Tools · Process

Blindspot & Bias Auditor

Most calculators compound the number you want to believe. This one haircuts it. Enter the growth rate you are assuming, how sure you are, and how much recent results are driving the thesis. The engine applies published persistence and calibration facts — not a price target, not a buy/sell call.

Assumed CAGR
Audited CAGR
Blindspot gap ($)
Composite bias load
Terminal wealth: assumed — → audited —

Assumed path vs audited path

Bias load (0–100)

Overconfidence
Recency
Confirmation
Concentration

Are these calculators investment advice?

No. They do arithmetic. AlphaLens does the research. Every decision is yours.

Do the calculators store my portfolio?

Calculator inputs at /tools/ are processed in the browser. They are not treated as an account profile.

How to read this number

The audited CAGR is not a prediction — it's what your assumed growth rate looks like after three well-documented tendencies get applied to it: high growth rates rarely persist as long as they're assumed to, a recent hot streak gets over-weighted into the future, and high confidence itself is a risk factor, not a credential. The blindspot gap is the dollar difference between compounding your original number and compounding the haircut number over the same years. A wide gap doesn't mean the thesis is wrong — it means the thesis is doing a lot of unexamined work.

The engine

g = assumed CAGR as a decimal · T = years · C = confidence (0.50–0.99) M = recent-growth multiplier · principal = starting dollars persist_raw = clamp(0.06 + 0.018*max(g-0.08,0)/0.01 + 0.012*max(T-5,0), 0.06, 0.55) persist = clamp(persist_raw * profile_fade, 0.04, 0.62) recency = clamp(0.10*(M-1), 0, 0.36) overconf = clamp((C-0.55)*0.45, 0, 0.20) audited_g = clamp(g*(1-persist)*(1-recency)*(1-overconf), 0.03, g) assumedEnd = principal * (1+g)^T auditedEnd = principal * (1+audited_g)^T gap = assumedEnd - auditedEnd profile_fade: tech 1.15 · turnaround 1.25 · cyclical 1.20 · quality 0.75 · defensive 0.70 Bias load, 0-100 (radar + bars): overconfidence = clamp((C-0.5)/0.5*70 + max(g-0.12,0)/0.28*30, 0, 100) recency load = clamp((M-1)/3*100, 0, 100) confirmation = clamp((thesis is moat or growth ? 18 : 8) + (C-0.5)*90 + max(g-0.15,0)/0.25*25, 0, 100) concentration = clamp(weight/0.4*70*profile_vol + max(g-0.1,0)*80, 0, 100) composite = round(0.30*overconfidence + 0.25*recency load + 0.25*confirmation + 0.20*concentration) profile_vol: tech 1.20 · quality 0.80 · cyclical 1.25 · turnaround 1.35 · defensive 0.70 "weight" is entered as a percent of portfolio (2-60) and used above as a decimal (weight/100). Coefficients are disclosed and illustrative, not fitted to the ticker you typed — the ticker field is a label only.

Why the haircuts look like this

None of the three haircuts are invented for this page. Persistence: Hendrik Bessembinder's research on long-run stock returns found that the great majority of lifetime wealth creation in the market comes from a small minority of stocks, and that high growth rates are structurally hard to sustain for a decade or more — the persist term fades the assumed rate harder the higher it is, and harder for profiles (tech, turnarounds) that historically fade fastest. Recency: Daniel Kahneman's work on the availability heuristic documents that recent, vivid results get over-weighted relative to their actual predictive value — the recency term penalizes theses built mostly on the last stretch of hot results. Overconfidence: Philip Tetlock's forecasting research found that the most confident predictions are, on average, the least well-calibrated — the overconfidence term treats high stated certainty itself as a haircut, not a credential.

The specific coefficients above are illustrative and disclosed in full in the formula box — they are not fitted to NVDA, or to whatever ticker you typed, and they don't get more or less severe based on the label you enter. Change the ticker and nothing about the math changes; only the profile, thesis, and sliders move the number.

Maps onto the 15 frameworks

If this is the actual concern…Run this
Growth rate looks unsustainableFair Value Stress Test · What's Priced In
Thesis is a moat story with no counter-caseBull vs Bear + Moat · Competitor Moat Comparison
Earnings quality behind the growthEarnings Quality Analyzer · Accrual Screen
No pre-committed kill switchThesis Invalidation · Long-Term Thesis
Large single-name betPortfolio Risk & Fit
Moving the goalposts after the factTrading Journal Audit

Common questions

Is this investment advice?

No. Arithmetic. AlphaLens does the research.

Does it store my portfolio or fetch the ticker?

No. Browser only. Ticker is a label — there is no live fetch of prices, filings, or news for the ticker you type.

Why doesn't it say key-man risk for NVDA?

Because that would be a hallucination. Live data or nothing. This tool has no live data on any specific company — the CTA above names the framework that can actually look at company-specific risk on live filings.

Can I embed this?

Yes. iframe to https://norrisai.us/tools/bias-auditor/?embed=1 at height 920.

Go deeper

The haircut is arithmetic. Whether this specific company can beat it is a research question — not a slider.

Run this idea through AlphaLens →   Start free trial

Embed this calculator

Drop this on your own site. It links back to the full calculator on norrisai.us.

Disclaimer: NorrisAI AlphaLens tools are educational calculators only. They are not investment advice, a recommendation to buy or sell any security, or a substitute for a financial advisor. NorrisAI is not a registered investment adviser, broker-dealer, or fiduciary. Results are estimates based on the numbers you enter and the assumptions shown. Markets, taxes, fees, and company fundamentals change. You are solely responsible for your decisions.
Educational calculator only. Not investment advice. — NorrisAI AlphaLens
Calculator by NorrisAI AlphaLens

We'll send your numbers — nothing else.