Disclaimer & risk notice

Not financial advice

Nothing on this site, in any daily brief, shock dossier, case file, or accompanying communication constitutes investment advice, financial advice, tax advice, legal advice, or a recommendation, solicitation, or offer to buy or sell any security or other financial instrument. The operator of Schmatz is not a registered investment adviser, broker-dealer, or financial professional. No content published here should be construed as personalized advice.

You are solely responsible for evaluating the merits, risks, and tax implications of any investment decisions you make. Consult your own qualified financial, legal, and tax advisors before acting on anything you read on Schmatz.

Paper trading only

The portfolio you see described as the "V7 portfolio," "shadow portfolio," or any variant thereof is a simulated paper portfolio. It does not hold, buy, or sell any real security. No real money is at risk in any holding shown on Schmatz. The paper portfolio is executed in recommend_only mode against a paper brokerage account and serves only to track the strategy's performance under realistic-execution assumptions.

Beta participants are observers of this simulated portfolio. You are not authorizing Schmatz to trade for you, and Schmatz does not have the technical or legal capacity to do so. If you choose to mirror any position you see published here in your own brokerage account, you do so entirely at your own discretion and risk.

Backtested and simulated performance

The headline strategy ("V7") is supported by a five-year walk-forward backtest covering 2021 through 2026 on a curated 81-stock universe. Reproducible results from this backtest are pinned to output/v7_canonical_result.json and regenerated by scripts/100_v7_canonical_backtest.py. The current canonical numbers are:

• Sharpe ratio: approximately 1.06 (walk-forward out-of-sample, 15bp round-trip costs)
• Pooled out-of-sample mean return per trade: approximately +1.29% (t-statistic 3.45, p ≈ 0.0006)
• Annualized return: approximately +9.5%
• Maximum drawdown: approximately −8.5%
• Portfolio beta vs. SPY: approximately 0.00

Honest caveats on these numbers

Backtested performance is not a forecast of live performance. Live execution typically realizes 15-30% less than paper backtest figures suggest, due to slippage, market impact, borrow availability, and unmodeled frictions. Past performance does not predict future performance. The backtest window (2021-2026) covers only one major macroeconomic regime; the strategy's behavior in regimes outside that window is unknown.

A Fama-French 3-factor regression of the strategy's daily returns shows statistically significant exposure to the size (SMB) and value (HML) factors. After adjustment for these factor exposures, the strategy's residual idiosyncratic alpha is positive (approximately +5.5% annualized) but not statistically significant at the 5% level (t-statistic 1.39, p ≈ 0.16). A meaningful portion of observed returns is therefore attributable to documented factor risk premia rather than to a unique strategy edge.

Independent statistical-rigor checks are summarized at docs/PROFESSOR_PROOFING.md in the project repository and include López de Prado & Bailey's deflated Sharpe ratio, which falls below the conventional 0.95 significance threshold given the number of strategy variants tested during development. Selection bias from iterative strategy development is a real and acknowledged risk.

Look-ahead bias and methodological caveats

The Schmatz backtest engine (src/backtest_engine.py, src/strategies/*) is point-in-time on price data and macro indicators. Trades enter at the next session's open after a signal fires, exit at end-of-session per the strategy's exit rule, and feature inputs are pulled from data available before the entry date. However, fundamentals sourced from third-party vendors (earnings, analyst grades, financial-health scores) are timestamped to filing date, not the as-of-knowledge date. Backtests that rely on fundamentals may capture small amounts of look-ahead bias on this dimension; this is a known limitation, not a hidden one.

Universe selection is also subject to survivorship bias. The V10 universe used throughout Schmatz includes the ~9,500 currently-listed US equities; it does NOT include delisted tickers. Backtests run on that universe will systematically overstate returns versus what would have been realized including failures and delistings. Independent analysis indicates V10 returns are inflated 15-30% by this effect on event-driven strategies; see memory/project_survivorship_bias.md. This is common to most retail backtest tooling but should be acknowledged explicitly.

The user-driven backtests run in /lab use the same engine and inherit the same caveats. The numbers shown for any user-run backtest are illustrative of what the strategy would have done on the chosen tickers and dates in our corpus — not what a live account would realize today.

Forward-looking statements

Any forward-looking statements made on Schmatz — predicted shock attributions, projected mean-reversion outcomes, probability estimates, or strategy expectations — are inherently uncertain. They are produced by statistical models and large-language-model analysis applied to historical data and current filings, and they are subject to error. They should not be relied upon as predictions of future events or returns.

No fiduciary relationship

Your use of Schmatz does not create a fiduciary, advisory, or agency relationship between you and Schmatz, the operator, or any contributor. We owe you no duty of care with respect to your investment decisions. You are not our client.

No personalized recommendations

Schmatz publishes research impersonally — the same backtest engine, the same shock dossiers, the same Ask responses are available to every reader. Nothing on Schmatz takes into account your age, income, existing portfolio, tax situation, risk tolerance, or investment objectives. Schmatz never tells "you" personally to buy or sell any security. Any wording you encounter that could read like such a recommendation is editorial framing, not advice.

Allowed (and what Schmatz does): "Here is a backtested mean-reversion model that historically beat SPY in the breadth-gated regime. The model's most recent simulated trades are X, Y, Z."

Not allowed (and what Schmatz does not do): "Based on your portfolio and risk profile, you should buy Stock X tomorrow."

Publisher's exemption

Schmatz operates as a financial publication, not as an investment adviser. We rely on the publisher's exemption from registration as an investment adviser set forth at 15 U.S.C. § 80b-2(a)(11)(D) of the Investment Advisers Act of 1940 and the analysis in Lowe v. SEC, 472 U.S. 181 (1985). To qualify for this exemption, a publication must be:

• Bona fide — a genuine publication, not a sham device to evade registration. The Schmatz daily brief and case files are published on a recurring schedule as editorial research, not as gates to specific securities transactions.
• Of general and regular circulation — the same content is distributed to every Schmatz reader on a recurring cadence. Schmatz operates this way today and will continue to do so as the product matures.
• Non-personalized — Schmatz does not provide individualized advice. The brief discusses the same set of model-flagged opportunities for all readers. We do not tailor recommendations to any reader's personal financial situation, portfolio composition, risk tolerance, tax situation, or investment objectives.
• Disinterested — Schmatz does not use the publication to promote securities held by the operator or by affiliated parties. The simulated paper portfolio discussed in Schmatz dispatches is publicly identifiable and not used as a vehicle for personal trading by the operator ahead of publication.

For the avoidance of doubt: this exemption analysis is operator-asserted. Schmatz has not requested or received a no-action letter from the Securities and Exchange Commission on this specific publication. Readers seeking definitive advice on the regulatory status of any third-party publication should consult an attorney.

Data sources and licensing

Schmatz’s customer-facing surface is built on public-domain and freely-redistributable data sources. Subscribers do not need to bring their own API keys.

• SEC EDGAR (public domain) — corporate filings: 8-K, 10-K, 10-Q, Form 4 insider transactions, 13D activist filings, and XBRL-tagged financial statements used to compute fundamentals.
• Federal Reserve FRED (public, free for any use) — macroeconomic indicators (VIX, ^TNX, DXY, breadth proxies, treasury yields).
• Freely-redistributable price archives — historical US equity OHLCV bars used to power charts, backtests, shock detection, and the Echoes pattern matcher.

For internal research only (never displayed directly to subscribers), the operator separately maintains paid-tier subscriptions to third-party market-data providers used as a gold-standard validation reference against the public-data pipeline. See docs/SCALING_PLAYBOOK.md in the project repository for the architecture details.

What Schmatz publishes: our OWN derived research output — backtest summary statistics, classified shock events, AI-composed cross-source attribution narratives, and editorial commentary. The methodology is reproducible by subscribers running our open-source code against the same public-domain data sources we use.

What Schmatz does NOT publish or sell: downloadable bulk exports of any underlying historical price archive, bulk-export of the fundamentals corpus, or any general-purpose “query the database” endpoint. Subscribers consume derived analytics; we are an analytics publisher, not a data reseller.

Jurisdictional considerations

Schmatz is operated from the United States and the content is targeted at sophisticated US-based readers. Nothing on Schmatz should be construed as soliciting business in jurisdictions where the operator is not qualified to do so, and the content may not be appropriate for, or available in, all jurisdictions.

Limitation of liability

To the maximum extent permitted by applicable law, the operator of Schmatz disclaims all liability for losses or damages — direct, indirect, incidental, consequential, or punitive — arising from your use of, or reliance on, any content published here. You assume all risk associated with any decisions you make in connection with Schmatz output.

Early-access notice

Schmatz is currently in an early-access release with free self-signup. Features, content, methodology, and data presentation may change without notice. This disclaimer is itself subject to revision; you will find the date of last revision near the top of this page.

Contact

Questions about this disclaimer or about any specific item of content should be directed to the operator through the beta channel by which you received your invitation.