Hold a 3x S&P 500 ETF while the index is above its 200-day moving average; hold cash otherwise. One instrument, one daily signal — and a drawdown that will test anyone who runs it.
Most trading systems are a tangle of indicators, regimes, and discretionary overrides. Mercurio is the opposite. It runs a single, fully mechanical rule: when the S&P 500 closes above its 200-day moving average, hold UPRO — a 3x-leveraged S&P 500 ETF. When it closes below, hold cash (SHV, a short-term Treasury fund). That is the complete strategy. There is no stock picking, no second instrument, no machine learning in the loop.
The idea is not new — it is a well-known approach sometimes called 'Leverage for the Long Run.' The thesis is simple: leverage amplifies returns, and amplified returns are wonderful in a sustained uptrend and catastrophic in a crash. A long, slow trend filter — the 200-day average — tries to keep the leverage switched on during the good stretches and switched off during the bad ones. It will never be perfect, because a moving average reacts to prices that have already happened.
Take the index's long-run trend as a regime switch. Be leveraged when the trend is up, in cash when it is not. Accept that the switch is slow and the leverage is violent — and size your expectations accordingly.
Once a day, the S&P 500's close is compared with its own 200-day moving average. Above the line, the portfolio holds the 3x ETF (UPRO). Below it, the portfolio sits in cash (SHV, T-bills). That is the entire strategy.
Because the signal is the index's own trend rather than a forecast, the system is honest about what it can and cannot do. It cannot predict tomorrow. It can only follow a line that summarizes the last several months, and rotate to safety once that line is broken.
Replayed over five years of real UPRO daily returns — including the actual cost of leverage decay and financing — on $25,000 of paper capital, the strategy grew to about $56,234. That is a +124.8% total return, a 17.7% compound annual growth rate, and 2.25x the starting capital. The S&P 500 itself, with dividends reinvested, returned about +84.9% over the same window.
This is the single most important sentence on the whole site. The strategy's Sharpe ratio — return per unit of risk — was 0.66. Simply buying and holding the S&P 500 scored 0.77. The leveraged strategy made more money, but it made it less efficiently and far more violently. Higher return is not the same as a better investment.
Leverage cuts both ways, and the 200-day filter is slow. The worst peak-to-trough loss in the backtest was about 52% — more than half the account, gone, before it recovered. That is roughly double the S&P 500's own 25% worst drawdown over the same period. A single day lost as much as 9%; a single week lost 18%.
The year-by-year view shows where the pain lands. 2021 was a leveraged bull and the strategy gained about +41%. Then came 2022 — and even with the timing rule, the strategy lost -41.5%. The 200-day average does not catch the top; it catches the trend after it has already turned, so the account rides part of the decline down before rotating to cash.
No moving average, no AI, no reactive layer can prevent a leveraged crash. Markets gap; information arrives after prices move. A 200-day signal will always be late to the worst single-day shocks. Anyone who tells you a timing rule makes leverage safe is selling something. The rule reduces time spent in deep bear markets — it does not abolish the tail.
Earlier versions of Mercurio leaned on machine intelligence to pick trades. This version does not, and it would be dishonest to pretend otherwise. The rule is mechanical. The only honest role left for an AI layer is commentary and monitoring — summarizing the current market-risk picture, flagging that the index is approaching its 200-day line, narrating what the rule is doing and why. It does not choose the trade, and it cannot prevent the crash. If it were switched off entirely, the strategy would behave identically.
If it is riskier than the index on a risk-adjusted basis, why bother? Because for an investor who specifically wants higher absolute returns, can tolerate a ~50% drawdown without panic-selling, and accepts that the edge is fragile, the long-run trend filter is a disciplined, rules-based way to take that bet. It is not a free lunch, it is not a guarantee, and it is not for most people. It is a transparent, testable expression of one specific risk appetite.
The most important caveat: this is a backtest, not a forecast, and Mercurio runs it on paper capital only. Read exactly how it was tested in the backtesting deep dive, and how fragile the edge is to small changes in the fragility analysis.
Disclaimer. Every figure here is from a historical simulation on paper capital. It is not live trading, not a forecast, and not financial advice. Leveraged ETFs carry substantial risk, including large and rapid losses. Past performance does not guarantee future results.
