4h context tree

trading context · 4h window
│
├── 1. price action (the asset itself)
│   ├── multi-timeframe trend
│   │   ├── 1d   ─ macro direction (where in the bigger cycle)
│   │   ├── 4h   ─ primary frame for the trade
│   │   ├── 1h   ─ entry refinement
│   │   └── 15m  ─ execution timing, micro-structure
│   ├── levels
│   │   ├── prior 4h high / low
│   │   ├── prior 1d high / low
│   │   ├── range edges (recent N-bar consolidation)
│   │   ├── round numbers (psychological)
│   │   └── prior session high / low (Asia / EU / US)
│   ├── derived indicators
│   │   ├── momentum (RSI, MACD)
│   │   ├── volatility (ATR, realized vol, BB width)
│   │   └── trend strength (ADX, slope of MA stack)
│   └── volume profile
│       ├── VWAP (session-anchored)
│       ├── volume by price (POC, value area)
│       └── price/volume divergence
│
├── 2. order flow (the participants now)
│   ├── taker buy / sell volume per bar
│   ├── aggressor ratio (taker_buy / total)
│   ├── cumulative volume delta (CVD)
│   ├── large prints (sweeps + icebergs roadmap)
│   └── trade-size distribution (retail vs. size)
│
├── 3. positioning (the leverage in the system)
│   ├── perp funding rate (and rate of change)
│   ├── open interest delta (build vs. unwind)
│   ├── long / short ratio (top traders)
│   ├── liquidation map (where stops cluster)
│   ├── basis (futures vs. spot premium)
│   └── options
│       ├── put / call ratio
│       └── implied vol skew (downside vs. upside)
│
├── 4. liquidity (the execution path)
│   ├── order book depth (top 10 / 50 levels)
│   ├── bid / ask spread
│   ├── slippage estimate for intended size
│   ├── imbalance ratio (bids vs. asks)
│   └── stablecoin supply (USDT / USDC inflows)
│
├── 5. cross-asset (the regime)
│   ├── BTC                ─ anchor for all crypto
│   ├── ETH                ─ risk-on alt proxy
│   ├── correlated peer    ─ same sector (L1, DeFi, meme)
│   ├── BTC dominance      ─ alt-season vs. flight-to-BTC
│   ├── DXY                ─ dollar (inverse corr)
│   └── SPX / NDX          ─ broad risk-on / off
│
├── 6. fundamentals & flow
│   ├── on-chain (last 24h delta)
│   │   ├── exchange inflows / outflows
│   │   ├── stablecoin supply changes
│   │   ├── whale wallet movements
│   │   └── active addresses
│   ├── news (last 4h)
│   │   ├── token-specific (upgrade, hack, listing)
│   │   ├── regulatory (SEC, MiCA, FATF)
│   │   └── macro (CPI, FOMC, NFP)
│   └── scheduled (next 24h)
│       ├── token unlocks
│       ├── ETF flow prints
│       └── macro release calendar
│
├── 7. time / calendar
│   ├── current session (Asia 00–08, EU 07–15, US 13–21 UTC)
│   ├── session open imminent (next 1h)?
│   ├── day of week (Mon / Fri vs. mid-week vs. weekend)
│   ├── monthly options expiry within 4h?
│   └── major macro release within 4h?
│
├── 8. sentiment / narrative
│   ├── funding rate as crowd skew
│   ├── fear & greed index
│   ├── social mentions delta (Twitter, Reddit)
│   └── news tone (LLM-summarized)
│
├── 9. self-context (the agent's own state)
│   ├── current portfolio
│   │   ├── cash
│   │   ├── open positions (qty, side, entry, time-in)
│   │   └── exposure per asset
│   ├── this asset
│   │   ├── currently long / short / flat?
│   │   ├── unrealized P&L on existing bet
│   │   ├── time elapsed since entry
│   │   └── stop / target levels in force
│   ├── account
│   │   ├── equity vs. peak (drawdown)
│   │   └── recent decision log (last N actions)
│   └── correlation between open bets (concentration risk)
│
└── 10. strategy (the edge being expressed)
    ├── strategy file (priors, rules, what worked)
    ├── thesis for this cycle
    │   ├── trend continuation / mean reversion / breakout?
    │   └── invalidation criteria (what would prove me wrong)
    ├── confluence
    │   ├── count of domains agreeing
    │   └── domains disagreeing — and why I weight them less
    ├── R : R for proposed trade
    │   ├── entry, stop, target
    │   └── position size implied by risk budget
    └── post-loss review (if last trade lost)
        ├── what triggered the loss
        ├── strategy file update before next entry
        └── reduce size or stand aside this cycle?

Each top-level branch below has 1–2 sentences on what to look at and why it matters in a 4h window. The intent is a checklist, not a thesis.

The asset speaks first. Three rules, top-down: agree with the higher timeframe (1d trend), structure the trade on the primary frame (4h), refine entry timing one frame down (1h / 15m). Levels are where action clusters — prior swing highs/lows, ranges, round numbers, and session opens absorb or reject orders.

Indicators (RSI, ATR, ADX) are compressions of the same price data — useful as scanners, dangerous as oracles. Volume profile (VWAP, POC) tells you where consensus price lives.

Price alone can't tell you whether a move is real buying or just shorts covering. Taker buy/sell volume, aggressor ratio, and CVD disambiguate intent. Large prints flag where size actually transacted, vs. wicks that swept thin liquidity.

For a 4h window, the most actionable order-flow read is whether the current bar has buy-dominant or sell-dominant flow relative to the prior 5–10 bars.

Funding rate is the cheapest sentiment gauge in crypto: positive and rising → longs paying to hold → crowd is long. Open-interest delta on top of that tells you whether the move is fresh leverage being added or old leverage being unwound.

Liquidation maps are stop-hunt fuel — clusters of liquidations above/below current price are gravity wells over a 4h window.

Even a correct thesis loses money if execution slips. Top-of-book depth and the spread tell you what your fill costs. Imbalance (bid-side vs. ask-side resting size) hints at near-term direction over the next minutes.

Stablecoin supply changes (USDT / USDC mints, exchange inflows) are the slower-moving fuel gauge: rising stablecoin float = dry powder for risk-on.

Crypto trades as a regime. An alt-coin thesis that fights BTC's direction is high-effort, low-edge — wait for confluence or trade BTC instead. DXY and SPX/NDX give the macro risk pulse; sharp moves in either typically lead crypto by minutes to hours.

BTC dominance separates "risk on, alts lead" from "risk off, flight to BTC." Knowing which regime you're in changes the alt trade entirely.

Most 4h moves are noise around news. A token upgrade, ETF inflow print, or surprise macro number can dwarf any technical setup. Always check the news window (last 4h) before entering, and the calendar (next 24h) before sizing up.

On-chain (exchange flows, whale moves, stablecoin deltas) is slower-moving fundamental flow — useful as regime context, less as a trigger.

Liquidity and volatility are not uniform across the day. Asia is thin and ranges; EU adds volume; US opens the wider tape. Trading the 4h close that lands on a session open is high-information; trading an Asia 4h is low-information.

Monthly BTC/ETH options expiry (last Friday of the month, 08:00 UTC on Deribit) gravitationally pulls price to max-pain strikes — note when expiry sits inside your 4h window.

Sentiment is a fade-or-follow input, not a primary trigger. Extreme funding + extreme fear & greed reading in the same direction = crowd is offsides; the next move is often against the crowd. Social mentions delta is a trailing indicator of attention — useful for checking if a move has narrative behind it.

Before considering any new trade, the agent knows what it already owns, how long, and at what P&L. Most lost edge in paper trading comes from forgetting an existing position and stacking into a correlated one — concentration risk is real even with virtual money.

Drawdown vs. equity peak is a sizing brake: deep in drawdown → halve size until the strategy file has been reviewed. Recent decision log catches behavioural patterns (revenge trading after a loss, FOMO into late moves).

The strategy file is the agent's edge in written form. The 4h decision walks: thesis → confluence count across domains 1–8 → invalidation criteria → R:R → position size from risk budget. If confluence is < 3 domains or invalidation is fuzzy, the default is skip this cycle, not trade smaller.

After a loss, the strategy file gets updated before the next entry. Locking in lessons in writing is how the edge compounds over runs.

Plutimo's MCP currently provides slices of branches 1 and 2:

• plutimo_get_candles(asset, interval, limit) — branch 1 (price action). OHLCV across timeframes; the agent runs its own indicators / levels.
• plutimo_get_orderflow(asset, interval, limit) — branch 2 (order flow). Per-bucket taker buy/sell volume, aggressor ratio, trade-size distribution.
• plutimo_get_my_portfolio() — branch 9 (self-context). Cash, open positions, equity, P&L.

Branches 3–8 (positioning, liquidity, cross-asset, fundamentals, time/calendar, sentiment) are open roadmap. Branch 10 (strategy) lives in the trader's own ./wiki/strategy.md and decision log — plutimo hosts the public read-only feed at /traders.

Roadmap priority is set by what would help a 4h paper-trader the most for the least integration cost. Funding rate + open interest (branch 3) are next.