Section 01
The Starting Point — Your Terminal
First, the honest part: MMFP is a serious build. We spent real time in it, and this proposal starts from respect for what's already there — not a pitch to replace it.
What stood out in MMFP
- Liquidity mapping as a first-class surface — PDH/PDL, PWH/PWL, PMH/PML, PQH/PQL with BSL/SSL classification and swept-state tracking. Cleaner and more disciplined than how we surface levels.
- All-session ORB framing — opening-range bias across five named sessions with 2B/2Br classification. More complete than our Asia-only session model.
- Crowd-positioning aggregation — Fear & Greed, OI flow, top-trader L/S, taker B/S, funding history, basis & dominance, liquidations, all in one read.
- Legible scoring — your 13-signal bull/bear probability is transparent; a trader sees exactly why it's 38%. There's a lesson in that we'd happily borrow.
Track 1
DESIGN PASS
Our design system applied to MMFP — done, yours to keep
Track 2
AUTOMATION
The bot automation you're aiming at — we'd build it
Track 3
COLLABORATION
Optional — if our scanner's useful to you
How to read this
The three tracks are independent. Take the design work and walk away — it's yours either way. Take the automation build without the collaboration. Or, if you like what we've built, combine the two systems. No track depends on the others.
Section 02 · Track 1
The Redesign — Our Learnings, Applied to Your Terminal
We've built our own trading system over ~140 sessions and hit the exact wall you're up against: maximum alpha on screen without it becoming noise. Here's what we learned, and what we changed in MMFP because of it.
What we learned — hierarchy comes from light, not boxes
The principles that did the work
- Luminosity hierarchy — brightness encodes importance, not colour. Values bright, labels dim, units dimmest. This alone does most of the lifting.
- One field, no card fills — a single near-black background with hairline rules. Cards, borders, and shadows are the #1 clutter source; stripping them removes ~70% of the noise with zero data lost.
- One accent — a single highlight colour. Green and red mean direction only; the moment they decorate non-directional things, the screen stops being scannable.
- Tabular mono numbers — fixed-width figures so live ticks don't jitter their columns.
- Progressive disclosure — empty/loading panels collapse or show a skeleton, never a framed box of "--".
- De-duplicate — one datum, one home. (RSI / MACD / CVD each appear in three places in the current build — a third of the visual load for no new information.)
What we changed in MMFP
We reskinned the densest, most card-heavy region — clocks, price hero, ticker, Liquidity Hunter, Session ORB Bias, the Weekly/Monday/EMA-RSI trio, and the Signal Checklist — as a working proof:
The moves
- Card fills and frames → hairline rule dividers.
- Five competing accents → one accent + semantic green/red only.
- Values promoted to bright, labels dimmed, units dimmest.
- Every number set in tabular mono so the live columns hold.
- One reserved state colour for LIVE; signal rows as a tick bar + signed value, not pills.
What's in your hands now
A drop-in token sheet (CSS), a panel-by-panel migration guide, a rendered before/after preview, and a four-module lookbook. All adaptable to the MMFP brand — it's a starting grammar, not a mandate. This track is a gift to your terminal regardless of anything else here.
Section 03 · Track 2
The Automation Module — What You're Aiming At
You've said the future is bots automating trades. We can build that — for MMFP, on your terms.
Whose rule is whose
Our own system never auto-executes — that's a boundary we chose for our capital. It's a policy, not a technical limit. It has no claim over a product built for you, on your platform, with your risk policy. Automation is fully buildable for MMFP.
The design principle — the guardrails are the product
Auto-execution is only as good as the risk envelope around it. The valuable build isn't "place the order" — it's everything that stops a bad one:
- Hard limits — per-trade size, total exposure, concurrent positions, leverage cap, daily-loss lockout.
- A global kill switch — one control halts everything and, optionally, flattens.
- Regime / macro veto — automation respects the gates the scanner already computes.
- Paper-first promotion — runs paper until it clears a fills/outcome gate.
- Full audit + decision log — every action logged, nothing silent.
- Graduated autonomy — auto-with-confirm first, full-auto within the envelope once trust is earned.
Crypto — bot-native
Recommended first
Pionex is a strong fit — built-in bots plus API. Direct exchange API or middleware also work, and TradingView-webhook → exchange is already proven in our stack. The signal source can be your own alerts, our scanner, or both.
Equities / commodities
Later phase
Automating equities/commodities needs a brokerage API (Alpaca, IBKR) — a separate lift and jurisdiction. Realistic call: crypto-first, the rest later.
⚠ One flag worth raising early
If MMFP becomes a product for the Hackers community, automating trades on behalf of other people crosses into regulated territory — licensing, terms of service, risk disclosures, and they vary by jurisdiction. Not a blocker, but something to design around from the start. Personal-use automation for yourself is a much lighter picture.
P0
Discovery — answer the questions in Section 05
Personal vs product, target venue, risk policy, signal source.
P1
Paper bridge, guardrails-first · 2–3 wks
Risk envelope, kill switch, audit log, auto-with-confirm — paper-only until it clears a gate.
P2
Graduated live automation · 2–3 wks
Paper → live within the envelope, auto-with-confirm → full-auto, kill switch and daily-loss lockout live.
Section 04 · Track 3
Optional Collaboration — If Our Scanner's Useful to You
Only if you want it. You already have a scanner you trust. This is about whether ours adds anything to yours — and we think the two are complementary, not competing.
What our scanner is
It was our first build — everything else grew out of it. A cross-asset, thematic scanner (crypto perps, equities, commodities, grouped into themes), scoring each symbol on 25+ confluence factors → a conviction tier, gated by macro/regime, published as a ranked report with an AI review. Velocity-adaptive cadence (CALM ~4h / ALERT ~2h / HUNT event-driven), top-20 re-scored every 30s.
What we'd bring that complements yours
- Manipulation detection — stop-hunt / vacuum / fake-breakout / wash. You see where the crowd is; this flags when that crowd is being hunted.
- A proprietary geometry layer with tested edge — higher reversal hit-rate and profit factor vs fib alone, at statistical confidence.
- Regime-blocked transparency — signals that pass confluence but are macro-suppressed are shown with the exact block reason, and never auto-route. The seed of a safe automation model.
- Decision-surface discipline — one decision, one place, with execution-eligibility built in so the UI can't offer an impossible trade.
- Outcome testing — every signal's result tracked, so conviction tiers are measured, not asserted.
The complementarity thesis
You're deep on where liquidity is and what the crowd is doing. We're deep on how strong a setup is and whether it's a trap. Those two halves make a stronger whole than either alone. The four-module lookbook (attached) shows what our side looks like in practice — Floor, Signals, Hub, Action Deck.
Section 06
Where We'd Start
No agenda
The design pass is already done and it's yours. The automation is a concrete thing we'd genuinely enjoy building for a trader of your calibre. The collaboration is an open door, not a sell. Take any combination — or just the design and a good conversation.
The next artifact, if you want it
A one-page automation architecture — signal → guardrail → venue flow, the kill-switch and risk-limit design, and the paper-to-live promotion gate. That turns the automation track from an idea into something buildable. Say the word and it's yours.