Constructing an Indicator System for Bond Futures: A Quantitative Perspective

In discretionary trading, indicators serve as auxiliary decision-making tools; in systematic trading (CTA), they become the raw material for constructing factors. For Chinese government bond futures (T/TF/TS), naively applying generic technical analysis is insufficient. Instead, one should build an indicator system grounded in three dimensions: macro inertia, volatility regime, and term structure.

Trend Factors: Capturing Macro Inertia

The underlying asset of government bond futures is the risk-free interest rate. Interest rate cycles are driven by macroeconomic fundamentals (GDP, CPI) and monetary policy operations (MLF, OMO), exhibiting strong trend persistence. Trend-following factors aim to capture this structural momentum.

MACD (Moving Average Convergence Divergence)

MACD simultaneously encodes trend-following information (via moving averages) and momentum information (via the divergence from the signal line), making it one of the most versatile trend indicators.

The MACD line is defined as the difference between a fast and a slow exponential moving average:

The signal line (DEA) is a smoothed version of the MACD line:

The histogram represents the gap between the two:

Three distinct usage modes emerge from this decomposition. First, the zero line acts as a trend filter: when $\text{MACD}_t > 0$, the market is in a bullish regime, and only long signals should be entertained; conversely, $\text{MACD}_t < 0$ indicates a bearish regime. Second, the crossover between the MACD line and the DEA line serves as an entry trigger. A golden cross above the zero line — where the MACD line crosses above the DEA from below — signals trend acceleration with the highest win rate, whereas a golden cross below the zero line often corresponds to a short-lived rebound. Third, histogram divergence provides a reversal warning: when price makes a new high but the histogram area shrinks, the trend is losing momentum, signaling an opportune moment to exit.

Moving Averages

Moving averages of different horizons carry distinct structural interpretations in the bond market. The 5-period and 10-period MAs (roughly corresponding to weekly and biweekly horizons) represent the speculative cost line — a breach below MA5 typically indicates dissipation of short-term momentum. The 20-period MA (monthly line) serves as a trend demarcation: it coincides with the middle band of the Bollinger Bands, and holding above MA20 signals a healthy intermediate trend. The 60-period MA (quarterly line) aligns with the rebalancing cycle of institutional investors and often marks the bull-bear boundary.

The effectiveness of moving averages in bond futures is not merely a technical artifact. The market hosts a substantial amount of trend-following capital, and moving averages frequently constitute material support and resistance levels.

Volatility Factors: Risk Control and Regime Identification

Government bond futures are typically low-volatility instruments, yet they exhibit脉冲式 (pulse-like) behavior around policy announcement dates — central bank reserve requirement ratio cuts, MLF operations, and the like. Volatility factors serve the dual purpose of risk management and regime identification.

ATR (Average True Range)

ATR measures the true thermal intensity of the market. The True Range is defined as:

where $H_t$, $L_t$, and $C_t$ denote the high, low, and close prices at time $t$, respectively. ATR is the moving average of TR, typically over 14 periods.

ATR serves three roles in a systematic framework. As a risk management foundation, ATR enables dynamic stop-loss: rather than using a fixed point-based stop, one should adopt $2 \times \text{ATR}$, which automatically widens during volatile periods and tightens during calm ones. Position sizing follows a similar principle:

ensuring that higher volatility translates into smaller position sizes. As a momentum validator, ATR helps distinguish genuine breakouts from false ones. A price breakout accompanied by declining ATR (a contraction breakout) is suspect; only the combination of rising price and rising ATR constitutes a confirmed breakout. As a noise filter, intraday price movements smaller than $0.5 \times \text{ATR}$ can be classified as noise and suppressed from generating trading signals.

Bollinger Bands

The Bollinger Bands integrate trend (middle band) and volatility (upper and lower bands) into a single indicator. With the standard parameterization $(N=20, K=2)$:

Under the assumption of normality, prices fall within $\pm 2\sigma$ approximately 95.4% of the time. A breach of the 2-sigma band thus constitutes a low-probability event, signaling either the onset of an extreme move or a limit of overbought/oversold conditions. At 3 sigma, which encompasses 99.7% of the distribution, reversion becomes highly probable — a classic entry point for mean-reversion strategies.

Two quantitative applications deserve attention. The volatility squeeze is detected when the bandwidth, defined as $\text{Bandwidth} = (\text{Upper} - \text{Lower}) / \text{Middle}$, falls to a historically low level (e.g., the 5th percentile over the past six months). A squeeze signals impending regime change; the recommended approach is to await a directional breakout before entering. The mean-reversion mode applies when bandwidth is stable (non-squeezing): bond futures tend to oscillate between the bands, and combining a band touch with an oscillator signal (e.g., price touching the upper band with RSI overbought) yields a high-conviction mean-reversion trade.

Momentum and Reversal Factors: Entry Timing Refinement

Oscillators such as RSI and KDJ are effective for pinpointing entries, but only when deployed in conjunction with trend indicators. The guiding principle is “follow the major trend, fade the minor trend” — taking oscillator signals only in the direction of the prevailing trend. Using KDJ crossovers in isolation leads to frequent stop-outs due to the intraday noise characteristic of bond futures.

RSI (Relative Strength Index)

RSI measures the relative balance between bullish and bearish forces and is inherently smoother than KDJ:

where $\Delta P^+$ and $\Delta P^-$ denote the magnitude of positive and negative price changes, respectively. Two parameter regimes serve different purposes. With $N=6$, RSI becomes a sensitive mean-reversion tool: in range-bound markets, RSI above 80 signals overbought conditions (short opportunity) and RSI below 20 signals oversold conditions (long opportunity), particularly suited for pairs trading. With $N=14$, RSI functions as a trend health monitor: during an established uptrend, RSI typically oscillates in the $[40, 80]$ range; a break below 40 signals potential trend exhaustion and warrants ceasing long exposure.

KDJ (Stochastic Oscillator)

KDJ introduces the highest and lowest prices into its computation, making it the most responsive of the three oscillators. The fast K line and the smoothed D line are computed as:

where $L_n$ and $H_n$ are the lowest low and highest high over the lookback period $n$. The J line, being the most sensitive, serves as a precise entry trigger: in a pullback within an uptrend, the J line emerging from negative territory ($J < 0$) and turning upward marks a high-resolution entry point. The K and D lines determine short-term trend direction: as long as $K > D$ with an expanding spread, the short-term trend remains upward. Crucially, when $J > 100$ or $J < 0$, the oscillator enters a passivation zone, indicating extreme trend strength. Trading against the trend in the passivation zone — shorting when $J > 100$ or buying when $J < 0$ — is empirically hazardous and should be avoided.

Term Structure Factors: The Defining Feature of Bond Futures

Term structure analysis distinguishes bond futures from commodity futures. The relationship between contracts of different maturities encodes information about the yield curve and liquidity expectations.

Inter-Commodity Spread

The yield curve’s shape changes — steepening or flattening — are captured by cross-tenor spreads. The canonical construction is $2 \times \text{TF} - \text{T}$ (steepening/flattening the curve). The 5-year contract (TF) is more sensitive to liquidity conditions, while the 10-year contract (T) responds more to fundamental macro expectations. When liquidity is accommodative but economic expectations are pessimistic, going long TF and short T captures the bull steepening dynamic.

Inter-Temporal Spread

The calendar spread between near-month and far-month contracts reflects market expectations of liquidity conditions and carries information about rolling costs. During the contract roll period, the spread exhibits regular convergence or divergence patterns that can be systematically exploited.

Microstructure: Auxiliary Confirmation

Open interest and volume provide supplementary confirmation. Rising prices accompanied by increasing open interest indicate active long accumulation and a healthy trend; rising prices with declining open interest suggest short covering, which may signal trend exhaustion. Breakouts at key levels (prior highs or lows) require significant volume confirmation — a volume spike — otherwise the breakout is likely false.

Strategy Framework: Intraday Spread Mean Reversion

As a concrete illustration of how these indicators combine, consider an intraday spread mean-reversion framework targeting the $2 \times \text{TF} - \text{T}$ spread. This is a prototypical mean-reversion strategy: while single-sided trend trades rarely generate ten high-win-rate opportunities per day, spreads tend to oscillate intraday, making them natural candidates for high-frequency mean-reversion.

Instrument Construction

The spread series is synthesized as $\text{Spread}_t = 2 \times P^{\text{TF}}_t - P^{\text{T}}_t$, where $P^{\text{TF}}_t$ and $P^{\text{T}}_t$ are the mid-prices of the 5-year and 10-year contracts, respectively. The notional ratio of 2:1 (TF to T) approximates a duration-neutral position.

Multi-Timeframe Configuration

A multi-timeframe architecture separates regime identification from entry timing. On the 15-minute chart, Bollinger Bands $(20, 2)$ determine the prevailing regime: if bandwidth is stable, the spread is oscillating and suitable for mean-reversion; if bandwidth is expanding rapidly, trending conditions prevail and counter-trend trades should be suspended. On the 1-minute chart, KDJ $(9, 3, 3)$ and RSI $(14)$ provide the fine-grained entry signals.

Entry Logic

For a long spread entry, three conditions must align: (1) the 15-minute Bollinger Bands indicate a non-trending regime (spread near or below the middle band, bandwidth flat); (2) the spread touches the lower band on the 1-minute chart; (3) the 1-minute KDJ J line is below 0 and turning upward, with RSI below 30. The logic is that the higher timeframe rejects the bearish hypothesis while the lower timeframe confirms an oversold reversal. The short spread entry is the mirror image: the 15-minute regime is non-trending with spread near or above the middle band; the spread touches the 1-minute upper band; and the J line exceeds 100 while turning downward with RSI above 70.

Exit Logic

Two take-profit mechanisms operate in parallel: a fixed target where the spread reverts by a specified number of ticks, and an indicator-based target where the spread returns to the 1-minute Bollinger middle band. Stop-loss is governed by ATR: the stop is placed at $\pm 2 \times \text{ATR}_{1\text{min}}$ from the entry price. A time-based stop liquidates the position if it remains unprofitable after a defined holding period, consistent with the principle that intraday mean-reversion trades should not be held through adverse moves.

Factor Library Construction

Building a robust factor library does not require deploying every indicator simultaneously. A layered approach is recommended. The core layer consists of one trend factor (e.g., MACD) and one volatility factor (e.g., ATR). The filter layer adds a spread factor or an oscillator (e.g., RSI) to eliminate low-quality signals. The risk control layer ensures that ATR-based position sizing is enforced systematically, calibrating exposure to realized volatility at all times.

References

1. Wilder, J. W. (1978). New Concepts in Technical Trading Systems. Trend Research.
2. Bollinger, J. (2001). Bollinger on Bollinger Bands. McGraw-Hill.
3. Murphy, J. J. (1999). Technical Analysis of the Financial Markets. New York Institute of Finance.
4. Chan, E. P. (2013). Algorithmic Trading: Winning Strategies and Their Rationale. Wiley.
5. Faber, M. T. (2007). A Quantitative Approach to Tactical Asset Allocation. Journal of Wealth Management, 9(4), 69–79.
6. Avellaneda, M., & Lee, J.-H. (2010). Statistical Arbitrage in the U.S. Equities Market. Quantitative Finance, 10(7), 761–782.
7. Baltas, A.-N., & Kosowski, R. (2015). Demystifying Time-Series Momentum Strategies: Volatility Estimation, Position Sizing and the Impact of Overlapping Returns. Journal of Alternative Investments, 18(1), 50–70.