How to use MIZAN.

00 · Start here

Five things, in order. The rest of this guide can wait until you have watched the engine run.

1. Add it. Once your access is granted, open TradingView: Indicators, then Invite-only scripts. You do this once.
2. Open any chart you already trade. Futures, stocks, ETFs, crypto, forex, any timeframe. The channel adapts to the chart it is on.
3. Pick the preset. Settings, section 1, Preset: Balanced. This is the only setting most people ever need to touch.
4. Check the clock. Settings, section 9, Session Timezone: set it to your exchange. On crypto and forex the session filter switches itself off.
5. Watch for a week before you trade it. Note what fires, what gets refused, and why. The engine explains itself on the chart.

01 · The one idea

MIZAN draws a channel around fair value and grades what happens at its edges. When price drifts outside the channel, that is a setup candidate. The engine scores it 0 to 100 on the confirmed close, refuses the weak ones right on the chart with the reason, and keeps score of both, the fired and the refused.

One sentence to carry through everything below: conviction is stacked evidence, not a win probability. It sits in the tooltip of every score on the chart because it is the most important sentence in this guide. The evidence families behind the score are named on the methodology page; the calibration is not, and never will be.

02 · The bands, and how I trade them

The channel is a live estimate of equilibrium: two bands and a mid line. Price spends most of its time inside. The moment this engine was built around is when price drifts outside a band, not a touch, a drift beyond it. Here is how I trade that, step by step.

Why this works often enough to build an engine around: price that has stretched away from equilibrium tends to come back toward it. That is the premise, not a promise. The receipts on your own chart show you how the premise is holding on your instrument, this month, in this regime. Read them.

03 · Reading the marks

• Triangle + number — A graded setup. The number is conviction on a 0 to 1 scale. Hover it for the evidence list and the reward multiple.
• Three dotted lines — Entry, stop, target, with the R multiple on the target.
• x-cross — Blocked by the phase gate: a confirmed breakout or hard momentum. Fading here is refused outright.
• Small cross — Blocked by conviction (evidence did not clear your preset) or by geometry (no room: reward-to-risk below your minimum). The label says which.
• Small o — Direction lockout. A recent signal this way just failed, so the engine sits out a few bars rather than re-enter a losing idea.
• Continuation arrow — The opposite of a fade: a pullback into a flipped band that held on quiet volume. These show up precisely when fades are being refused.
• Pinch and Flare — Volatility compressed into a squeeze, then released. The bars after a Flare are an expansion window.
• Phase ribbon — The strip along the bottom colours every bar by market phase, so you can see where conditions changed.

04 · The dashboard, one line each

• Phase — Ascent, Descent, Equilibrium or Unstable. Trend up, trend down, range, or transition.
• Balance — Where price sits in the channel. 0% is the lower band, 100% the upper.
• Volatility — Coiled, Releasing, Shocked, Expanding or Settled.
• Pressure — The session’s cumulative buy-versus-sell estimate against its own recent scale.
• Clarity — How much identifiable structure recent movement has. Noisy means little to work with, which is information.
• Gates — Six dots: session, news, grind, squeeze, lockout, cooldown. Filled is open, hollow is blocking.
• Conviction — The strongest setup currently on the table. Confluence, never win probability.
• Gate edge — Average R of fired minus refused, once both cohorts pass 20 samples. Positive means the refusing is earning its keep.
• Ladder — Institutional only. Which order-flow tier fed this bar, and the bar’s delta as a share of its volume.
• Bottom line — The dominant condition in plain English. A description, never an instruction.

05 · The receipts

Turn on Show Stats Table in settings section 7. Every signal is logged with its entry, stop and target, then resolved: stop hit, target hit, or expired after your resolution window. Every signal the gates refused is logged too, resolved against the geometry it would have had. Those trades were never taken. The point is the Gate edge line: proof, on your own chart, of whether the refusals are helping.

What the table is not: it says so in its own footer. Signal resolution on this chart’s history, not trading results. If you want to watch it resolve in real time before you trust it, there are nine unedited sessions on the chart page, losing days included.

06 · Settings you will actually touch

The menu is numbered. Work top to bottom, and stop at section 3 unless you have a reason to keep going.

• 1 · Start here — The preset. Conservative fires least, Aggressive most, Balanced is the recommended start. While a preset is selected it overrides the conviction gate and the reward-to-risk gate everywhere else, on purpose, so a half-changed menu cannot produce behaviour nobody chose. Want more or less activity? Move the preset one step before touching anything else.
• 2 · Signals — Evidence toggles: sweeps, divergence, model blending, session filter, bar-close confirmation, label size.
• 3 · Risk — The reward-to-risk gate, the stop buffer, the target mode. No room, no trade, no matter how good it looks. Changing the target mode changes every R multiple the engine reports, so change it knowingly.
• 4 · Filters — The momentum gate that stops knife-catching, continuations, grind filter, lockout, cooldown, squeeze filter, and the news blackout. There is no automatic calendar: you type the times.
• 5 · Order flow — Institutional only. Leave the resolution on Auto.
• 6 · Levels — Prior day high, low, settle, overnight high and low, session VWAP, and nine manual gamma slots. Levels affect scoring whether or not you draw them.
• 7 · The receipts — Section 05 above.
• 8 · Appearance — Purely cosmetic. Turning things off changes what you see, never what the engine decides.
• 9 · Advanced — Session hours and timezone, the higher-timeframe filter (completed bars only, it never repaints), cross-market confirmation, the volatility anchor, and the JSON alert switch.
• 10 · Calibration — Two print-altering switches, both off by default. Leave them until you have watched default behaviour long enough to compare.

07 · Institutional order flow

Institutional reads the buying and selling that actually happened inside each bar instead of inferring it from the bar’s shape. It uses the best data your account can see and steps down automatically if a source goes quiet: native exchange footprint, then a reconstruction from intrabar data, then the geometric estimate every chart can produce. The Ladder row always names the tier that fed the current bar, because a reading is only worth what its source is. Stacked imbalances at the band, delta conviction, absorption into a sweep and unfinished business at an extreme each add evidence to conviction, on confirmed bars only. Institutional is the second tier on the pricing page; everything else in this guide is in both.

08 · Alerts and automation

Every meaningful event has its own alert condition, all firing on bar close. To automate: turn on JSON payloads in section 9, create one alert on Any alert() function call, and point the webhook wherever you want. Every signal then delivers a structured payload: engine, tier, version, signal, ticker, timeframe, price, conviction, pattern, reward multiple, stop, target, regime, and on Institutional the order-flow tier and delta that fed it. The payload names its data source for the same reason the dashboard does.

09 · Questions people actually ask

Does it repaint?

Signals confirm on bar close when Confirm Signals on Bar Close is on, which it is by default and should stay for live trading. The higher-timeframe and cross-market checks read completed bars only. Live-bar dashboard readings update until the bar closes, like every live reading everywhere.

Why is it so quiet?

Because most bars do not deserve a trade, and the engine is built to say so. Want more activity? Move the preset one step toward Aggressive rather than editing individual gates.

Which markets and timeframes?

Any liquid market, any timeframe you already trade. Futures, stocks, ETFs, crypto, forex, seconds to weekly. On 24-hour markets the session filter and the cross-market check disable themselves.

Why does my friend’s chart show a different conviction on the same bar?

Different loaded history, different tracker state, or on Institutional a different intrabar-resolution setting. See the honesty notes in section 05.

What does MIZAN not do?

Predict, promise a win rate, read your risk tolerance, or know about news you did not type in. Conviction is confluence, not probability. Anyone who tells you their indicator’s score is a win probability is selling something the math cannot deliver.

10 · If something looks wrong

• The stats table shows different numbers than yesterday — It rebuilds from the bars loaded on your chart, every load. Scroll-back depth and plan history limits change the sample.
• Signals near the session boundary behave oddly — Check that the Session Timezone in section 9 matches the exchange you trade. The engine reads your settings, not your intentions.

Anything beyond those two: send a screenshot with the settings gear open to qasim@mizanquant.com and it gets checked against the build.