Methodology
Last updated: April 28, 2026
A zone is a price level where multiple independent technical categories agreed on the same area, scored by how many agreed. Past data established how often price reached the next zone, reverted from a swept extreme, or held at a tested level. Those past rates are what get displayed. None of it predicts the future.
The Six Categories
Every "agreement zone" you see drawn on chart was placed because at least three of the following six categories agreed on a price area. The score is just the count of categories that agreed, normalized by what the script considers each category's reliability.
- Moving averages — multi-timeframe (5min, 15min, 1h, 4h, daily, weekly, monthly) clusters where two or more MAs converged.
- Fibonacci retracements — drawn from the most recent significant swing on each visible timeframe.
- Session levels — Asian / London / New York session highs and lows.
- Volume profile — multi-source aggregated for forex (FXCM + OANDA + Pepperstone + Eightcap + Capital.com), single-feed for equities and crypto.
- Fractal support / resistance — Bill Williams fractals validated as held levels.
- Momentum compression — multi-timeframe squeeze states marking volatility-contraction zones that historically expanded.
Why 86% Maximum
The visible score caps at 86. There is no seventh category. Six confluence categories at full agreement = 86 because the script weights each category by its independent reliability rather than treating all six as equally informative. The remaining 14 points are reserved for what the script does not measure: order flow, futures basis, specific options-positioning, news catalysts. Those are not part of this product.
A zone scoring 86 is rare — it is the top decile of confluence in the dataset, not a guarantee.
How the Probability Numbers Got There
When the on-chart narrative says "reach 64% backtested," that number was computed from a backtest that observed price behavior on 1,824 tickers across four timeframes (1min, 5min, 15min, 1h). The script picks the closest matching ticker class and timeframe, then reports the historical reach rate.
Reach is defined as: did price touch the named next zone within the timeframe-appropriate window, regardless of direction. That is the either-touch rate. Some narratives also frame favored-side reach (the closer zone in the named direction); favored-side rates are typically lower than either-touch.
Slope-based directional claims are gated. If the historical class rate for a slope continuation is below 55%, the script suppresses the slope narrative entirely rather than print a coin-flip number. Crypto on small timeframes typically falls below this gate.
Per-Ticker Tuning
The 21 most-traded tickers (BTCUSD, ETHUSD, SOLUSD, EURUSD, GBPUSD, USDJPY, XAUUSD, WTIUSD, NVDA, TSLA, AAPL, MSFT, AMD, AMZN, META, COIN, MARA, GOOGL, PLTR, QQQ, SPY) ship with backtest-derived rates baked into the script as a "DeepProfile." For these, the displayed reach percentage is the per-ticker observed rate, not a class average.
Custom-ticker subscribers can supply additional ticker symbols and receive per-ticker tuning generated from the same backtest pipeline.
What Got Tested vs Not Tested
Tested: zone-touch rates, sweep-revert rates, break-to-next-zone rates, slope continuation rates — all per asset class and per timeframe, on a fixed historical window. The numbers reported on chart are bounded by what these tests measured.
Not tested: live trading P&L, slippage, broker fill quality, account-level risk management. The displayed rates are level-touch rates, not trade returns. Whether a touch is a profitable entry depends on the trader's stop placement, position sizing, and execution — none of which the script knows about.
Honest Probability Framing
Every percentage in the on-chart narrative is past-tense by construction:
- "backtested" — the rate observed in the historical dataset
- "historically" — same, used for older or sparse samples
- "hist." — abbreviation when narrative space is tight
You will not see "will," "expect," "predict," "buy," "sell," "long," "short," "target," "setup," or any phrase suggesting a recommendation. The narrative describes where price is and what historical rates apply at that location. A daily linter scans every alert for forward-looking or directive language; the alert pipeline halts on a violation.
Calibration and Limits
Calibration is checked against live alert outcomes on a rolling basis. The aggregate calibration gap (observed minus claimed reach) was +5.1 percentage points across 5,797 v17.x alerts as of 2026-04-27, slightly conservative-direction. Bucket-level calibration is less precise — at the high-confidence claim band (60-89%), observed reach can run 20-30 percentage points lower for crypto and forex specifically. This is documented in the transparency page.
The structural fix for bucket-level calibration is a per-class × per-timeframe lookup table from the full backtest. That is on the roadmap as a separate companion product (working name: ADA), not an update to this indicator. The current indicator's compile budget cannot fit the lookup table.
Repaint Behavior
Zones are computed at session lock and do not move retroactively. Once a session closes and a zone is drawn, that zone's price levels and confluence score are fixed for the life of the chart. Replay mode in TradingView occasionally renders zones at slightly different positions than live charts due to a known TradingView platform behavior with multi-timeframe data — live charts are the source of truth.
Probability claims displayed in narratives reference the script version that fired the alert, captured in the |V:vXX.YY| meta tag at the end of every alert payload. A claim displayed under v17.20 will not change if the script is later updated to v17.30 — historical alert records preserve the version they were generated under.
Source Citation
Every numerical claim in the on-chart narrative resolves to a row in backtest_results/b1_full_dataset_1min.json through 1h.json. The full provenance for any individual claim — which dataset, which row, which formula — is reproducible from the script source plus the meta tags in the alert payload.
Marketing claims on this site reference the same dataset. The composite "agreement zone" framing is original to this product; the underlying technical categories (moving averages, Fibonacci, fractals, volume profile, session levels, momentum compression) are well-known traditional analysis tools.
Important: This page describes how the indicator computes its outputs and what its outputs mean. It does not promise any particular trading outcome. Past observed rates are not predictive of future rates, and the gap between historical reach percentages and your individual trading P&L is generally large. Use of this product is governed by the Terms of Service and the Refund Policy.
Questions
Methodology questions: hq@stickymoneyai.com