All VIP Trading features in one place. Follow the categories to learn step by step.
Log in with the ID and temporary password issued by the administrator, then immediately change your password in your profile to secure your account. This is the first task to complete within 5 minutes before using the screen.
Enter the ID and temporary password issued by the administrator via Telegram or email into the two input fields on the login page (/login), then press Login.
If the password is in the format of 6 uppercase English letters (e.g., ZXKQPA), it is a temporary password. General user registration is not available; all accounts are issued directly by the administrator.
Immediately after logging in, click the area at the bottom of the left sidebar where Username + Role (VIP) is displayed. You will be taken to the /profile page, where you can manage all personal settings including password changes, email, contact information, and AI API keys.
On the Change Password card of the profile page, enter the new password twice and click Change Password. A minimum of 6 characters, mixing letters, numbers, and special characters is recommended.
On the profile's Account Information card, verify that your email and contact details are correct. Auto-betting alerts and system notices will be sent through this channel.
Your role is one of vip (general), manager, or admin (administrator). Your role determines the scope of available features.
Once the login and password have been set up, the next step is exchange API registration. You need an API key from at least one exchange for automated betting and real-time balance inquiries.
Register API keys and secrets from exchanges (Binance, Bitget, OKX, Bybit, etc.) to enroll in VIP Trading. The registered keys enable automated bet placement, real-time balance, and position synchronization.
Navigate to the left sidebar Live β Exchange Connection (/trading/exchanges). Only exchanges activated by the administrator will be listed in the Exchange Adapter card at the top of the screen.
For beginners, we recommend a latecomer (API-friendly, fast KYC) such as Bitget or Bybit. Binance has Korean IP restrictions, so a separate VPN or a Japanese account may be required.
Log in to the exchange website β Go to the API management menu β Create a new API:
On the exchange connection page, in the Register New API Key card, enter the following:
Once saved, it will immediately send a ping to the exchange API to automatically verify whether balance inquiry is possible.
Press the Test button next to the registered key to call the exchange API for balance inquiry. If successful, it will display as β Connection Successful β N Assets.
If it fails, check the following:
You can register multiple API keys for a single exchange (e.g., main account + sub-accounts). When setting up automated betting, you can select which key to use based on the exchange and symbol combination.
After logging in, the dashboard (/dashboard) automatically displays your current status at a glance across 9 panels. Familiarizing yourself with what each panel means will make a daily 5-minute check sufficient.
The largest card, Position, is a list of open positions by exchange and symbol. Each row displays Symbol / Side (long/short) / Entry Price / Quantity / Leverage / Unrealized P&L.
Prices are updated every minute using the freshest values from the exchange feed, and unrealized P&L is calculated as (Current Price β Entry Price) Γ Quantity Γ Β±1 (long=+, short=β).
Risk Gauge cards show your risk level at a glance.
Three cards summarize the market environment.
The Trading Engine card is the master switch for the automated betting bot. When ON, it evaluates the active config every minute and places an order immediately upon signal generation. When OFF, all automated evaluations are paused.
The AI Health card shows the status (up/down/degraded), average latency (ms), and priority of each registered AI provider. Administrators can adjust priority by dragging.
Exchange Β· Symbol Profit visualizes the ROE of settled trades from the last 7 and 30 days as per-exchange bars. It shows at a glance which exchanges and symbols are performing well.
Create a single, simplest auto-betting setup with 1 recommendation algorithm + 1 exchange + 1 symbol, and verify it actually works. Start with a dry run (virtual fill) to safely get accustomed, then switch to live trading.
Go to the left sidebar Labs Β· Algorithms (/labs). For new accounts, four recommended algorithms verified by the operations team are automatically activated.
At the top toggle [View All | Active Algorithms], click Active Algorithms to see only the algorithms available to you.
Navigate to the left sidebar Live Β· Auto Trading (/trading/auto). Click Add New Auto Trade at the top right.
When the modal opens, select the following:
Same modal's Bet Size section:
Press Save to immediately add a new row to the auto-bet list. Make sure the toggle in the Run column is set to on.
Now, every minute, a background task evaluates the algorithm and automatically executes a virtual trade when a signal occurs.
Typically, a 5βminute trendβfollowing strategy generates an entry signal about once every 1β3 hours. After waiting briefly, proceed to Live / Bet Results (/trading/results).
Selecting only Dry Run in the top Mode Filter displays only virtual execution results. Clicking on any row shows the entry rationale, chart, and KPI.
The Trading Engine toggle at the top of the dashboard is the master switch for the auto betting bot. You need to understand exactly how it operates when ON and what stops when OFF in order to run it safely.
The engine is a background process that runs every minute, iterating through all of your active auto-betting configurations and asking each algorithm, "Is this a good time to enter now?"
When the engine is ON:
When the engine is OFF:
When OFF is appropriate:
If you toggle back to ON, evaluation will resume immediately from the scheduler call at the 0th second of the next minute. This takes effect instantly without any separate command or restart.
VIP Trading has two decision-making systems β Algorithm (quantitative rules based on Pine Script) and Portfolio Recipe (AI-based multi-asset management). They are used for different scenarios and can be selected as the "basis" in automated betting.
Algorithm is a quantitative rule written in Pine Script v5. Clear entry and exit conditions, such as "if RSI is below 30 and the closing price is below the Bollinger Band lower line, go long," are expressed in code.
Portfolio Recipe manages multiple assets simultaneously through AI autonomous decision-making. The AI decides at each cycle "how to allocate risk and safe assets under the current macro environment."
Selection criteria:
Difference in decision output format:
At the very top of the new "Live Β· Auto Bet" modal, there is a Base radio button.
Determines how to set the amount for each auto bet β choose between a fixed absolute amount (fixed) or a set percentage of the balance (percent). This depends on your account size and risk management style.
fixed: Fixed amount, e.g., "100 USDT per entry." The bet amount remains unchanged even if the balance increases or decreases.
percent: "10% of balance per entry". As the balance grows, the bet amount increases proportionally.
In percent mode, what "balance" refers to differs by mode.
The fixed/percent determines the margin. Actual market exposure is that value multiplied by leverage:
When BTC market price is 60000, 300 USDT exposure = 0.005 BTC quantity.
You can also modify an existing config. Click the Edit button on the existing row β change the Amount Mode radio β save. The new mode will apply from the next entry (current open positions are unaffected).
Understanding how the "Current Price" and "Unrealized PnL" on the dashboard position card are calculated, and why they may differ from the actual execution price. Misinterpreting this can lead to meaningless alerts and overreactions.
Mark price is the reference price used by the exchange for forced liquidation and unrealized profit/loss calculations. It may differ from the actual bid/ask prices.
VIP Trading uses the exchange ticker or the closing price of the freshest candle as the mark price β whichever data point is the most recent is automatically selected.
/admin/ingest) may be paused. Refer to Troubleshooting.
Unrealized PnL formula for open positions:
long PnL = (Current Price β Entry Price) Γ Quantity
short PnL = (Entry Price β Current Price) Γ QuantityExample: BTC long 0.01 BTC, Entry Price 60000, Current Price 62000:
This value is immediately reflected in the PnL column of the position card on the dashboard.
Upon liquidation, the realized PnL deducts trading fees from unrealized PnL.
realized PnL = price difference PnL β (entry fee + liquidation fee)
= (price difference PnL) β (fee rate Γ entry price Γ quantity Γ 2)Default dry-run fee rate 0.0004 (0.04%, based on Binance futures taker). For live exchange trading, the actual fee from the exchange API response is used.
If leverage is N times, with the same margin, the quantity becomes N times larger. PnL also becomes N times larger.
Quantity = (Margin Γ Leverage) / Entry Price
PnL = Price Difference Γ Quantity = Price Difference Γ (Margin Γ Leverage) / Entry PriceExample: Margin 100 USDT, Leverage 5x, BTC entry 60000 β Quantity = 100 Γ 5 / 60000 = 0.00833 BTC. 1% price increase (60600) β PnL = 600 Γ 0.00833 = +5 USDT (5% of margin).
The exchange automatically liquidates positions when unrealized losses exceed the margin + maintenance margin rate limit. Since VIP Trading does not display the liquidation price, monitor indirectly via the Margin Usage Rate gauge on the dashboard.
Using the Labs backtest engine, run a single algorithm against historical data to verify its performance. This is a standard workflow for validating algorithm suitability before live trading.
Go to the left sidebar Labs Β· Algorithms (/labs). In the top market view filter, select Mean Reversion to see mean reversion algorithms such as RSI and Bollinger Bands.
Click Run Backtest at the bottom right of the algorithm card to go to New Simulation (/labs/simulate), where the form opens with the algorithm pre-selected.
Key values to enter:
When you click Start Simulation, the backtest job is registered in the queue, and the Progress gauge progresses from 0% to 100%. At the same time, the equity curve is drawn in real time on the Chart View tab on the right.
After completion, the Result Summary card displays the final ROE, total number of trades, win rate, and maximum drawdown. In the left Trade History tab, click each row for entry and exit pairs β the chart zooms in at a 1:1:1 ratio.
Rather than using all four recommendation algorithms automatically applied to new accounts, activate only those that fit your own scenario to reduce noise. Even for the same algorithm, activation status is independent for each user.
Go to Labs Β· Algorithms (/labs). Click View All from the top toggle [View All | Active Algorithms] to list all registered algorithms.
The Active toggle at the top right of the algorithm card shows the activation status for you.
The activation status is stored per user β another user can keep the same algorithm deactivated.
Top filter market view:
It is advisable to balance your capital allocation between the two groups β if you concentrate on just one side, losses can skyrocket when the market character changes.
4 recommendations verified by the operations team:
You can directly register Pine Script v5 by clicking Add Algorithm at the top. After entering the code and saving, the transpiler will immediately compile and verify it. The registered algorithm is only available to you (it will not be exposed to other users).
The AI generates a report analyzing the quarterly macro environment. Rather than simply listing data, it produces output in the style of a securities firm analyst, including judgments such as "Which is more attractive right nowβrisk assets or safe assets?
Go to the left sidebar Portfolio Β· Macro Report (/portfolio/macro-report). Initially, the Create New Report tab is active.
Select the quarter to analyze in the Period Selection UI.
During the selected quarter, a comprehensive analysis is performed on macro trends using sampler 10 tickers + high importance macro events + the 12 hypothesis catalog.
Press Generate Report to start the AI call.
Once completed, the report body will be displayed automatically.
Standard Report Structure:
You can save as PDF via Print at the top of the screen. Sharing the report URL (/portfolio/macro-report/{id}) allows other users to open the same report (authentication required).
Backtest multiple portfolio recipes over the same period and seed capital to compare which strategy best fits your capital size and risk profile. An essential step before switching to automated betting.
Go to the left sidebar Portfolio Β· Recipe Catalog (/portfolio/recipe). There are 24 seed recipes categorized into 6 categories.
Turn on the active toggle for the recipe cards to compare. Only active recipes will appear in the dropdown of the new simulation form.
Portfolio Β· New Simulation(/portfolio/recipe/simulate) Navigate to. Set the following identically:
Start Simulation. Completes in 1β3 minutes.
Note the results of the first completed simulation (final ROE, win rate, DD), and run the second candidate with New Simulation under exactly the same seed/period/leverage/duration.
Portfolio Β· Simulation History (/portfolio/recipe/history). Both simulations created above are displayed. Compare the following:
Once the winner has been determined through comparison, proceed to add a new auto bet (/trading/auto) β Base: Portfolio β Select Winner Recipe β Run in Dry Run mode for 1β2 weeks in real-time β Switch to live trading if satisfied.
Explains in detail how to register, modify, delete, and verify exchange API keys. Also includes security permission settings and key rotation best practices.
Main screen consisting of two cards:
First Exchange API Registration covers the initial key registration flow. Here, the focus is on operational management.
Exchange-specific permission items:
The Test button next to a registered key makes a single balance inquiry call to the exchange API. Results:
Security best practice: Rotate API keys every 3β6 months. Steps:
Since the auto-betting config references the key ID, editing alone is sufficient β no need to add a new entry.
Deleting a key from the key list:
You can register multiple keys on a single exchange.
In the auto-betting modal, select which key to use from the Exchange dropdown. Even with the same exchange and symbol, each key operates independently.
Adjusts the priority of AI providers used for AI review, macro reports, and recipe simulations. If the first-priority key fails, it automatically falls back to the second and third priorities.
In the AI API Key card of /profile, click Manage API Keys. A modal opens showing the list of registered keys and priority β²βΌ buttons.
/admin/ai-providers. This is used when there is no user key.
In the "New Registration" section:
The key list displayed in the modal is in priority order (1st priority at the top). Use the β² (move up) / βΌ (move down) buttons in each row to change immediately.
When making a decision, use from 1st priority to active keys in order:
The test on the key row calls a short phrase ("Say hello") using that key. If the response is normal, a β mark is displayed. If it fails, an error message is shown.
Temporarily OFF with the Active toggle on each key row. The key is not deleted, only excluded from the priority fallback. Toggle back ON to restore immediately.
Create a Telegram bot to receive real-time updates on key VIP Trading events (new proposals, executions, liquidations, forced liquidations, etc.), enabling immediate response even when away from the desk.
Search for @BotFather on Telegram and open the chat. Type /newbot β Answer two questions:
_bot (e.g., vip_alerts_bot) β globally uniqueBotFather will issue an HTTP API token. Format: 123456789:AAExx.... Save it in a secure place.
Search by bot username β Add to your chat room (direct chat for individuals, invite the bot as a member and grant permissions for groups). Send any message in that chat room.
Call the following URL in your browser:
https://api.telegram.org/bot{TOKEN}/getUpdatesThe result[].message.chat.id value in the response JSON is your chat_id.
-123456789)987654321)Go to the left sidebar Live Β· Telegram Bot Alerts (/trading/telegram). Click Register New Channel:
Event Category:
liquidation + position_closed. The rest are up to personal preference. Turning on too many may cause notification fatigue.
When you press Test on the registered channel line, a short test message is sent to the chat room. If the message arrives in Telegram, β Success.
Multiple channels can be registered. Example:
Event selection for each channel is independent, so notification branching is flexible.
Set the visual theme (dark / light) and language (Korean, English, Japanese, German β 4 options) of VIP Trading to your preference. Settings are saved immediately and retained on subsequent logins.
Every page has a π / βοΈ toggle in the top-right header. Clicking it instantly switches between dark and light. CSS variables update without any page flicker.
The translate icon button in the top-right header opens a dropdown with 4 languages (νκ΅μ΄ / English / ζ₯ζ¬θͺ / Deutsch). Selecting one immediately reloads the page with the new language applied.
The selected language is saved in two places:
In other words, if you switch to English on your work PC, it will still appear in English when you log in from home.
Translated items:
Not translated:
Explains in detail all the fields of the automatic bet setting modal. After the initial setup, this is a page you will frequently refer to during the actual operation phase.
Modal top basis radio:
Fields below the modal dynamically change based on selection. See concept difference.
The Exchange dropdown only displays exchanges for which you have registered API keys. The Symbol dropdown only shows symbols mapped to that exchange.
Amount mode: fixed vs percent (concept).
Real Operating Time Limit:
Multiple time zones can be added: e.g. 09:00~12:00 + 14:00~17:00.
AI Validation Pipeline checkbox β when enabled, AI performs additional review after algorithm signals. When disabled, bets are placed based on algorithm signals as-is.
When the checkbox is enabled, the Checklist section becomes active. Mix the operations team's presets (e.g., "Macro environment suitable", "Sufficient liquidity", "News sentiment positive") with your own custom items.
Pressing Save closes the modal and adds a new row to the auto-betting list. The Run column of the row contains a toggle.
It may be in OFF state right after saving β toggle it ON immediately if that is your intention.
This page explains how to read the KPIs, charts, and tables on the /trading/results page and reflect them in operational decision-making. It is the core screen for the daily 5-minute check.
Four KPI cards at the top of the page:
All PnL values are net of fees (realized_pnl column).
Top mode filter chips:
Since mixing dry runs and live trades is meaningless, always filter by one mode before analyzing.
The chart below the KPI shows the daily net PnL for the last 30 days. Bar colors:
Observe the overall balance to identify trends. If there are 7 or more consecutive red days, the algorithm needs to be checked.
The Performance by Config card in the middle section shows, for each auto-betting config:
When there are multiple configs, it provides an intuitive comparison of which config is performing well.
Consolidated transaction history table below:
Currently for on-screen display only. CSV export will be added later. Temporary workaround: Right-click on the transaction history page β "Save as" β paste the table from the page source into Excel.
This covers what the AI Health card on the dashboard measures and how to interpret it. If the AI provider becomes unstable, auto-betting and macro reports may be affected.
AI Health card on the right side of the dashboard. Displays the status of all registered providers, one row per provider:
Status criteria:
An automatic health check runs every 5 minutes to update the status and is preserved as a time series.
When an AI call is needed, the system:
This operates automatically without any additional action required from the user. The AI Health card is for monitoring purposes only.
Top-tier provider degraded for several consecutive days:
If the problem persists, it is recommended to swap the top-tier provider with the second-tier provider.
Average Latency Interpretation:
In AI review for real-time auto betting, if latency is too high, signals may be missed due to timeout. Prioritize fast providers.
Letβs look at what prompts the AI receives to generate macroeconomic reports and what response structure it produces. Understand which aspects of the prompt can be adjusted when the report quality is unsatisfactory.
Basic structure of the macro report prompt:
The 12 hypothesis catalog is a macro thesis maintained by the operations team. Each hypothesis:
The AI matches the hypotheses with current data to determine OVERWEIGHT/NEUTRAL/UNDERWEIGHT β that is the core of the report.
/admin/portfolio-macro).
Multilingual processing method: Keep the prompt body (written in English) as is, and only append the following at the end.
[LANGUAGE]
Respond ONLY in Korean (ko).
All narrative fields, reasoning, and analysis MUST be in Korean.
Keep machine-readable enums, code values, JSON keys, and ticker symbols
in English / original form.This eliminates the need to translate the prompt body four times, minimizing token costs while keeping only the response narrative in the user's locale.
Example output JSON schema:
{
"summary": "Quarterly summary narrative",
"active_hypotheses": [
{"code": "h_dollar_weakness", "stance": "OVERWEIGHT", "rationale": "..."}
],
"asset_views": [
{"symbol": "BTC", "stance": "OVERWEIGHT", "narrative": "..."}
],
"portfolio_template": {
"holds": ["BTC", "ETH"],
"avoids": ["JPY"],
"hedges": ["Gold"]
}
}Signals of a good report:
If you request the same period again, it will be returned immediately from the cache (AI calls: 0). To force a new generation:
How to review what decisions the portfolio recipe made at each turn and why it made them. This is a key step in judging and tuning recipe suitability.
Portfolio Β· Simulation History β Click a simulation to review on (/portfolio/recipe/history) β Navigate to the detail page (/portfolio/recipe/simulate/{id}).
In the timeline section of the detail page, cards are listed per turn. Each card:
Click on Stage 1 sentiment card β Modal:
Check if the narrative aligns with current macro facts. If it deviates, also question the Stage 2 decision-making.
Stage 2 strategy card:
{
"decisions": [
{"asset": "BTC", "action": "enter", "side": "long", "weight": 0.4, "leverage": 3, "reason": "..."},
{"asset": "Gold", "action": "hold", "reason": "..."},
{"asset": "JPY", "action": "close", "reason": "Risk-off termination"}
]
}
When AI decisions are not executed as-is:
Check the differences in the "Execution Result" section of each turn card.
Aggregation at the top of the detail page:
If the "hold" action is 80% or more, the prompt is too conservative β market opportunities are being missed. If the "enter" action is 60% or more, it is too aggressive β excessive noise entries occur.
This covers how to read the rationale narrative displayed in the trading details modal at /trading/results. There are two types: algorithmic rationale and AI rationale.
Click a row in the trading history table at the bottom of /trading/results β a detail modal opens. At the bottom of the modal, there are two buttons: Algorithmic Basis / AI Basis.
Click Algorithm Rationale β narrative panel opens. Sections:
Only applies to trades in configs where the AI verification pipeline was enabled.
For recipe-based automatic betting, the basis is the following section:
In the upper right corner of the basis panel, toggle View Raw Data. Exposes the entire payload in JSON format:
If you read 5 to 10 justifications daily, you will start to see patterns in which market conditions the algorithm handles well and where it misfires. This pattern recognition leads to:
A glossary of frequently used terms in VIP Trading, organized in alphabetical order. If you encounter an unfamiliar term in the manual or on a screen label, quickly look it up here.
api_key + api_secret (+ passphrase). Management guide.If auto betting is turned on but there have been 0 positions for several days, check the following causes in order. Most of these are configuration or environment issues; itβs actually rare for the algorithm itself to produce no signals.
Most common cause. Check the status of the Trading Engine card on the dashboard.
When OFF, no automated evaluation is carried out. Refer to Engine Mode concept.
Check if the Run column toggle for the corresponding config row in /trading/auto is ON. If it is OFF, even if the engine is ON, this config alone will be excluded from evaluation.
If the schedule is set to "Specify time range," the current time must be within the range. Example: If the setting is 09:00~18:00 and it is currently 03:00 in the morning, the evaluation is skipped.
If the engine and your auto-bet configurations are all ON but no positions have opened for several days, the background processing may have a temporary issue. Please contact the operations team.
Each config has a cooldown (usually 5β15 minutes), so if a new signal occurs too close to the previous entry, it is ignored. Check the recent entry time in the trade history:
If all 5 steps above are OK but there are 0 orders for several days, the market is likely in a sideways trend. The mean_reversion algorithm works well in a sideways market, but it is normal for trend_following to produce 0 orders.
If macro report generation, recipe simulation, or AI review takes longer than usual, check the provider status or prompt size.
First check. AI Health card's top priority provider status:
If the 1st priority is consistently slow:
When the provider limit is nearly exhausted, calls are throttled. Check directly on the provider site (console.anthropic.com, etc.):
Macro reports analyzing long quarters can have prompts of 50KB+, which naturally results in slower responses.
Start with a short period to quickly check results.
Although rare, the network from the VIP server to the provider API may be temporarily slow. Request the administrator to check ping/route.
If you press the header flag and the language remains the same or only certain areas change, check the following.
After switching the language, the page will reload once. If it does not reload automatically, force refresh:
Rarely, statically cached areas such as the login page may not switch immediately. Asking the operations team to clear the cache resolves it instantly.
F12 β Application tab β In localStorage, if keys like vip-theme exist, leave them as is. If locale-related keys are stale, delete them:
On the profile page, verify that the currently saved language matches what you want. If a different language is saved, reselect, save, log out, and log back in.
Example: The header is in English, but the inside of a modal is in Korean. In this case, that area's translation is missing. Report to the operations team with the following information:
If the total fees are eating into profits in dry runs or live trading, check the following. This occurs particularly often in short timeframe algorithms.
The dry-run default fee rate is 0.0004 (0.04%, based on Binance futures taker fee). If your actual exchange has a lower fee (e.g., 0.0002 = 0.02%), change the setting:
The 1-minute candlestick algorithm generates many signals, resulting in a very high trading frequency β accumulated fees. Switching to 5-minute or 1-hour candlesticks:
If algorithms generate signals frequently, about half are filtered through AI review:
AI call costs are incurred, but fee savings are greater (typically fee savings > AI costs).
If trading volume is sufficient, VIP tier fees of 0.02% or lower are possible:
Leverage N times also generates N times the fees (fees are based on exposure). Reducing leverage from 10x to 5x cuts fees in half. However, profits are also halved, so consider the balance.
When the progress gauge of a backtest, macro report, or simulation appears stuck at 0% or a fixed percentage, work through the steps below in order. Most cases are temporary; long stalls need the operations team.
Live updates run in the background, but the connection can briefly drop, making the displayed progress look frozen even when work is still proceeding.
The work may have already completed even when the progress display looks stuck. From the simulation / report history page:
After trying the steps above, if progress stays at the same percentage for more than 10 minutes, contact the operations team with the following info:
The operations team will inspect the backend and resolve it.