Rates Carry Proxy
Uses 10Y yield direction as a duration proxy, not a cross-asset carry engine.
This is a narrow rates-duration proxy that borrows the carry intuition. It should not be read as a general carry implementation across asset classes.
Implementation Scope
This is a narrow rates-duration proxy that borrows the carry intuition. It should not be read as a general carry implementation across asset classes.
The Intuition
Carry is one of the most durable risk premia in finance. In bonds, carry is the yield earned from holding a bond minus financing costs. When interest rates fall (bond prices rise), long bond positions earn both carry and capital appreciation. The Carry Trade strategy here operationalises a simple version: use the momentum of the 10-year Treasury yield as a signal for duration positioning in TLT (the 20+ year Treasury ETF).
The economic channel: when yields are falling (rate momentum negative), bond prices are rising — the primary driver is typically either economic slowdown (risk-off), falling inflation expectations, or central bank easing. All three scenarios are bullish for TLT. When yields are rising, the reverse holds. The strategy bets that rate trends, once established, persist for long enough to be tradeable.
Koijen, Moskowitz, Pedersen, and Vrugt (2018) documented the carry premium across 8 asset classes: bonds, equities, currencies, commodities, credit, options, equity factors, and global fixed income. The carry premium is positive on average in every class, though the risk-adjusted excess is most reliable in currencies and government bonds. The common thread: carry strategies are implicitly short volatility and tail risk — they earn steady returns in calm environments but suffer in sudden regime changes.
Key assumptions: (1) The 10-year yield (^TNX) is a clean signal for rate trends — in practice it can be noisy, and practitioners use yields across the term structure to construct a cleaner trend signal. (2) Rate momentum persists long enough to be captured by the lookback window. (3) TLT provides clean exposure to rate duration — it does, being a 20+ year Treasury ETF, but it also embeds credit risk and liquidity factors. (4) The momentum window (20 days default) is appropriate — some practitioners use longer windows (3–12 months) for rate signals.
The critical risk in carry strategies is the "carry crash" — sudden reversal in the carry direction, often coinciding with global risk-off episodes. In bond markets, carry crashes happen when markets suddenly price in rate hikes (as in 1994's "bond massacre" or 2022's aggressive Fed tightening). Position sizing and stop-losses are essential companions to any carry strategy.
The Math
Read this as a compact model summary: what the signal sees, what it ignores, and where fragility can creep in.
mom_rates(t) = (^TNX(t) - ^TNX(t-n)) / ^TNX(t-n)
Signal(t) = +1 if mom_rates(t) < 0 [rates falling → bond rally]
= -1 if mom_rates(t) > 0 [rates rising → bond decline]
Parameters
| Parameter | Type | Default | Description |
|---|---|---|---|
| window | int | 20 | Momentum window for 10Y yield direction |
Source Code
def run(ticker: str, start: str, end: str, **params) -> dict:
window = int(params.get("window", 20))
df = fetch_ohlcv(ticker, start, end)
# Use 10Y Treasury yield as rates signal
try:
df_tnx = fetch_ohlcv("^TNX", start, end)
common = df.index.intersection(df_tnx.index)
df_aligned = df.loc[common]
tnx_aligned = df_tnx.loc[common, "Close"]
tnx_mom = tnx_aligned.pct_change(window)
pos = pd.Series(0.0, index=df_aligned.index)
pos[tnx_mom < 0] = 1.0 # rates falling → long bonds
pos[tnx_mom > 0] = -1.0 # rates rising → short bonds
return run_backtest(df_aligned, pos, ticker=ticker, start=start, end=end,
strategy=METADATA["slug"], params={"window": window})
except Exception:
# Fallback: just use price momentum of the ticker
mom = df["Close"].pct_change(window)
pos = pd.Series(0.0, index=df.index)
pos[mom < 0] = 1.0
pos[mom > 0] = -1.0
return run_backtest(df, pos, ticker=ticker, start=start, end=end,
strategy=METADATA["slug"], params={"window": window})
Further Reading
- Koijen, R., Moskowitz, T., Pedersen, L. & Vrugt, E. (2018). Carry. Journal of Financial Economics, 127(2), 197–225.
- Ilmanen, A. (2011). Expected Returns. Wiley.
- Asness, C., Moskowitz, T. & Pedersen, L. (2013). Value and Momentum Everywhere. Journal of Finance, 68(3), 929–985.
When It Works / When It Fails
- Rate-trending environments with sustained directional cycles
- Macro regimes with persistent central bank rate moves
- Portfolios wanting explicit duration or carry exposure
- Range-bound interest rates with no directional trend
- Rate spikes driven by non-macro idiosyncratic factors
- When equity-rate correlation breaks down (e.g. stagflation)