← Projects·Quantitative Researcher
Time-Series Momentum + Vol Targeting
A time-series momentum strategy on a single crypto series, with and without vol targeting. The headline isn't alpha — it's risk management that doesn't break under drawdown.
The hypothesis
Time-series momentum (long positive-trend, flat/short negative-trend) is a documented edge. The risk-management question is more interesting: does vol targeting actually buy you risk-adjusted Sharpe on top of the raw signal?
What the project does
- Signal: sign of the 20-day return.
- Vol targeting: scale position to a 20-day realized-vol target.
- Cost model: 5 bps/side.
The result
| Metric | Raw signal | Vol-targeted |
|---|---|---|
| Sharpe | 0.27 | 0.39 |
| Max DD | −62% | −30% |
Vol targeting doesn’t generate alpha — it generates calm. The Sharpe lift and the DD halving are the same story at different frequencies.
What’s transferable
A 20-day realized-vol target is one of the cheapest risk-managements you can ship. Applied at the strategy level or the book level, it changes the equity curve shape more than it changes the headline Sharpe.
Capacity & participation rate
A vol-targeted BTC/USDT strategy at 30% annualized target vol has a capacity that is set by:
- Average daily volume (ADV): Binance BTC/USDT ≈ $20B/day (mid-2026)
- Position size at target vol: $1M notional × 30% / √(252 × 1) ≈ $19k daily P&L σ, scaling to ≈ 0.18 BTC notional at $60k BTC
- Linear-impact cost model: slippage ≤ 5 bps at < 5% ADV participation
Capacity estimate: ≈ $3–5M notional at <5 bps impact on BTC/USDT alone. Beyond $5M, the linear-impact assumption begins to understate true slippage; the cost model in project 08 would need a square-root-impact calibration.
Cross-asset capacity: the same signal on ETH/USDT or SOL/USDT is <1% of the BTC-tier capacity — these are venue/asset class specific. The strategy is not capacity-unconstrained; it’s capacity-bounded at the venue level.
Live-paper record
This strategy is currently running as a paper-trade. See /positions for the live equity curve, realized slippage, and weekly updates. The current live period is 50 trading days, +12% return, max DD −3.6% — confirming the calm-regime behaviour predicted by the backtest.