Daniel Johnston

[Daniel Johnston @Dannyj1010]

@0x_Anni crazy how full music workflows are running straight off a laptop now

[Daniel Johnston @Dannyj1010]

@crimson_moti the scale of these ai character workflows is kind of wild

[Daniel Johnston @Dannyj1010]

@choopyplug1 using agents to build the agents themselves is such a cool feedback loop

[Daniel Johnston @Dannyj1010]

Quant investors can use CAGR to turn different holding periods into one comparable annual return, and the comparison can change how you read every backtest.

That is why CAGR matters more than the victory screenshot.

A portfolio that doubles in 3 years compounds at roughly 26.0%.

The same doubling over 10 years is roughly 7.2%.
Over 30 years, it is roughly 2.3%.

The formula is simple: ending value divided by beginning value, raised to one over the number of years, minus one.

CAGR turns different holding periods into a comparable annual rate.
It is useful for comparing a stock, fund, property, or systematic strategy without letting calendar length distort the result.

But CAGR is also a smoothing machine.
It converts a jagged path into one steady-looking number and hides drawdowns, volatility, leverage, liquidity constraints, and the sequence of returns.

Two strategies can share the same CAGR while demanding completely different risk tolerance.
A 15% CAGR with a 60% drawdown is not interchangeable with a 15% CAGR with shallow, recoverable losses.

Use CAGR to ask how fast capital compounded.
Then inspect the path to ask whether you could have stayed invested.

Bookmark this before the next clean signal turns into a bad decision. Follow for the mechanism behind it.

[kocer @kocer_eth]

https://t.co/AEKMKjEou3

[Daniel Johnston @Dannyj1010]

@0xNEXUSs1 mcp tool integration making automation like this so fast is crazy

[Daniel Johnston @Dannyj1010]

A desk quant at a mid-tier prop shop ran the same polarizer brainteaser in every interview loop for three years, and every candidate who answered with cos²(alpha) got cut, because the formula was never the question.

here is the setup:

a photon hits a polarizer
at angle alpha to the axis

classically, a wave loses a fraction of energy
transmitted intensity: cos²(alpha)
clean, deterministic, no ambiguity

but a photon is indivisible

it cannot shed part of itself
it either passes through completely
or it does not pass at all

so cos²(alpha) does not disappear

it becomes the probability the photon passes
same number, completely different claim
no longer a ratio of energies
now a statement about irreducible randomness
in a single trial

that is the object that changes your answer

you have to recognize when a formula
migrates from classical determinism
into a probability statement
and why those two things look identical on paper
but are not the same claim

the market translation is direct

a model treating a probability as a guaranteed fraction
will be calibrated correctly in expectation
and wrong on every individual event

that is not a rounding error
it is a category error in how you size,
hedge, and interpret a single outcome
versus a distribution of outcomes

knowing cos²(alpha) is free

knowing when to stop trusting
deterministic framing in a live position
that part does not come from a blackboard

Bookmark this before you mistake a public source for a paid edge. Follow for the next quant breakdown.

[kocer @kocer_eth]

https://t.co/AEKMKjEou3

[Daniel Johnston @Dannyj1010]

@theSethian schrodinger always shows up right when things start making sense

[Daniel Johnston @Dannyj1010]

When a quant model resists decomposition, the bottleneck is almost always non-linearity in the underlying equations, not the complexity of the data or the effort applied.

That's the assumption.
It's backwards.

Classically: m * d²x/dt² = -V'(x(t))

V'(x(t)) is evaluated at x(t).
Cubic potential → quadratic derivative.
The equation becomes non-linear.

in your models or signal work.

Non-linear equations don't compose.
You can't add two solutions
and get a third valid one.

Newton solved the two-body problem exactly.
The three-body problem has no closed form.
Not a gap in effort.
A structural property of non-linearity.

Quantum replaces that with Schrödinger:
i * ℏ * d(ψ)/dt = Ĥ * ψ

Ĥ is the Hamiltonian.
It is a linear operator.

Scale a solution → still a solution.
Add two solutions → the sum is also a solution.

That's superposition.
Not a physical quirk.
A direct consequence of the math.

Maxwell's equations: linear.
Electromagnetism yields to systematic analysis.

Einstein's field equations: non-linear.
Exact solutions are rare.
Numerical relativity needs supercomputers.

When a system resists decomposition,
the bottleneck is usually non-linearity.

Not complexity. Structure.

BOOKMARK and FOLLOW this if linearity ever shows up

[kocer @kocer_eth]

https://t.co/AEKMKjEou3

[Daniel Johnston @Dannyj1010]

@theSethian risk management really is the hardest part of trading

[Daniel Johnston @Dannyj1010]

@crimson_moti the consistency across all those different lighting conditions is honestly wild

[Daniel Johnston @Dannyj1010]

@ScottyBeamIO shipping agency quality sites with agents like this is seriously impressive

[Daniel Johnston @Dannyj1010]

@0xDezo totally agree breaking down assumptions reveals everything

[Daniel Johnston @Dannyj1010]

MIT gave away the linearity lesson that paid quant packs keep reselling, but the interview edge is knowing when the market stops honoring it.

Barton Zwiebach is not selling a prep course here.

He is writing the thing paid interview packs keep repackaging: why linear systems can be broken apart, solved, and recombined.

here is what is actually inside

A linear operator has two promises.

scale the input, and the output scales with it:
L(au) = aLu

add two inputs, and the output is the sum of the two outputs:
L(u1 + u2) = Lu1 + Lu2

That is why superposition is not a vibe.
If u1 and u2 both solve Lu = 0, then alpha u1 + beta u2 also solves it.

In physics, that lets electromagnetic waves or quantum states combine cleanly.

In quant interviews, it shows up as the same mental test: can you separate signal components without pretending the system stays linear after the market starts pushing back?

A portfolio model, a factor regression, a covariance estimate, a PDE approximation all of them become easier when the operator behaves like this.
Small pieces can be analyzed alone, then recombined.

But the moment transaction costs, crowding, regime shifts, nonlinear payoff, or feedback enters, the clean blackboard stops being the market.

The math is free.

The edge is knowing when the math stopped applying.

[kocer @kocer_eth]

https://t.co/AEKMKjEou3

[Daniel Johnston @Dannyj1010]

@paul4jennii trying to slice light and getting a swarm back instead is incredibly wild

[Daniel Johnston @Dannyj1010]

@kocer_eth framing quantum mechanics around concrete math rules makes it so much clearer

[Daniel Johnston @Dannyj1010]

@tsukiema_ this math breakdown is super clean for a tricky interview question

[Daniel Johnston @Dannyj1010]

@kitsune_xbt local setups are getting so good it makes renting tokens feel unnecessary

[Daniel Johnston @Dannyj1010]

@theSethian the circular funding loop in AI infrastructure is absolutely wild to map out

[Daniel Johnston @Dannyj1010]

@tonyprediction building a permanent wiki completely fixes the constant context reset headache

[Daniel Johnston @Dannyj1010]

A 3D Quantum Morphology Surface can expose skew, clustering, and tail risk that the S&P 500 return line compresses into a single curve.

The video shows a “Quantum Morphology Surface” next to market regimes from the GFC to EU Debt to the 2022 bear phase.

The useful idea is not “quantum” as magic.

It is shape, not average return.

A price chart compresses the market into one path.

A distribution view asks a different question: where did the probability mass move, how fat did the tails get, and did the center of the market actually shift?

BOOKMARK this if you study markets through regimes, not headlines.

In a bull run, the surface can look stable while cumulative returns climb.

In a crisis, the same return line can hide a very different object: skew, clustering, and tail risk changing faster than the index level.

That is why regime models fail when they only fit yesterday’s slope.

They are not just forecasting direction.

They are assuming the shape of future uncertainty.

Practical rule: before trusting any signal, ask whether it survives a change in distribution.

If it only works when volatility, correlations, and tails stay polite, it is probably a bull-market artifact with math around it.

The chart is the output.

The distribution is the battlefield.

[kocer @kocer_eth]

https://t.co/AEKMKjEou3