Quantitative trading · Crypto markets · Live since Jan 2018
Precision
at scale.

A systematic trading engine for crypto markets. Strategies validated across three market regimes. Deployed only after walk-forward confirmation. Executed in under twenty milliseconds.

Status Live Trading
Live strategies 3 of 6
Venues Whitebit
Binance
Volume / day ~20M USDT
Trades / day ~250
Uptime · 90d 99.97%
Takumi — the Japanese word for the mastery that comes from decades of deliberate practice. The craftsman who refines a single discipline until it becomes second nature.
The ethos · Translated into hard rules
01 · Operating principles

Six rules. No exceptions.

These are not aspirations — they are the gates every strategy must pass through before it touches capital.

Rule 01
Sharpe ≥ 1.5 on out-of-sample data across three market regimes.
Bull, bear, and chop. Backtest results on training data are disregarded. Only walk-forward results count toward deployment criteria.
Rule 02
Max position size ≤ 2% of NAV per strategy at entry.
Capped at strategy level, aggregated at portfolio level. Correlation-adjusted during high-volatility regimes. No exceptions.
Rule 03
Fixed stop-loss defined before entry. Never moved against the trade.
Trailing stops in favor of the position are allowed. Widening stops after entry is forbidden — it is the fastest route to structural drawdown.
Rule 04
Kill switch triggers at −12% drawdown from peak.
Automated halt of all strategies. Manual review required before redeployment. Protects against regime shifts the models have not yet seen.
Rule 05
Every parameter is reproducible. Every trade is traceable.
Full audit trail from signal generation to order fill. Strategy configs versioned in git. No manual overrides that bypass the record.
Rule 06
Paper-trade for minimum 30 days before capital deployment.
Live conditions, zero risk. Validates execution assumptions that backtests can miss — slippage, partial fills, queue position, API quirks.
02 · Strategy portfolio

Five strategies.
Each with an explicit trigger.

Every strategy has a falsifiable hypothesis and a single mechanical entry rule. No discretion. No override. The trigger fires or it does not.

s-002
Mean reversion
Live

Short-term price extremes on liquid perps revert when order book imbalance contradicts the move. Avoids catching falling knives by requiring book disagreement with price direction.

Trigger:|z-score(15m)| > 2.5σ · book imbalance opposes move > 60/40
Sharpe ratio (OOS)1.84 / 3.0
Sharpe (OOS)
1.84
Pairs
Top 10
Since
Mar 2025
Win rate
59.8%
Avg trade
+0.31%
More detail
s-003
Funding rate arbitrage
Live

Extreme funding rates on perpetual contracts can be captured market-neutrally with a delta-hedged spot position. Isolates the funding payment as the primary return source.

Trigger:|funding| > 0.08% per 8h · basis < 30 bps · top-3 venue
Sharpe ratio (OOS)1.67 / 3.0
Sharpe (OOS)
1.67
Venues
3 CEX
Since
Jan 2025
Market neutral
Yes
Avg APR
18.4%
More detail
s-004
Cross-exchange basis
Paper

Temporary price dislocations between spot venues during volatility spikes mean-revert within seconds. Execution is highly sensitive to latency, withdrawal times, and inventory balance.

Trigger:inter-venue spread > 12 bps · volatility regime = high
Sharpe ratio (paper)1.52 / 3.0
Sharpe (paper)
1.52
Venues
4 CEX
Since
Apr 2025
Status
30d paper
Deploy ETA
Jun 2026
More detail
s-005
Liquidation cascades
Research

Cascading liquidations exhaust when large open interest clusters are cleared. The subsequent bounce is tradeable if cascade termination can be identified in real time.

Trigger:TBD · backtest phase over 2Y liquidation data
Phase
Backtest
Data
2Y history
Since
Mar 2026
OOS target
May 2026
Data source
Coinglass
More detail
03 · System

Built for speed, safety, scale.

A modular pipeline engineered from first principles. Market data flows from exchanges into a normalized store. Strategies subscribe to signals, a shared risk layer sizes positions, and execution routes orders to venues.

Data ingest
WebSocket streams
Tick-level history
Feature store
ClickHouse
Redis cache
Signal engine
Strategy modules
Python + Rust
Risk layer
Position sizing
Kill switches
Execution
Smart routing
CEX APIs
Core
Node.js
Rust (hot path)
C++
Data
MongoDB
NATS
Redis
Infra
Docker
Prometheus
Grafana
Venues
Whitebit
Binance
14 ms
Median signal → fill latency
38M
Tick-level events / day
99.97%
System uptime · 90d rolling
3
Venues active simultaneously
04 · Performance

Live metrics.
No hype.

Figures pulled directly from the trading database. Unlevered. Net of fees and slippage. Updated every thirty minutes from exchange APIs.

Synced 13:59 UTC · 19 Apr 2026
Sharpe ratio
2.18
Rolling 90-day
Win rate
63.4%
128 trades · 30d
Max drawdown
−8.7%
Below 12% kill-switch
Net return
+34.2%
YTD · unlevered
Equity curve
Takumi system BTC benchmark
Source: internal trading database · reconciled daily against Binance + Bybit exchange reports
Recent executions
Live
PairStrategySideP&LTime
BTC/USDT s-002 LONG +0.63% just now
BTC/USDT s-001 LONG -1.35% 1m ago
BNB/USDT s-002 LONG +0.33% 2m ago
SOL/USDT s-001 LONG +0.87% 3m ago
AVAX/USDT s-002 LONG +0.98% 4m ago
ETH/USDT s-001 SHORT +0.51% 5m ago
Monthly returns · 2026 YTD
Jan
+4.8%
Feb
+7.2%
Mar
−2.1%
Apr
+6.4%
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
hover a month to see running total

This page documents an in-house quantitative trading system operated as a private company. It is not a solicitation, investment offering, or performance advertisement. Past performance does not predict future results. Quantitative trading involves substantial risk of loss.

05 · Build log

What shipped recently.

A running record of strategy deployments, risk framework changes, and infrastructure work.

12 Apr 2026
Research
Orderbook microstructure study initiated
Exploring queue dynamics and trade flow at 1–30 second horizons. Tick-level data collection active across Binance and Bybit. Early results show predictive power but margins are thin after fees and slippage.
+ 2.4 TB data collected · feature study ongoing
28 Mar 2026
Risk
Adaptive sizing for high-correlation regimes
Deployed rolling correlation measure across active pairs. When cross-asset correlation exceeds 0.75, per-strategy size is scaled down by 30%. Reduces drawdown during synchronized crypto moves.
Impact: −34% peak-to-trough on the 14 Mar correlated sell-off
15 Mar 2026
Deploy
Liquidation cascade strategy entered backtest
Hypothesis formalized. Backtesting across two years of liquidation data from Coinglass and exchange APIs. Out-of-sample validation scheduled for May 2026.
s-005 · backtest phase
02 Mar 2026
Infra
Feature store migrated to ClickHouse
Previous Postgres setup became a bottleneck at ~30M rows/day. ClickHouse reduced query latency on common aggregations by roughly 40× and simplified columnar research workflows.
P95 query latency: 1.2s → 31ms
18 Feb 2026
Deploy
Cross-exchange basis strategy in paper trading
Two months of backtest results passed review. Now running live with simulated fills to verify execution assumptions, particularly around withdrawal times and inventory constraints.
s-004 · paper phase · min 30d before capital deployment