A physics engine, and a verification tool enforced on top of it. PhysWall inverts a closed, published, non-linear physical law at a single measurement point — and refuses when the inverse is not unique. Seven laws, one engine, and the same refusal in all of them.
The answer is a band rather than a point, and for the same reason it is one in metrology: three of the four inputs were chosen rather than measured.
The technique has a name: a reverse DCF. A discounted cash flow model asks what a share is worth; run backwards it asks what implied growth rate the price already assumes. Rappaport and Mauboussin called it expectations investing.
It does not predict a price, recommend a trade, or say whether a company will grow. Nothing here tells you whether to buy.
We looked at market-direction prediction and rejected it: there is no closed law to invert, and a tool that guesses the future would contradict every other page on this site. What is here instead is the one thing that IS a calculation.
Somebody quotes a price. Working backwards from it gives the growth rate that price implies:
price 140, earnings 5, discount 9% -> the price already assumes 9.6% a year
If you assume 20%, you agree with the market and more. If you assume 8%, your figure is below what the price implies. The number is the market's assumption, not ours — we only extracted it.
Four inputs go into that calculation and three of them are chosen rather than measured. Each one moves the answer more than the price does:
discount rate 7% to 11% -> 4.3% to 14.0% terminal growth 1% to 4% -> 12.0% to 8.1% horizon 5 to 20 yr -> 15.3% to 7.0% earnings the one measurement
Reporting one number out of three guesses is the failure this whole site exists to refuse. So it sweeps every combination a practitioner would defend and reports the envelope:
45 combinations -> 3.9% to 20.8%
a spread of 17 points
That width is not a defect in the method. It is the finding — and any tool handing you a single figure here has hidden three assumptions inside it.
ABOVE ALL your assumption is higher than
every defensible combination
-> you disagree with the market
under any reading of it
BELOW ALL lower than every one
-> your figure is below every one
of them, in every reading
NOT SETTLED inside the envelope
-> the numbers do not decide it.
Your assumptions do, and
those are yours
UNDECIDABLE the model does not apply --
loss-making, or no earnings
NOT SETTLED is the useful one, and it is the same refusal this engine gives everywhere else, in an investor's words. A disagreement that survives every assumption is real. One that appears at 11% and vanishes at 9% was never about the company.
The calculation divides by earnings. At zero it returns 200% and at −1 it returns 200% again — numbers that are artefacts of the arithmetic, not facts about the company.
So it refuses. Not because the answer is large, but because this method does not apply there and saying so is worth more than a figure that looks like an answer.
A record of returns is a stated summary with data behind it, and stated summaries are what this engine checks:
+50% -30% +20% -10% +15% they report 9.0% the simple average you received 5.5% what actually compounded 1,000 became 1,304 -- not 1,539
Their arithmetic is correct and the figure is wrong. It is the same shape as a published measurement guide that stated an average its own five readings did not give — and that one survived twenty-six years.
A manager beats an index by 3% a year for eight years. How long would the record need to be before that is distinguishable from luck?
edge 5% a year 41 years edge 3% a year 114 years edge 2% a year 256 years
At 16% annual volatility. That is not an opinion about the manager — it is what the record can and cannot support, and it is arithmetic you can repeat.
Not about the arithmetic — that is shown, and you can redo it. What we can be wrong about is whether the check fits your case at all.
A valuation model that suits a steady business suits a start-up badly. A volatility figure from a calm decade understates a turbulent one. Those are judgements, they are yours, and this page cannot make them for you.
PhysWall was developed and architected by Gadi Zion.
Built on PhysWall — the same engine reads antenna bandwidth, conductor loss, bit erasure and heat limits. It answers what the measurement implies, and refuses when the measurement cannot say.⚠ Check this instead of believing it. Every number here reproduces from a source that is named, and the claims that turned out wrong are still printed next to what replaced them. The same engine runs all of these — it asks how much a measurement allows you to conclude, and refuses the same way in every field. The same engine runs all of these — it asks how much a measurement allows you to conclude, and refuses the same way in every field. How to check each one →