alphy.live: an automated leveraged-tech trading strategy, live

The goal

Beat buy-and-hold SPY, QQQ, and TQQQ.

A leveraged-Nasdaq strategy, run by a bot, built to capture the upside of 3x tech without the drawdown almost nobody can actually hold through. Every trade and position posted here, daily. The red days too.

Account

since ·

Return

Pending

since inception

Today

last close

Exposure

28%

in risk assets

Equity

The dial

0.27

exposure dial

defensive · 0.15normal1.00 · full

Reset every morning from volatility, crash risk, valuation, and drawdown. It decides how much leverage to carry that day. Whatever it holds back sits in T-bills earning yield. The exact rules are below.

Current position

TQQQ27.8%

T-bills72.2%

risk 28% T-bills / cash 72%

The bar: beat buy-and-hold

The benchmarks are the three plain ways to own tech: the S&P (SPY), the Nasdaq (QQQ), or the Nasdaq levered 3x and held straight through (TQQQ). Here is how each did over two windows, real returns you can verify yourself.

The last 10 years2016-2025 · a bull that bounced back from every dip

Hold	CAGR	Max DD
SPY	+14.9%	-34%
QQQ	+19.7%	-35%
TQQQ	+37.9%	-82%
alphy	+37%	-27%

A full cycle2000-2025 · dot-com crash and 2008 included

Hold	CAGR	Max DD
SPY	+8.0%	-55%
QQQ	+8.2%	-83%
TQQQ	+0.8%	-99.9%
alphy	~+20%	-44%

Real total return: SPY and QQQ actual; TQQQ actual since its 2010 launch, 3x-daily synthetic (conservative, understates the drag) before that. alphy: frozen, pre-registered backtest, whole account including its T-bill ballast.

Here is the trap. TQQQ's +37.9% only exists because the last decade handed leverage a bull with no crash that failed to bounce back. Put a real cycle around it and 3x-and-hold becomes the worst of the four: +0.8% a year, after a drawdown that took $10,000 down to about $4. It fell 99% in the dot-com crash and bottomed at the 2009 low, and digging out takes the better part of two decades, if you even held through it. alphy targets roughly 20% a year across that same cycle. Matching 3x in its best decade at a third of the risk, and beating it many times over across a real one, is the point. A backtest is not a promise, which is why the live curve above is the real scorecard.

Why 3x is a trap, and how to use it anyway

Leveraged ETFs earn their bad reputation. TQQQ resets every single day. It multiplies each day's move by three, not the year's. In a choppy market that quietly bleeds you out, even when the index ends flat.

The index, round trip

Down 10%1.000 → 0.900

Up 11.1%0.900 → 1.000

Netflat

The 3x fund, same two days

Down 30%1.000 → 0.700

Up 33.3%0.700 → 0.933

Net-6.7%

The index went nowhere and the 3x fund still lost money. Run that through a sideways year and leverage grinds you down for free. That is where the "never hold overnight" rule comes from, and for most people it is good advice.

But the bleed is not random. It lives in high volatility and downtrends. The mirror image is just as true: in a calm, steady climb, 3x compounds to more than three times the index. And the Nasdaq has climbed for decades, software and chips keep eating a bigger share of the economy. So the edge is not holding forever, and it is not day-trading. It is holding the leverage only while the trend is intact and volatility is low, and stepping aside when it is not. The drawdowns are still real: a 35% dip is a normal bad stretch. The point is not to dodge every one. It is to skip the worst of the decay and survive to compound on the other side. That switch is the dial at the top of this page, reset every morning, frozen before it was ever tested.

Inside the dial · the actual rules

No black box. Four checks run every morning. Each returns a number between 0.15 and 1. Multiply them together, cap at 1, and that is the day's exposure, exactly the value shown up top.

01 Volatility target

E_vol = min(1, 0.40 / recent vol)

Hold a steady amount of risk, not a steady number of shares. When TQQQ's trailing 20-day volatility runs hot, carry less of it; when markets are calm, carry up to a full position. The target is 40% annualized.

02 Crash brake

E_regime = 0.15 in a crisis, else 1

A fast, binary switch. When the tape tips into a high-volatility crisis, the setup that tends to precede the worst days, cut leverage to a token 15% until it clears.

03 Valuation cap

E_val = 0.5 if QQQ > 1.35 × 3-yr avg

When the Nasdaq is stretched more than 35% above its own three-year average, the 1999 setup, halve the position. Expensive markets fall harder and further.

04 Drawdown stop

E_cb = 0.15 after a -35% drawdown

Down 35% from its own high, the strategy slams leverage to 15% and holds it there, re-engaging only after the Nasdaq spends 40 straight sessions back above its 200-day average. This is the rule that survives a long, grinding bear.

Dial = vol × crash × valuation × drawdown, capped at 1.0. Whatever falls out, roughly 0.15 fully defensive to 1.00 all-in, is how much TQQQ the bot holds that day. The rest sits in T-bills earning yield.

These are the live rules, unchanged from the spec that was frozen before any testing. Writing them down first is the whole discipline: they cannot be quietly tuned to fit the past.

Trade log · every fill, timestamped

Timestamp (ET)	Side	Sym	Qty	Fill	Signal
No trades yet. Deploys when the account funds.