gates change participation and can reduce losses and cost exposure; the eligible-asset intervals do not establish a dependable advantage or equivalence. This evidence cannot support a blanket claim that every other indicator is unnecessary.

The studies do not establish that pivots contain every other data layer, that every touch reverses or that a strategy is profitable. The 100-tool catalogue and the 100-configuration screen make the comparison inspectable, including cases where additional information changes the result.

The TradeLean proposition is a map-first process: know the relevant price areas before the encounter, let the price response determine the interpretation, and retain an additional indicator only when it demonstrates a useful contribution.

For that process, one primary indicator can be enough to keep the chart coherent. Whether it is enough for a particular trading decision is the test—not the assumption.

Questions behind the one-indicator hypothesis.

Earlier diagnostics.
A separate observation unit.

The historical checks below use nominal reference contacts or earlier setup definitions. They remain inspectable as secondary evidence; they are not substituted for the new full-band measurements.

Inspect the definitions.
Keep the evidence.

The research package includes the full matched decision audit, all 20,700 paired rows, the fixed protocol, source hashes and verification, together with the four selected case studies, their code, all 57 close/reclaim candidates and the selection specification, alongside the full-zone code and specification, asset-level band profiles, the 85-configuration state arrays, the 100-tool evidence audit, the broader screen and the archived comparison tables. Source code, verification checks, exclusions and coverage records remain together. It contains no exchange credentials or private account data. The replay inputs come from the locally cached sources recorded in the manifest. No independent source refresh or outside replication is claimed.