The edge is in the implementation.
MIZAN™ is built on three proprietary pillars, with a fourth layer in the Institutional tier.
The families are named.
The implementation is described.
The calibration is private.
01 · Adaptive equilibrium modeling
A self-correcting fair-price engine that fuses multiple volatility models in real time, weighted by which has been tracking price more closely rather than averaged.
The bands are not drawn from a fixed setting: width comes from a volatility forecast rather than a plain average range, and it is anchored to the prior bar, so a single outlier cannot widen the channel it is about to be judged against.
Model influence is a normalized share of recent tracking accuracy, exponentially smoothed, and the fused center weights each model by the inverse of its smoothed error variance — influence is earned across bars, never seized by one.
02 · Multi-regime validation
Every setup is tested against the current market state. A mean-reversion signal in a trending regime is not a mean-reversion signal; it is noise.
The engine reads which game is being played before it scores a move, and a setup that fails a gate is refused and marked with the gate that stopped it — a named output on the chart, not a silent absence.
03 · Evidence-based scoring
No directional price forecasts, and no implied probability of profit. The engine estimates state and volatility only to normalize the conditions a setup is judged under; it does not predict where price will go.
A 0–100 score measures weight of evidence on the confirmed close — not a win probability, not a forecast.
Seven validation layers combine bounded observations from distinct evidence families.
Those observations cover band position, momentum extension, participation, model agreement, regime, rejection quality, wick structure, divergence, higher-timeframe alignment, volatility expansion, cross-market confirmation, liquidity sweeps, and level proximity.
Each family is capped, so repeated expressions of the same market condition cannot become unlimited votes.
Two hard floors then apply: conviction, and reward-to-risk — a setup can carry heavy evidence and still be refused when the geometry does not justify the risk.
The score explains the setup. The gate decides admissibility. The block is the feature.
04 · Order-flow confluence INSTITUTIONAL TIER
Three data tiers keep the source of each read explicit: native footprint, intrabar reconstruction, and structural delta.
On supported TradingView plans, the engine reads the native footprint feed. When that data is unavailable, it reconstructs each bar’s footprint from lower-timeframe intrabars. The structural delta model sits beneath both.
When the native feed is live, the reconstruction is rescaled to its totals so the tiers report on the same scale.
Where supported data is available, the ladder steps down automatically, and the on-chart dashboard and every webhook alert name which tier fed that signal, because a reading is only worth what its source is.
One boundary stated plainly: MIZAN freezes its decision at the confirmed bar close, while TradingView classifies native footprint volume from intrabar price movement and may recalculate historical footprint data as available intrabar granularity changes.
With bar-close confirmation enabled, the decision is confirmed at the close. Historical recalculation can still reflect changes in available data, loaded history, or settings; the active data tier stays visible.
TradingView’s footprint data documentation ↗
Order-flow evidence is tracked as its own cohort in the on-chart ledger, against signals that had no order-flow backing.
That is the question this tier exists to answer, and it is answerable from your own chart rather than from anyone’s screenshot.
What we do not disclose
The exact parameter sets. The fusion weighting logic. The calibration thresholds. The proprietary extensions to standard models.
This is intentional, and it is the whole boundary. Everything above describes what the engine decides and why, which is enough to judge the approach before deciding whether the numbers are worth paying for. The vocabulary is public. The calibration is ours.
What it cannot do
It does not predict. It reads what has already transacted and grades it.
Native footprint requires TradingView Premium or Ultimate. Fallback data does not remove that platform requirement. For another plan, confirm compatible access before choosing Institutional. The panel identifies the data source used for each read.
Reconstruction quality falls on thin instruments where intrabar volume is sparse.
The on-chart ledger rebuilds from the bars loaded on your chart, so history depth changes the numbers.
Losses are counted conservatively: a bar that tagged both stop and target is recorded as a loss unless the intrabar path positively proves otherwise, and that setting ships switched off.
On authorship
The evidence families and footprint concepts named above are public methodology. The implementation is not borrowed: every line is written from scratch, and no code in it is taken from another author’s script, on any platform.