AI ASSET MANAGEMENT TECHNOLOGY

KLab AI Trade

Our proprietary AI models analyze the crypto market as it moves 24/7.
AI predicts when to buy and sell, enabling smarter investment decisions.

Cumulative Return (vs BTC Gain)

BTC gain over same period —

2026.07 –

AI BTC Hold

BTC prices converted at $1 = ¥162

Improving & Tuning the AI Model to Resume Live Trading

Key Features of Our AI

In the Bitcoin market, trading constantly is not always the optimal strategy.
What truly matters is extending profits in rising markets while avoiding unnecessary trades and limiting losses in dangerous ones.
KLab's AI trading system draws on a wide range of market data to judge whether current conditions call for offense or defense, autonomously managing trades and positions to match the market environment.

FEATURE 01
hub

Reads Diverse Market Data

Beyond price alone, the AI comprehensively analyzes a total of 96 features, including technical indicators such as EMA spreads and RSI, volatility, derivatives data such as funding rates and open interest, on-chain data, and macro indicators.

FEATURE 02
psychology

AI Assesses the Market

Based on the features from FEATURE 01, the AI predicts whether the current market most closely resembles an uptrend, a range-bound market, or a downtrend, and makes trading decisions accordingly.

FEATURE 03
monitoring

Manages Trades and Positions by Market Conditions

In trending markets it holds positions longer to let profits run; in risky conditions it moves to cash to limit losses. Through continuous trailing stops, position sizing, and exit decisions, it maximizes returns while containing large losses.

We have developed multiple AI models with these characteristics. Descriptions and test results for the two core models, Mamba-bull and Mamba-basic, are shown below.

  • Mamba-bull — A long-term, long-only model that captures medium- to long-term uptrends. By filtering out noise and riding major trends, it steadily accumulates returns.
  • Mamba-basic — A short-term model that trades both long and short on short-term price movements. It reacts quickly to market trends, aiming to profit in both rising and falling markets.

A detailed technical explanation of how the AI works will be published soon.

Mamba-bull Long-term model
Mamba-basic Short-term model
BTC Hold Benchmark
swap_vertTrade Direction
Long only Long / Short Buy & hold
scheduleHolding Period
Long-term Short-term Continuous hold
trending_upBacktest Return
+857.5% return +579.7% return +83.7% return
arrow_downwardMax Drawdown
-21.89% max DD -19.28% max DD -67.6% max DD

* Backtest period: January 2022 – December 2025. BTC Hold shows the return and max drawdown of buying and holding BTC over the same period.

By combining the characteristics of these AI models, we are building a trading system capable of adapting to diverse market environments. After thorough validation through backtesting and demo trading, we began trading with our own funds using the integrated system on July 1, 2026.

01step
history

Backtesting

Simulation-based validation using historical market data. We evaluate how the AI makes decisions against actual price movements.

02step
show_chart

Demo Trading

Validation that trades simulated funds against live market movements. We evaluate the AI's performance under near-real conditions.

03step
rocket_launch

Trading with Our Own Funds

Models that pass validation are deployed with real funds. We continuously monitor and improve their performance.

04step
group

Private Fund

We plan to offer investment opportunities to external investors through Special Business Activities for Qualified Institutional Investors, etc.

Track Record

We publish live trading results achieved with real capital.
For backtest and demo trading results, see our trade records.

Total Trades
 
Win Rate
 
Cumulative P&L
 
Max DD
 
Cumulative Return (vs BTC Gain)
--
AI
BTC Hold
Improving & Tuning the AI Model to Resume Live Trading
View Trade Records east

Latest News

Loading news...

Get in Touch

For media inquiries, partnership opportunities, and other questions, please feel free to contact us.