Proprietary quantitative trading · Singapore
We trade our own capital, systematically.
Kaching Ventures Pte. Ltd. is a Singapore-incorporated proprietary trading company. We design, validate and operate systematic strategies across digital assets, listed futures and equities, using the company's own capital exclusively.
Company
Kaching Ventures Pte. Ltd. was incorporated in Singapore in 2025. We are a small, privately held company, and every position we take is funded with the company's own money.
Our work is quantitative research and engineering. We build the data pipelines, statistical models and execution systems that turn observations about market microstructure into rule-based trading strategies.
Every decision rule is specified in code, tested against market history and then audited against live fills. A strategy enters production only after it has passed each stage of validation, and it stays there only while the evidence still supports it.
Markets
Digital assets
Spot and perpetual futures on major centralised exchanges. Research focus: order flow and large-trade dynamics.
Futures
Exchange-listed equity-index futures. Research focus: intraday price and volume divergence, and liquidity around contract rolls.
Equities
Exchange-listed shares. Research focus: multi-factor selection models.
Research process
Hypothesis
Each strategy begins as a stated, falsifiable hypothesis about market behaviour, usually drawn from order-flow or microstructure data.
Data
We maintain our own archive of tick-level trade and quote data, cleaned and documented so that every test can be reproduced.
Simulation
Event-driven backtests at tick resolution, with conservative fill assumptions and explicit fees and slippage.
Validation
Walk-forward and out-of-sample testing, with correction for multiple comparisons. Ideas that do not survive are rejected and the result is recorded.
Shadow trading
Before any capital is committed, a candidate strategy must first run on live data without capital, and its simulated fills must reconcile trade by trade against the backtest.
Production
Capital is allocated only then, at limited size, and live results are reconciled against broker statements.
Methods
Order flow and microstructure
We study signed trade flow, the clustering of large trades and divergence between price and volume across several time horizons. Slower mean-reverting behaviour is modelled as an Ornstein–Uhlenbeck process and traded on standardised deviations.
Two execution conventions
Every limit-order strategy is simulated twice: once filled when the market touches the price, and once only when the market trades at least one tick through it. Strategies are ranked under the stricter convention.
Data-snooping control
Parameter grids of several hundred configurations are evaluated jointly. A result counts only if its statistic exceeds what the best of that many random configurations would be expected to reach, and only if its ranking holds from in-sample to out-of-sample periods.
Rules replayed in sequence
Filters and gates are evaluated by replaying them inside the simulation, in time order, and never by deleting trades from a finished backtest, which lets later information leak into earlier decisions.
Mark-to-market, not realised
Exit rules are judged on mark-to-market drawdown and return-to-drawdown ratios. Realised profit alone flatters any rule that postpones closing a losing position.
Volatility-conditioned exposure
Position size is scaled by trailing realised volatility, which in our data predicts the magnitude of the next move but not its direction. Exposure is measured per trade and not per day.
Simulation against live data
Shadow strategies are reconciled against the backtest trade by trade until the difference is zero. What remains is then attributable to operational events such as restarts and disconnections.
Liquidity-based roll handling
Futures rolls are detected from where volume is trading and not from the calendar. Rolling windows are flushed at the switch and open positions are adjusted by the measured basis between contracts.
Technology
Capture and storage
Feed handlers are built with data-freshness watchdogs, so that a stalled feed is detected and not mistaken for a quiet market. A historical tick store serves as the single source for research.
Automated order management
Order handling through broker and exchange APIs, with pre-trade checks and position reconciliation before orders are sent.
Monitoring and reporting
Service health checks, automated alerts, and scheduled profit-and-loss and reconciliation reports. Measured execution cost per contract is tracked against ideal fills.
StackPython for research and execution, SQL on a columnar analytical store, Linux services on cloud infrastructure.
Risk and governance
- Limits. Every strategy runs within preset position and loss limits that are enforced in code.
- Reconciliation. Positions and orders held in our systems are checked automatically against the broker's records.
- Human control. Trading can be halted at any time through a manual override that requires two-step confirmation.
- Records. Orders, fills and research results are logged, including the hypotheses we tested and rejected.
Important information
- Kaching Ventures trades only with its own capital.
- We do not accept money from outside investors, manage assets for others, or offer any fund or investment product.
- We do not provide investment advice or any other financial service to the public.
- Nothing on this website is an offer, a solicitation or a recommendation to buy or sell any financial instrument.
Company information
| Legal name | Kaching Ventures Pte. Ltd. |
|---|---|
| UEN | 202500748G |
| Incorporated | 7 January 2025, Republic of Singapore |
| Company type | Private company limited by shares |
| Business | Proprietary trading for the company's own account |
| Contact | [email protected] For business and general enquiries. We cannot respond to requests to invest. |