Methodology

Core principles

1. Primary sources only. Every quantitative claim links to a primary source. Secondary sources (other personal finance sites, press coverage, forum threads) are not acceptable as the basis for a published number. If we cannot find a primary source, we omit the claim.
2. Math is published, not hidden. When we say a card is the best 2% cash back on rotating categories, we show the effective rate after the category cap. When we say a Roth IRA conversion is worth it at a given tax bracket, we publish the year-by-year cash-flow table.
3. Affiliate payout never enters the ranking model. The model is run on product features only. Affiliate relationships are checked after the ranking is finalized, only to confirm the affiliate link points to the same product the model ranked.
4. Reader-relevant math beats academic optimality. A theoretical strategy that requires a 30-step quarterly optimization is worse for most readers than a 90%-as-good strategy they will actually execute. We say so.

Primary sources we use

finbarrow's quantitative claims are anchored in the following primary sources. We do not rely on aggregator data unless explicitly noted as an aggregator estimate.

• CFPB (Consumer Financial Protection Bureau) — credit card agreement database, mortgage data, complaint data, regulatory bulletins.
• FDIC — bank deposit insurance limits, national deposit rate caps, bank failure data, weekly national rates and rate caps.
• NCUA — credit union deposit insurance, share account data.
• FRB (Federal Reserve Board) — H.15 selected interest rates, G.19 consumer credit data, FOMC dot plot for forward rate guidance context.
• IRS — contribution limits (Roth IRA, Traditional IRA, 401(k), HSA, SEP-IRA, SIMPLE IRA), tax brackets, RMD tables, MAGI thresholds, Publication 590-A and 590-B for retirement accounts, Publication 936 for mortgage interest, Form instructions where dispositive.
• FICO — official FICO scoring factor weights, score range definitions, version differences (FICO 8 vs FICO 9 vs FICO 10), industry-specific FICO scores (auto, bankcard).
• FINRA — broker check data, suitability rules, margin rules.
• SEC — Form ADV for RIA disclosure, investor.gov for retail-investor education, Regulation Best Interest framework.
• Issuer terms — for any specific product (credit card sign-up bonus, APY, mortgage points), the issuer's current marketing page is authoritative. We link to it and capture the date checked.

Credit card ranking model

The credit card ranking model computes an effective rewards rate for each card across a published spending profile, plus a fee adjustment, plus a sign-up bonus value normalized over a 5-year horizon.

Inputs

• Base earn rate, category earn rates, and category caps.
• Annual fee, foreign transaction fee, balance transfer fee, late fee structure.
• Sign-up bonus value, minimum spend requirement (MSR), and timeframe.
• Redemption mechanics: cash back vs. points (with conservative cents-per-point valuations from publicly observed redemptions, not aspirational valuations).
• Issuer-specific approval rules where relevant (Chase 5/24, AmEx 2-in-90, Bank of America 2/3/4 etc.).

Math

For each spending profile (we publish profiles for grocery-heavy, gas-heavy, travel-heavy, flat-spend), the model computes annual rewards = Σ (category spend × category rate, capped at the category cap) minus annual fee, plus sign-up bonus amortized over the 5-year horizon. The card with the highest effective rate for the profile wins for that profile. The points valuations are documented per card with a "conservative" and "optimized" figure; the ranking uses the conservative figure.

Savings / CD / money market ranking model

For deposit accounts, the model uses APY as the primary signal, with explicit penalties for:

• Minimum balance requirements that materially exceed the typical reader's likely deposit.
• Direct-deposit or activity requirements that effectively gate the headline rate.
• Promotional / introductory rates with a known step-down date (we publish the post-promo rate).
• Bank-specific service-fee structures that can erode yield.

FDIC or NCUA insurance status is a hard requirement, not a scoring input — uninsured products are excluded entirely. Cross-bank verification of headline APY is performed at the issuer's own rate page on the date of publication.

Investing / broker / retirement account ranking model

Broker rankings weight expense ratios for the broker's flagship mutual funds and ETFs, commission structure (with stock and ETF commissions normalized to zero where applicable), fractional-share availability, IRA fees, transfer fees, money-market sweep yields, and investor protection (SIPC limits, supplementary insurance). For retirement account strategies (Roth vs Traditional, mega backdoor Roth, in-service withdrawal), the model is bracket-aware and produces a year-by-year cash flow that the article publishes in full.

Loan and mortgage ranking model

Loan rankings use APR as the primary signal — not interest rate — because APR reflects origination fees, points, and other costs the borrower actually pays. For mortgages, the model considers points (and the break-even period for paying them), PMI triggers, prepayment penalty terms, and rate-lock policies. Where available, we anchor the rate environment to the FRB H.15 series.

Credit / FICO content standards

Credit-score articles use only FICO's published factor weights (payment history 35%, amounts owed 30%, length of credit history 15%, credit mix 10%, new credit 10%) — not myths or vendor talking points. Where finbarrow takes a position that diverges from common online advice (e.g., that the "30% utilization rule" is wrong — the data suggests <10% is materially better for high scores), we cite the FICO source and show why.

Update cadence

Rate-sensitive articles (HYSA picks, CD picks, mortgage rates, credit card sign-up bonuses) are reviewed monthly at minimum. Concept articles (how Roth IRA works, what FICO is) are reviewed at least every 12 months. Every article carries a "Last reviewed" date in the byline area. If something material changes between scheduled reviews — an issuer pulls a card, the IRS updates contribution limits, FICO releases a new scoring model — we update immediately.

Corrections

Errors happen. When we publish a number that turns out to be wrong, we correct it, update the "Last reviewed" date, and log the change with a short note at the foot of the article. We do not silently rewrite history. The correction policy is detailed in the editorial policy.

Educational content only. finbarrow is an independent editorial publication, not a licensed financial advisor, broker, tax preparer, or attorney. Always verify rates, terms, and eligibility with the issuer or a licensed professional before acting. See disclaimers and funding disclosures for details.