Independent simulated wallet
Perpetuals Research ↗
Stocks, options and Savings Portfolio. Switching views does not stop trading.
SYNTH84 // SEPARATE MARGIN WALLET
PAPER TEST · LIVE OFFPerpetuals // Validation
BTC + ETH. Real market data, simulated positions. This test does not spend your funded wallet.
BTC Spot · Paper comparison WAITING FOR DATA
Actual BTC prices, simulated coins. Each model starts with its own $100 paper wallet; up to $20 per position. No real funds or broker orders.
Independent simulated wallet
No observations yet.
Simulation assumptions + limits
EMA pullback entries, 2× ATR stop, 4× ATR target, completed-trend reversal or time exit (6h fast / 24h baseline). Entries wait for the next observed ask; exits use an observed bid, with 0.01% adverse slippage each way. Assumed fees: 0.25% per side; stress: 0.60% per side. These are scenarios—not your verified account fees. No leverage or shorting. Misses, losses and data gaps stay recorded; gap-affected outcomes still affect paper cash but are excluded from scored P&L. Separate wallets must not be added together. Neither model has proven an edge or verified real fills.
Revised forward test · cost-calibrated trend WAITING FOR DATA
Separate experiment. Old history stays intact; no live trading.
Training uses new completed outcomes, including no-trades and losses.
Tests whether the next 15-minute move follows, fades or ignores the previous trend. Uses only earlier completed observations. Both candidates pay assumed spread, fees and slippage; no-trade baseline $0. Forty examples start forecasts—not live approval. Snapshot fills, funding and liquidation remain unvalidated.
Forward qualification CHECKING EVIDENCE
Training, forecast accuracy and after-cost performance are separate checks. No check here grants order authority.
Cost-aware strategy lab
WAITING FOR DATAThree separate paper candidates · 15-minute forecasts · estimated costs × 1.5 entry hurdle · no-trade baseline $0.
Checking forecast collection. No paper entries does not mean collection has stopped.
Entry checks + latest forecasts
Waiting for 30 minutes of continuous observations.
Entry checks + latest forecasts
Waiting for 30 minutes of continuous observations.
Entry checks + latest forecasts
Collecting future outcomes—not backfilled predictions.
Uses only completed, same-market observations. After 40 training examples, it compares 24 similar past conditions. This starts prediction testing, not live approval. Probabilities and snapshot fills remain unvalidated.
Learning progress unavailable.
Forecast scores unavailable.
How these candidates differ
Trend tests continuation when 5-minute and 30-minute moves agree. Range reversal tests a return toward the mean in a choppy regime. The learning model uses volatility-normalized patterns and only outcomes known before its forecast. Each candidate has its own one-contract-per-market simulation, not a combined allocation.
Forecast movement must exceed 1.5× visible spread, current taker-fee assumptions and slippage. Funding windows, stale costs and missing observations are not treated as free trades. Results include double-cost stress and excluded outcomes. These hypotheses are not validated live strategies; no wallet funds are spent.
Original five-minute baseline History preserved · different horizon, not a like-for-like ranking
Settled real funds in the separate margin wallet—not simulated P&L or guaranteed available margin. This paper test does not place real orders.
Hypothetical closed-position P&L after spread, assumed taker fees and slippage. Snapshot depth is not a verified fill. Late or unverified funding outcomes are excluded and shown separately.
Profitable scored positions divided by all scored positions, including flat outcomes. Open and excluded positions do not enter the denominator. Win rate alone does not qualify a strategy.
BTC—
ETH—
Live readiness —
Live execution is not connected. Verified access or wallet funding alone cannot activate trading.
Experiment rules + data quality
Frozen test v1: BTC/ETH, one simulated contract per market. A 0.1% reference move over about five minutes selects long or short; otherwise no trade. Decisions run every 15 minutes, with a five-minute target hold. No historical backfill.
Both legs include current taker-rate assumptions and one tick of slippage. Stress P&L doubles fees and slippage. Funding crossings are skipped or left unscored. These are research assumptions—not validated live execution.
Stress result unavailable.
Checking public-data collector
Waiting for a snapshot.
Checking the read-only account receipt.
No trade count alone unlocks live mode. Held-out performance, execution modeling and a tested live adapter are still required. Perpetuals API documentation ↗
SYNTH84 trading operations terminal
Measured signals. Defined risk. Synthetic Trading Engine
SYNTH84EDGESIGNAL SEEKING
SYS 84-26ALINK ACQUIRING
New York market clock
ET · 24H
Next session
Next bell in--:--:--
Awaiting synchronized event tape
operatorviewer
loading
market state checking
Open Synthy's read-only evidence console. Synthy is in shadow mode and has no order authority.
Turn terminal cues on or off. Volume, individual cues, and previews are available on the Settings page.
Turn terminal motion, including Synthy's animation, on or off. CRT intensity, density, and accessibility preferences are available on the Settings page.
System authority state
Market
checkingEngine
checkingNew entries
blockedBroker sync
checkingStock and crypto trading engine overview
STOCKS // ALPACA PAPER
Open stock cockpit ↓Stock Trading Engine
syncingNet liquidation—
Buying power—
Session P&L—
CRYPTO // KALSHI DEMO 24/7
Open crypto cockpit ↓Crypto Trading Engine
syncingDemo wallet equity—
Available cash—
Realized P&L—
KALSHI // PRODUCTION // REAL MONEY
offlineLive Crypto Lane
Production status unavailable. Live orders are not presented as active.
Production equity—cash + portfolio value
Production cash—separate from demo
Current / effective tierT— / T—evidence unavailable
SYNTH84 // ISOLATED 24/7 DEMO MARKET
CRYPTO TRADING ENGINE
BTC, ETH, and XRP run reviewed 15-minute paired-edge execution in the isolated Kalshi demo wallet. XRP uses its own evidence policy. DOGE runs beside them as a production-public observer only, with no order authority and no effect on execution evidence.
connectingexecution lockedisolated demo
Markets observed0 / 4
Two-sided books0 / 4
Qualified execution pairs0
Demo connectionchecking
Orders this cycle0
Execution ledgerchecking
Bitcoin // 15M
demo execution marketWaiting for the first sanitized demo snapshot.
Ethereum // 15M
demo execution marketWaiting for the first sanitized demo snapshot.
XRP // 15M
separate demo execution policyWaiting for the first sanitized XRP demo snapshot.
Dogecoin // 15M
production-public observer · no ordersWaiting for the first sanitized DOGE public-market snapshot.
Advisory operator preflight
Operational readiness
Combining fresh broker evidence, order certainty, protection references, risk budget, and research gates. The engine remains the sole source of trade authority.
checkingReview event tape
Waiting for enginechecking
The first synchronized snapshot has not arrived.
Execution protection incident
Incident pendingNew entries are locked
Safety reason pending.
ScopeAll new entries are blocked.
Automatic actionMonitoring and risk-reducing exits remain active.
Latest broker evidenceWaiting for reconciliation.
Net liquidation value
—
starting equity pending
Buying power
—
cash and leverage pending
Session gross realized P&L
—
loss budget pending
Loss sequence
—
shutdown threshold pending
This week gross realized P&L
—
New York week pending
All-time gross realized P&L
—
paper fill ledger
Open Book // Managed positions
checking Broker reconciliation pending.equity guard checking
Ranked inside the active universe
Top 3 live signal graphs
cycle pending
Rolling one-minute observations. Charts explain rank; they do not place orders.
Waiting for ranked price observations.
Independent control lenses
Three-way trading audit
Signal evidence, execution integrity, and portfolio risk are scored separately. One green card cannot override another card's block.
Audit 01
collectingSignal edge
Buy labels—
5m avg—
Win rate—
Evidence gate pending.
Audit 02
checkingExecution integrity
Filled—
Rejected—
Incidents—
Broker-state and data-freshness controls pending.
Audit 03
checkingPortfolio risk
Budget used—
Position cap—
Closed trades—
Session loss controls pending.
Risk Monitor // Paper Controls
loadingOperational status stays here • profile selection lives in Settings • no profile change resumes the engineoptions checking
Loading active profile
Manage in SettingsWaiting for current paper-risk limits. This terminal view is read-only.
Intraday equity—Per-trade and session budget.
Equity swing—Gap-adjusted overnight exposure.
Options premium—Maximum premium at risk; long contracts only.
Owner paper control / deliberate override
loss count pendingReset consecutive-loss pause
Existing positions remain monitored and protected while paused. This queues a one-time request to reset only the consecutive-loss count; the engine stays paused until its next clean cycle revalidates the broker evidence and accepts the request. It does not erase session P&L, trades, history, or safety evidence. Daily-loss, market-hours, stale-data, timing, and safety-freeze pauses cannot be cleared here.
Activity Gates // Paper Tuning
active profileLower thresholds create more candidates; they do not increase dollar risk or bypass stale-data and safety gates.
Equity watch—
Equity entry—
Full-size score—
Momentum floor—
ATR floor—
Option quality—
Option direction—
Option scan cadence—
Source of truth:
[strategy] and [options] in config.paper-smoke.toml. Changes require an engine restart and remain paper-only.Latest entry scores
Options Lab // Researching
sizing research · 1 / 2 / 5 contractsOffline ask-to-bid experiments; larger sizes and broader filters do not change execution approval.
research status loadingCollecting liquid-chain observations with no order authority.Methodology + safety ↗
End-of-day Swing Desk
Profit Vault // Long-Term Paper Portfolio
checking plan10% weekly · same paper account · cash-funded only · purchases held
This week’s allocation preview
—
Estimate only · nothing reserved or transferred
Prior losses to recover
—
New net-profit highs fund the plan, not gross wins
Funded by this savings plan
—
Existing holdings stay separate · targets are not purchases
Waiting for accounting evidence.
Personal picks // Your watchlist
Watchlist only · no allocation and no automatic buying.
How the weekly rule works
Review each Monday in New York time. The preview applies 10% to new cumulative realized-profit highs after configured share-fee estimates, carrying prior losses forward. Of that allocation, target 60% VTI, 30% VXUS and 10% uninvested cash. Fractional-cent remainders stay in cash. Target amounts are illustrative, not booked investments.
The approved plan uses the same paper account, capped by available cash and buying power—not margin borrowing. It is not a separate protected wallet. Kalshi demo funds cannot fund this account. The planned quarterly review directs new contributions toward underweight holdings; it is not an active rebalancing service.
Before purchases: verify broker net-of-fee accounting and complete duplicate-safe recurring purchases and lot reconciliation. This release does not reserve cash or place orders. Your personal picks have zero allocation and do not change VTI/VXUS weights. This stock-heavy portfolio can lose value; it is not guaranteed savings.
View existing long-horizon strategy ↓Long-Horizon Starter Book
checkingSmall fractional paper positions with an independent allocation cap.
News Sentinel // Live Intelligence
gate checkingConnecting licensed and official evidence sources.
Checking engine health.
Sentinelstarting
Alpaca liveconnecting
Nasdaq haltsconnecting
SEC filingsdisabled
Stream received0
Stored evidence0
Commodity ETF Desk // Paper Trading
paper armedGLD, SLV, USO, DBA and DBC use the equity signal, bracket, stale-data and reconciliation controls. Direct commodity futures are not connected.
User Backend // Access Control
local authBroker keys remain environment-only and are never stored in user records.
Safety & execution event tape
Waiting for the first event snapshot.Exact repeats will be grouped.