Rules-Based Trend Investing

The leveraged fund earns more on paper. Almost nobody holds it long enough to collect.

In our 20-year backtest, holding 2x Nasdaq straight through returned more than our system. It also fell 83% along the way, and almost no real person holds through that. They sell near the bottom and never see the recovery. Sphinx Edge trades some of that paper upside for a ride a disciplined investor can actually sit through: trend rules, cash when risk isn't paid for, and a tested worst-case loss roughly a third as deep. The return you can hold is the only return you get. No predictions. No discretionary calls. And every idea that failed testing is buried in public.

One strategy is live, trading real money since June 2026. The track record is being built in public.
CAGR17.4%
Max Drawdown-34%
Sharpe0.74
Test Window~20 yrs

Figures above are backtested (2007–2026) under realistic execution timing — orders modeled the morning after each signal, the way the live bot actually trades — net of estimated costs. The frictionless version scores about a point higher; we quote the number a real account can earn. Backtested results are hypothetical, not a live track record — see disclosure.

The Sphinx 30/70 Strategy

The core system pairs tactical leveraged growth exposure with a diversified trend-following sleeve. Every position is governed by an objective trend rule. When an asset's trend breaks, that money steps aside to cash — it never shorts and never holds leverage through a downtrend.

The 70% — the engine

Rides leveraged growth exposure while the trend is up, and moves fully to cash the moment the trend breaks. This is where most of the return comes from — and most of the risk, which is why it's governed by a hard rule, not a feeling.

The 30% — the ballast

Spread evenly across several unrelated things — growth, gold, long bonds, commodities — each held only while its own trend is up. When markets break, at least one of these usually holds the line.

Cash — the whole trick

When a trend breaks, the system steps aside, not down. It doesn't short and it doesn't buy inverse funds. Sitting in cash through the worst stretches is how it sidesteps the drawdowns that wreck most leveraged strategies.

The catch — what you give up

There's no free lunch. In choppy, trendless years the system gets whipsawed: it sells after a dip, then the market snaps back without it. It will underperform a simple buy-and-hold in those stretches. That's the deliberate trade — give back a little in the quiet years to avoid the catastrophic losses in the bad ones. We show you the trade-off instead of hiding it.

Research Summary

Live Record — Real Money
Live sinceJune 15, 2026
Trading days58
Since inception-4.9%
QLD same period-7.3%
SleeveSignalWeight
QLD — engineLong71.2%
QQQ — ballastLong7.6%
DBC — ballastLong8.9%
GLD — ballastCash — exited Aug 316.2% held in cash
TLT — ballastCash — below trend6.1% held in cash

Updated weekly, by hand, red or green. As of September 4, 2026. This was the week after the system's first realized loss, and nothing about the machine changed — which is the point. It took the small gold loss on Monday by rule, then finished the week at its second-best mark since launch, still down less than its benchmark at every close, as it has been for all fifty-eight days. Gold, for what one week is worth, has traded below where the bot sold it. The quiet story remains the commodity sleeve, now up 13% since inception while the tech engine recovers — uncorrelated ballast carrying the book through a tech drawdown, exactly the job it was hired for. Two sleeves sit in cash because their trends never confirmed or failed; that cash is the cushion. Fifty-eight days, four trades, zero overrides. This record will include every losing week, and every losing trade.

The goal isn't the biggest raw return — it's a better trade-off between growth, drawdown, and risk-adjusted performance. The table below is the full 20-year backtest against four honest benchmarks, net of estimated costs.

Growth of $1 — Sphinx 30/70 vs. holding the leveraged fund Backtested Jun 2006–Aug 2026 · weekly · log scale
$0.5 $1 $2 $5 $10 $20 $50 $100 2008 2010 2012 2014 2016 2018 2020 2022 2024 2026 3.4 years underwater 2008–2012: the 2x holder trails the system -83% from peak Sphinx 30/70 · $29 QLD hold · $96
Sphinx 30/70 QLD buy & hold (2x Nasdaq)

Real backtested daily data, downsampled weekly. The shaded region is the price of holding the leveraged fund: from September 2008 to January 2012 — 3.4 years — the QLD holder sat below the system, bottoming 83% under its peak in March 2009. The flat stretches in the sand line are the system sitting in cash while its trends were broken. QLD ends higher; the chart shows what a holder had to sit through to collect it. A live equity curve will be published once there is meaningful live history — the strategy went live in June 2026.

PortfolioCAGRMax DrawdownSharpeVolatility
Sphinx 30/7017.8%-34.9%0.7722.6%
QQQ Buy & Hold16.3%-53.4%0.7222.1%
SPY Buy & Hold11.3%-55.2%0.5719.4%
QLD Buy & Hold24.8%-83.1%0.6944.0%
60/40 SPY/IEF8.7%-31.4%0.6611.1%

All five rows computed in one run (June 2006 – July 2026), same window, same daily-close convention, so the comparison is apples to apples. The honest story is in the QLD row: holding 2x Nasdaq made more raw return (24.8%) — and handed you an 83% drawdown almost nobody can survive. Sphinx 30/70 gives up some of that return to cut the worst-case loss roughly in half. It beats the S&P 500 on return and risk, and beats the Nasdaq on drawdown. Realistic execution timing shaves roughly one point off the Sphinx row, as quoted at the top of the page. Backtested, net of estimated costs.

The Research, Honestly

Where the rest are buried.

You can trust the one that lived because of how many did not. Every idea here looked promising. Each was tested against a pre-committed bar, a fair benchmark, and real trading costs. Each failed honestly, and was put down. This is the research — most of it is a graveyard, on purpose.

Fifteen buried · one under watch · one survived
Crypto direction bets
"We went looking for an edge and found a casino."
No durable edge · closed
The prediction-market arb
"Fourteen thousand markets, zero free money."
Market too sharp
Rotation bot
"A flattering backtest built on looking at the future."
Beaten by buy-and-hold
MACD crossover
"Four lucky trades are not a track record."
Loses over a full history
Pairs trading
"Market-neutral, and neutral-ly losing."
Dead before costs
Risk-parity weighting
"Clever weights that lost to splitting it evenly."
No edge over equal weight
RSI-2 mean reversion
"It works. It just adds nothing new."
Same bet, lower payoff
Managed-futures funds
"Two of three failed. The third earned a stay, not a seat."
Under watch · re-test scheduled
Low-volatility factor
"Not an edge. A smaller bet wearing a costume."
Crowded away
The trend system's parent
"We tested the father. The son won."
Reacts a month too late
Cross-asset momentum
"A diversified basket in disguise. It lost to the basket."
Selection added nothing
FOXY · currency fund
"It passed every test. Then we measured what it cost to trade."
Un-tradeable spread
Breakout trading
"Beat holding on twelve of twelve funds. Then we made it wait one day."
Look-ahead · faked the edge
Read the teardown →
The copper / AI trade
"The story was real. The fund still lost to one we already owned."
Redundant & already run
Leveraged version of the fleet
"More borrowing, more bleeding, no more edge."
Margin is dead weight
Volatility selling
"Pennies in front of the steamroller."
Capped upside, full downside
The Sphinx 30/70 — the survivor
"Attacked six ways to break it. It didn't break."
Live · trading real money
The edge was never a secret strategy. It was the discipline to bury the ones that didn't work.

Why the Edge May Exist

Sphinx Edge is built around repeatable market behavior, not predictions. We don't claim to be smarter than the market — we claim to be more patient and more disciplined than the average participant.

Trend Persistence

Markets often keep moving in one direction longer than people expect. The system only participates while the long-term trend is favorable, and steps aside when it isn't.

Drawdown Avoidance

Avoiding the deepest losses matters more than catching every upside tick. A 50% loss needs a 100% gain just to break even. Cash is treated as a risk-control tool, not a failure to be invested.

Paid for Patience

The return isn't a clever prediction — it's compensation for sitting through discomfort while others panic or chase. That's a durable reason to get paid, and it's why the bad years are part of the deal, not a bug.

Important Disclosure

Backtested performance is hypothetical and does not guarantee future results. All performance figures on this site are backtested over 2006–2026 unless explicitly labeled as live. The strategy began trading real money in June 2026; a meaningful live track record does not yet exist. The research period was unusually favorable to U.S. technology and leveraged Nasdaq exposure, and forward returns may be materially lower. Taxes, account restrictions, execution differences, liquidity, slippage, data errors, and investor behavior can all affect real-world outcomes.

Sphinx Edge is a personal systematic-investing research project intended to communicate a process transparently. Nothing on this website is individualized investment advice, a guarantee of performance, an offer, or a recommendation to buy or sell any security. Sphinx Edge is not a registered investment adviser and does not manage money for others.