Learn
Plain-English explainers, deep dives, and backtest results on the SMA200 and adjacent trend-following topics. Everything written by a builder, not a guru.
TLT has been below its 200-day SMA for 85 days. The 5-year analog set.
Long bonds tripped the SMA200 trend filter back in March and have not recovered. 85 calendar days below is the fifth-longest streak of the last five years; the longest was 648 days through the 2022 bond bear. Here is what the framework actually says to do, what the recent analogs look like, and why TLT trips this filter more often than equity ETFs do.
Read article →I open-sourced the synthetic-LETF return engine. It includes the borrow-cost fix.
If you have ever run a synthetic TQQQ backtest using the standard L * daily_return formula, your numbers are roughly 60% too high. sma200-bt is the corrected version, MIT-licensed, calibrated to within 5% of real TQQQ over 2015 to 2024. Install with pip, use it for any leverage and any underlying.
Read article →For Leveraged ETFs, the Underlying Matters More Than the Leverage
The conventional wisdom is 'use less leverage to reduce drawdown.' That's correct in isolation but wrong as a cross-product comparison. 27 years of synthetic-LETF backtests show 3x SPY (UPRO) with the SMA200 filter beats 2x QQQ (QLD) on both CAGR AND max drawdown. The underlying's tail-bear profile dominates the leverage level. Practical implication: pick underlying first, then size leverage.
Read article →The Semi Collapse Scenario: What the SMA200 Actually Does to TQQQ and SOXL if Semis Have Their Own Dotcom
Most TQQQ and SOXL holders are quietly comforting themselves with 'I'll use the SMA200 to bail if it cracks.' 27 years of synthetic backtests including the actual dotcom collapse show that escape plan works for UPRO (-67% max drawdown) but doesn't save TQQQ (-95%) or SOXL (-99%). The underlying's vol matters more than the filter when leverage is high.
Read article →The Third Defensive Bucket: Managed Futures as an Alternative to Cash and Gold
When the SMA200 trend filter says 'go flat' on your leveraged equity sleeve, the conventional defensive options are cash (boring), gold (good), or long bonds (broke in 2022). There's a third option most retail strategies don't consider: managed futures via DBMF. The 7-year data shows it adds +0.076 Sharpe over cash and delivers the best single-asset drawdown protection of any defensive bucket tested.
Read article →Why One Indicator Isn't Enough (and Why Three Tuned Indicators Aren't Better)
The SMA200 gives you a binary read on long-term trend. Useful, but incomplete. Most traders try to fix this by stacking RSI, MACD, and Bollinger bands until the chart is unreadable. There's a better way, and a walk-forward test reveals why most "tuned" confluence strategies are statistical noise dressed up as edge.
Read article →The Hidden Cost Every Leveraged-ETF Backtest Ignores
Most synthetic LETF backtests overstate returns by ~60% over 10 years. The bug is one missing term in the daily-return formula. This piece shows the calibration proof against real TQQQ, the corrected long-window numbers, and what the SMA200 trend filter actually does once leveraged ETFs are modeled honestly.
Read article →Does the 200-day Moving Average Actually Beat Buy-and-Hold?
A look at 16 years of data on QQQ, TQQQ, SOXL, UPRO comparing the simplest trend-filter rule against passive buy-and-hold. The answer is more interesting than either side wants to admit.
Read article →