TLT algorithmic trading strategy

TLT (iShares 20+ Year Treasury Bond ETF) tracks long-duration U.S. government bonds with 20+ year maturities, serving as a critical instrument for portfolio diversification, risk management, and tactical trading. Treasury bonds typically move inversely to equities during market stress, making TLT an essential hedging tool against equity market downturns.

Data acquisition

The iShares 20+ year treasury bond ETF (TLT) has exhibited significant price fluctuations over the past 15 years, reflecting the inverse relationship with interest rates, with notable bull runs during economic downturns and periods of monetary easing, and steep declines during rising rate environments.

The TLT price history from 2010 to 2025 exhibits price fluctuations with a minimum of $57.75, maximum of $150.75, mean of $97.44, median $95.72 and 15.27% annualized volatility, reflecting the ETF's sensitivity to changing interest rate environments and macroeconomic conditions.

Trending signal generation

Trading signals are specific conditions or events that indicate optimal entry and exit points for trades based on predefined rules. In this TLT strategy, signals include short entries on month starts (yellow triangles), short exits five days later (red circles), long entries seven days before month-end (green triangles), and long exits one day before month-end (blue circles).

The signal generation function creates a calendar-based trading system that shorts TLT at month beginnings and goes long near month ends, with the plots visualizing both the full trading history and a zoomed-in recent year for clearer signal identification.

The full-period trading signals visualization (2010-2025) shows the systematic application of our month-based strategy across diverse market environments, generating 444 total signals (71 complete round-trip trades), 120 short entries, 71 short exits, 126 long entries, 127 long exits with an average of 50.83% success rate.

The recent-year chart provides clearer visualization of 28 trading signals with 3 complete trading cycles, showing how short entries at month beginnings achieve a 33.33% success rate, the average monthly signals are 0.46 short entries, 0.23 short exits, 0.69 long entries and 0.77 long exits.

Performance analysis

The performance analytics section quantifies and visualizes the TLT strategy's effectiveness against a buy-and-hold benchmark through equity curves and drawdown comparisons.

The performance visualization demonstrates buy & hold achieving 53.71% total return with 2.85% annualized growth and significant COVID-related peak, long strategy delivering minimal 4.45% total growth, combined strategy suffering -47.77% cumulative decline, and short strategy collapsing to zero, conclusively invalidating the calendar-based trading hypothesis.

The drawdown comparison reveals severe performance disparities with short strategy experiencing -100% maximum drawdown, long strategy maintaining moderate -21.09% drawdowns, combined strategy showing persistent -51.16% decline, and buy & hold exhibiting cyclical yet ultimately recoverable -48.35% drawdowns during 2010-2025.

Monthly returns analysis

This section aggregates the strategy and benchmark returns by month, calculating average monthly performance.

The average monthly returns chart shows January has the strongest returns for both strategies, with buy-and-hold (+0.38) significantly outperforming the combined strategy (+0.15), while October shows the largest negative returns for buy-and-hold (-0.09%) with the strategy slightly mitigating losses, and May-June represent the only period where the combined strategy (pink bars) shows competitive performance relative to buy-and-hold (blue bars).

Performance simulator

The simulator allows testing different investment amounts and timeframes on TLT trading strategies, demonstrating the calendar-based approach's performance and suggesting future research on selective implementation or trend-following integration.

Summary

A calendar-based TLT trading strategy was backtested, combining short positions at month beginnings (days 1-5) and long positions near month-ends (last 7 days) across a 15-year period.

Outcomes

The calendar-based strategy significantly underperformed buy-and-hold with poorer risk-adjusted returns. January showed the strategy's largest gap in performance (capturing only +0.15% vs buy-and-hold's +0.37%), while the short component performed disastrously, ultimately losing the entire investment.

Future research directions include selective monthly implementation, trend-following filters to avoid adverse trades, dynamic position sizing based on volatility, performance analysis across different interest rate environments, and reducing trading frequency to minimize costs.