Start with the instrument, not the coin name
This page concerns BTC-USDT-SWAP, the instrument identifier used to request an OKX perpetual-market chart. It is not the BTC-USDT spot instrument, nor a BTC-USD contract. A Bitcoin label alone does not establish which product, settlement currency or quantity unit you are viewing.
Before reading a signal, match the full instrument identifier on this page to the exchange screen. Confirm the current instrument status, contract value and quantity unit there. Tick size, minimum size and funding terms have not been live-verified for this tool; they are not replaced with remembered values.
| Object | What it represents | What to check |
|---|---|---|
| BTC-USDT | A spot-market instrument | Spot balances and available order types |
| BTC-USDT-SWAP | A perpetual-market instrument | Contract definition, quantity unit, margin settings and funding |
| This page’s chart | OHLC candles requested for the selected instrument | Exchange, candle interval, update time and candle completion |
Your immediate task: write down the full instrument ID, selected interval and chart timestamp. If any of them differ from the screen you intend to use, fix that mismatch before considering the indicator.
Read the two bands on the selected timeframe
The application uses two Bollinger windows: 20 candles and 90 candles, both with a two-standard-deviation distance. It also displays EMA20 and EMA50. The 90-candle window is an application setting, not a universal Bollinger default.
Period means candles, not automatically days
At 1H, a 20-candle window spans about 20 hours of evenly spaced candles; at 4H, it spans about 80 hours. Switching the interval changes the information in the window even if the label “20” stays the same. Comparing a 1H breakout to a 4H chart requires recording both intervals.
| Displayed item | Calculation / purpose | It does not establish |
|---|---|---|
| BOLL20 | 20-candle mean with upper/lower bands at ±2 standard deviations | A confirmed reversal or profitable entry |
| BOLL90 | Same calculation over 90 candles | A separate higher-timeframe chart |
| EMA20 / EMA50 | Exponential averages with different response speeds | A complete trend or risk model |
For the same window, the lower band must lie below the middle and the upper band above it, unless the price variation is zero. This is also a useful implementation sanity check: upper = mean + 2 × standard deviation; lower = mean − 2 × standard deviation.
A touch, a crossing and a closed candle are different
A high or low touching a band is not the same as the close crossing it. The chart checks whether one close moves from inside to outside a band relative to the preceding candle. This describes a price event; it does not decide whether you should take a position.
Check the candle completion flag
OKX candle responses include a confirm field: 0 denotes an unfinished candle and 1 a finished candle. The latest candle can still change. The chart calculates its crossing on completed candles and separately labels the live last price.
- Check data retrieval time and the last completed candle time.
- Compare the previous close with its own band and the new completed close with its own band.
- If price moves back inside later, record that later event; do not rewrite the earlier observation.
- If several bands cross together, keep both observations instead of forcing one label.
A trend can continue along a band and a breakout can fail. Record what would invalidate your interpretation, the timeframe you will reassess and the information still missing. A chart marker alone cannot supply those decisions.
Check an AI output before relying on it
The reviewed application asks AI for direction, takeProfit, stopLoss and reason from a multi-timeframe market payload. A valid JSON response only proves that a response can be parsed. It does not establish current prices, a tested strategy or a suitable position size.
| Check | Question to answer | If it fails |
|---|---|---|
| Identity & freshness | Is this the same instrument and candle set you selected? | Do not use levels from another instrument or an old request. |
| Directional ordering | For long: stop < entry < target; for short: target < entry < stop? | Reject inconsistent values rather than silently swapping them. |
| Reasoning | Does the explanation refer to the supplied timeframe and observable data? | Ask for clarification; avoid treating confident wording as evidence. |
| Execution assumptions | Which order type, trigger price and costs are assumed? | Check exchange settings separately; this response does not execute orders. |
AI analysis requires an authenticated account with an active membership or redemption quota. Check the instrument, timeframe and returned prices before using an output. The calculator below only checks the arithmetic of prices you enter.
Check your price plan’s arithmetic
Enter your own prices in the same quote unit. The checker validates their direction and calculates distance to stop, distance to target and their ratio. It does not choose prices, assess fill probability or calculate liquidation. Leave all fields blank until you have values to review.
The ratio is potential target distance divided by stop distance, before fees, funding and slippage. A larger ratio alone does not establish a better trade. Partial fills, gap moves and the exchange’s stop trigger can change the realised result.
Quantity and contract value belong in a separate exposure calculation. Do not treat a price-distance percentage as account-loss percentage, or a contract count as Bitcoin quantity without verifying the contract definition.
Know what a historical signal statistic measures
The old symbol-statistics endpoint reads AI analysis logs with result labels such as correct, wrong and none. Those labels are application records, not independent proof of executed trades. Before showing a win rate, define how each record is resolved and which records enter the denominator.
| Metric | Definition to document | Important boundary |
|---|---|---|
| Resolved signal rate | Correct / (correct + wrong), if that is the chosen rule | Unresolved records must be reported separately. |
| Return illustration | Entry/exit prices, direction, quantity and included costs | Adding unweighted price percentages is not an account equity curve. |
| Both levels touched | An ordering rule or lower-timeframe evidence | One OHLC candle may not reveal which level was hit first. |
| Record completeness | Successful, failed, repeated and unresolved requests | Removing failures can bias the sample. |
During the earlier read-only check, the BTC endpoint returned an empty record set. This page therefore makes no win-rate or profitability claim. When data is connected, show the period, sample count, costs, result policy and update time beside any number.
When the chart or signal does not look right
| What you see | First checks | Next action |
|---|---|---|
| Blank chart / data unavailable | Public API connection, exchange response, update time | Retry once; keep an error state instead of a zero price. |
| Chart differs from exchange | Full instrument ID, interval, timezone, completed candle | Compare the same instrument and candle timestamp. |
| No crossing | Completed candles and minimum window length | A valid chart can have no crossing; do not force a signal. |
| AI button is unavailable | Service connection, account entitlement, request state | Distinguish login, quota, market-data and AI provider errors. |
| Old AI levels remain visible | Instrument/interval changed or newer data requested | Clear old outputs when context changes; request a new review. |
When reporting a problem, include instrument ID, interval, retrieval time, last completed candle time and the exact error. Keep account credentials and API secrets out of the report.
Make a usable analysis note
Use this handover to review your own observations later. The checkboxes stay only on this page and are not uploaded or saved after reload. Print a copy if you need a record.
0 / 6 checks complete
Add your observation, the conditions that would change it and the next review time. An empty checklist is a prompt to collect evidence, not a reason to seek a stronger-sounding prediction.
Questions to settle before acting
Is BOLL90 the same as switching to a higher timeframe?
No. It uses 90 candles at the currently selected interval. Switching interval creates a different candle series.
Why can a live crossing disappear?
An unfinished candle can change before it closes. The chart bases its reported crossing on completed candles.
Does an AI target execute an order?
This website does not execute trades. A target and an exchange order are separate objects.
Does an empty history prove the strategy failed?
It establishes that no records were returned in that response. It does not prove a win rate, a loss rate or the absence of records in every system.
Can I use the same stop distance for every symbol?
This page supplies no universal setting. Contract definitions, price behavior, costs and your constraints require their own checks.
Sources, review scope and open questions
Reviewed 12 October 2026. Application-specific descriptions come from the existing frontend and PHP source inspected for this rebuild. External documentation explains candle fields and indicator definitions; it does not validate this application’s performance.
- OKX API documentation — instruments and candlestick fields ↗
- Fidelity — Bollinger Bands definition and limits ↗
- Still to verify: live BTC contract specifications and exchange fee/funding conditions.
- Still to verify: a successful market connection and matched completed candles.
- Still to verify: account integration, AI request freshness and quota behavior.
- Still to verify: historical outcome rules, sample integrity and included trading costs.