Desbloquear esta nota

Taiwan index futures quadrupled in 11 years. So why not just hold? We put a very common trend indicator through an honest backtest. Supertrend ships with most charting tools: price above the line means uptrend, below it means downtrend. This TXF strategy (TXF Supertrend Trend, official and free) does exactly one thing — go long Taiwan index futures when the 1-hour close holds above Supertrend, go flat when it breaks below — and sizes the position by volatility. Backtested January 2015 to October 2026: +2,375% total return, about 31.4% annualized, Sharpe 1.67, max drawdown −16.5%. It beat buy-and-hold in 9 of the 11 full calendar years from 2015 to 2025. How does Supertrend work on TXF 1-hour bars? Take the ATR (a smoothed average of true range, 20 bars) and draw a band 1.5×ATR above and below each bar's high-low midpoint. In an uptrend only the lower band matters: it can ratchet up but never loosen, and the first close below it flips the trend down. In a downtrend, the upper band takes over. The strategy holds TXF only while the trend is up and goes flat the moment it flips — it never shorts. Day and night sessions give about 21 one-hour bars a day; each close is evaluated. Size isn't a fixed contract count. The strategy scales exposure to target roughly 30% annualized volatility using the last 30 trading days of realized volatility, capped at 2×. Calm markets get a bigger position; violent ones get a smaller one automatically. Why long only? Because going flat on a breakdown hands the bear phases back to the market. In 2022, TXF fell 23%; this strategy lost 2.7%. Over the full period, buy-and-hold's worst drawdown was −32%; the strategy's was −16.5% — roughly half. Backtest performance (2015–2026, 1-hour bars) MetricValueTotal return+2,375% (buy-and-hold +425%)Annualized~31.4% (buy-and-hold ~15.2%)Sharpe1.67Max drawdown−16.5% (buy-and-hold −32%)Entries~150 a year, average hold under a dayMCPT p-value< 0.001 (none of 2,000 random shuffles did better) Includes 0.005% cost per side (futures tax + commission + 1 tick of slippage). The position is forced flat on the bar before each monthly settlement, so contract-roll gaps are never booked as fake profit. Flat about 44% of the time. Backtest results, not a guarantee of future performance. Four things to know before you run it First, the 1.5 multiplier was chosen on the same data. The ATR period barely matters — anywhere from 10 to 30, Sharpe stays between 1.59 and 1.67 — but move the multiplier one step (1.25 or 1.75) and Sharpe drops to about 1.4. It still passes our robustness gate (neighborhood average ÷ chosen cell = 0.89), but there is no out-of-sample period, so expect live results to come in below the backtest. Second, it doesn't win every year: it trailed buy-and-hold in 2016 (+7.1% vs +11.7%) and 2021 (+12.8% vs +23.5%), and is behind so far in 2026 (+54.9% vs +67.3%) — when a strong bull market chops it out, it leaves money on the table. Third, it trades often, about 150 entries a year; if your slippage is worse than one tick, that eats into returns. Fourth, 2015–2016 data covers the day session only; night sessions are included from 2017. The rule is simple. Running it every hour is the work. The logic is all above. What takes time is recomputing Supertrend every hour across day and night sessions, resizing contracts to volatility, stepping aside at settlement and placing orders without missing a bar. Blave Agent runs that chain for you: add this free strategy from the library, then tell your Agent to deploy it. See the TXF Supertrend strategy in the library →