LIVE CONTEXT / HISTORICAL EVIDENCE
Market explorer.
Waiting for feedUSDT / USDC proxy —
Watchlist
USDT pairsThe pilot scans up to 20 chosen pairs. Changing the chart does not expand the scan universe.
BTC / USDT
— Historical + live candles
Scroll to zoom · drag to pan · UTC
Connecting to public market data…
Trade modeller
SIMULATEDSelect a research signal, or define your own long-only paper scenario.
Targets are scenarios, not forecasts. “Conservative” describes distance, not safety. Stops are not guaranteed. No real order is sent.
WATCH → STUDY → SIMULATE
Developing setups
LIVE PRIORITY / RECORDED EVIDENCE
Opportunities in motion
Fresh setups rise and cool with price, spread and age. No score guarantees a profit. Current live proxies use a 24-hour horizon; minute research does not automatically enable new live models.
RESEARCH CONTROL ROOM
Ideas in.
Evidence out.
Prepare a job here, leave the screen, and return to the results.
Phone or desktop. One server. One auditable queue.
Waiting for schedulerNo resource snapshot yet
YOUR SAVED MARKET DATA
Coverage by coin & candle size
See which prices you already have, how detailed they are and the dates they cover. A 1m candle describes one minute; a 1h candle describes one hour. Derived candles are built from the smaller candles you own.
Loading saved inventory…
What does this report prove?
The report reads saved-file metadata. A date range is not proof that every candle in between exists. It flags gaps between files and overlapping ranges; missing candles inside a file require the integrity audit. “Source checksums recorded” means ingestion recorded hashes, not that this scan rechecked all bytes. “Hourly cross-check recorded” does not certify a profitable model.
A scan runs through the normal queue so a large warehouse does not freeze your phone or compete with another heavy job. It calls no exchange API. The snapshot stays visible until you run a new scan. Daily and hourly history remain separate from the minute pilot and minute warehouse; duplicate logical rows are not added into a misleading unique-market total.
EXTEND THE WAREHOUSE
Bring in more data, then build the larger candles
1. Choose pairs and finished months.
2. Review the capacity estimate and authorise.
3. Queue and watch server progress below.
Downloads use Binance’s official monthly archive, serial requests and bounded retries—not thousands of live chart calls. Verified saved sources are reused. A disagreement with hourly history stays flagged; it is not silently corrected. Missing/pre-listing months stay explicitly unavailable.
Preparing only fills the reviewed job editor below. Nothing runs until you preview, enter a fresh authenticator code and queue it. Research uses the conservative, audit-gated pilot; it may need required BTC/ETH context to be included.
Prepare a calculation
VALIDATED RECIPESLoad the supplied job file or paste a JSON job we create together. Parameters and hypotheses are editable. Python, shell commands, arbitrary URLs and app changes are rejected.
Available research features
Nothing downloads until you queue a reviewed job. Jobs never alter the live prediction model or place orders.
Calculation queue
SERVER OUTPUT / RESUMABLE CHECKPOINTS
Select a calculation
On-screen times use London (BST/GMT). Copied feedback and downloaded source logs remain in UTC.
Output will appear here. Closing this page does not stop a queued or running calculation.
Earlier server research
Read-only status for the existing JPW research scripts. These are not re-launched or silently migrated into this queue.
Mobile HTTPS
Loading connection profile…
When the HTTPS gateway is configured, use the same EntiTrade username, password and authenticator code from phone or desktop. Port 8790 remains loopback-only; app updates still go through desktop ChangeGuard.
These are the server’s historical research files, read without changing them. Relative rank, correlation and model discrimination do not establish a tradable directional edge.
Evidence browser
Select a row or bar to read its explanation. The ? buttons also work with keyboard focus and touch.
Every live-rule signal is recorded, including signals you never trade. Outcomes use sampled bids. Feed gaps are excluded from clean outcome totals; historical research rows are not presented as prospective predictions.
First-target barrier outcomes
Target first / stop first / neither. This is an observed sample, not calibrated predictive confidence.
Independent event catalogue
Closed 1h candles above the preceding 20-hour high, whether a prediction existed or not. Only the enabled watchlist is monitored.
0 observed eventsPrediction records
Export append-only ledgerTargets below are the recorded prices and potential price changes if reached, before costs. Pending is the outcome status—not a missing price forecast.
Blue = low priority or no longer active → Red = higher rules quality and remaining potential. Not a success probability.
When does Pending change? What does High / Med mean?
The current research proxy observes a setup for 24 hours from its recorded time. A target changes when a later fresh bid first reaches it or the stop. At the deadline unresolved targets become Neither, or Incomplete if the feed had gaps. End direction is measured at that deadline, even if a target was reached earlier. Missing worker updates are never guessed.
Quality / potential is not certainty: quality High ≥80, Med ≥60, Low <60 on the recorded rules score. Potential High ≥6%, Med ≥3%, Low <3% for your selected target. Success confidence is unrated until measured and calibrated. These are display bands, not validated investment rules.
Heat combines the rules score, remaining lift from a fresh ask to the selected target (scaled to 10%), and time remaining. Stopped, resolved, stale or gap-affected rows cool to zero. A high lift alone cannot establish a good trade. Target levels exclude fees/slippage and are not guaranteed.
Actual events and missed signals
Virtual USDT only. Filled on later observed bid/ask snapshots with your fee and slippage assumptions. This is not an exchange-matching simulator. Unobserved feed gaps require manual review.
Paper account
HOW ENTITRADE KNOWS WHAT IT KNOWS
From raw prices to accountable evidence.
A guide to the data, models and limits behind every number. Read the summary first, then explore the detail.
Loading knowledge…
LEARN THE WORKSPACE
A guided tour. No orders placed.
Choose a step, read what to look for, then show its location in the live workspace. This tour only navigates.
Private boundary
EntiTrade itself remains on 127.0.0.1:8790. Mobile access uses the exact HTTPS gateway entitrade.entiserve.com when configured; port 8790 is never exposed directly.
Argon2 password hashing, encrypted authenticator secret, replay protection and one-use recovery codes.
Polling does not extend the session. Absolute lifetime is 12 hours. Server restart logs everyone out.
No exchange credentials, signing functions, withdrawals, leverage or live order routes.
Feed and supervisor health
A hash chain helps detect record changes; it is not independently notarised or tamper-proof against the server owner. This private pilot has not had an independent penetration test.