Desbloquear esta nota

Tuning parameters into a worse result happens more often than you'd think — what matters is whether you check. Supertrend is an ATR-based stop-and-reverse system: stay long while price holds above the lower band, flip short on a close below it, always in the market, no flat state. The same logic was first tested with default parameters across five assets side by side — TXF, BTC, ETH, SOL, DOGE. SOL had the best Sharpe (1.09), but that's a low bar: all five underperformed buy-and-hold. Tuning parameters, and almost getting fooled by the tool itself Scanning ATR Period × Multiplier for SOL, the first "best combination" the scan returned actually dropped Sharpe from 1.09 to 0.86. The problem wasn't the strategy — it was how the parameter picker works: it maximizes the neighbourhood average (to avoid a lone lucky spike), not the single highest cell, and this time that pick was worse than the untouched default. Checking both against the same grid showed the unmodified default (10, 3.0) was in fact one of the best individual cells on the whole grid. The real problem was drawdown, not return After reverting to the better parameters, Sharpe came back to 1.09 — but max drawdown was still -94%. The position was always fully sized (±1.0) with no volatility adjustment, and SOL is volatile enough that this produces deep drawdowns. Adding volatility targeting (30% annualized target, 2x leverage cap) changed the signal's shape entirely, requiring a full rescan — done in three passes (coarse, fine, then a full sweep across Period 1–30 and Multiplier 0.25–5.0) to confirm the final parameters (Period=7, Multiplier=1.5) weren't a false signal sitting at the edge of the tested range. Backtest performance (SOL, daily bars, 2020-10-14 to 2026-09-14) MetricValueTotal return+1045.47%Annualized+50.96%Sharpe1.40Max drawdown-26.30%MCPT p-value< 0.001 (n=2000) Net of a real 0.05% Binance perp taker fee. Backtest results, not a guarantee of future performance. Where the real value is: keeping position volatility in check The point of this strategy isn't to push return as high as possible. It's to stop an always-in-the-market stop-and-reverse system from swinging with SOL's raw volatility: position size scales to a 30% annualized volatility target with a 2x leverage cap, so the rougher the market, the smaller the position. The listed version's backtest max drawdown is -26.30%. The strategy is free. The hard part is keeping it running The logic is above. But watching ATR stop bands every day, deciding entries and exits, placing orders automatically, and handling volatility scaling — that's the part that costs traditional quant traders weeks. Blave Agent does that part for you. What you're paying for isn't the strategy — it's the execution that keeps it actually running. Three steps to get it live today ① Install Blave Agent② Top up from 300 TWD to activate your Blave Agent③ Tell it "load the official SOL Supertrend strategy" → one-click deploy, running around the clock No need to build it yourself — the official version is free, load it with one sentence. Minimum top-up 300 TWD to activate, then pay per use: server fee from 1,440 TWD/mo (30-day basis, 2.0 TWD/hr), plus AI usage. Install Blave Agent and get this strategy running today →