--- title: "Portfolio checks: 98 fixed rules, one briefing" description: "98 fixed checks across portfolio, costs, German taxes and behavior. A weekly briefing and an AI chat over real data. Context, not prediction." canonical: "https://investboard.de/en/produkt/insights" language: "en" modified: "2026-07-26" author: "Investboard" --- # 98 checks. One briefing. No forecasts. Investboard reads your wealth against 98 fixed rules: concentration, costs, German taxes, behavior. A weekly briefing sums up what changed. And an AI chat answers questions over your real data: evidenced, bounded, in the EU. ## Insights are context with evidence. Investboard Insights check your wealth against 98 fixed rules: from concentration risk through costs and German tax deadlines to your own behavior. Anomalies are explained, grounded in the triggering numbers and framed against your mandate. The AI phrases; the computing happens first. - 98 deterministic rules in 14 categories, each with a named evidence base. - A weekly briefing: what changed and whether it matters for your plan. - AI chat with nine tools over real portfolio data, behind guardrails. ## 98 rules you can read. The rule engine works deterministically: 98 fixed checks in 14 categories, ranked by relevance and de-duplicated. No signal comes from a model's whim; each comes from a rule with a threshold and an evidence base. ### 14 categories From concentration and costs through dividends, earnings, fundamentals and net worth to German taxes and your own behavior: every rule checks one concrete, named condition. ### German tax depth Vorabpauschale, the saver's allowance, the one-year speculation period, legacy holdings, loss pots: the rule engine knows the deadlines and allowances that actually apply to German brokerage accounts. ### Not a language model Deliberately not a language model: signals are computed, not phrased. What the rules cannot evidence does not appear. ## Every morning: what changed and whether it counts. Five deterministic detectors per symbol spot anomalies. A weekly briefing puts the picture into words: calm, fully formed, without forecasts. ### Five detectors Per symbol, five deterministic detectors check for price outliers, volume spikes, news-sentiment shifts, earnings surprises and cross-asset divergence. ### One briefing, fully phrased An EU-resident model condenses rules, detectors and dates into a weekly briefing behind output guardrails. It frames, it does not predict. ### Dates in view Upcoming earnings dates of held positions flow into the briefing before they surprise. ## Answers over your real data. The AI chat answers questions about your actual portfolio: EU-resident, behind guardrails, with tool access instead of guesswork. It explains and frames, it does not recommend. ### Nine tools Portfolio summary, individual holdings, net worth, prices, allocation, drift, tax summary, market overview and active signals: the chat reads your data rather than answering in general. ### Explain where the question is Metrics, charts and signals can be explained right in context; follow-up questions keep the thread. ## Every number precomputed. Every answer logged. Trust here is not a footnote but architecture: the AI may phrase, but it may not compute, guess or advise. ### No invented figures Every number in an AI answer must have been computed from your data first; outputs are checked against that. If an answer fails the check, it is blocked or replaced by the deterministic version. ### EU-resident, logged The model runs in the EU, conversations stay in the EU, and every AI answer is logged: traceable, even months later. ### Advice boundaries Investment advice, order suggestions and forecasts are out of scope and actively constrained. Insights frame; deciding stays with you. ## Frequently asked questions ### Are Investboard Insights AI forecasts? No. Insights are context based on available data. 98 fixed rules and five detectors flag anomalies; they do not predict prices and do not replace investment advice. ### How do I know why a signal appears? Every signal comes from a named rule and is grounded in the triggering numbers from your data. What cannot be evidenced does not appear. ### Why does Investboard use fixed rules instead of open-ended AI? Fixed rules are auditable: same data, same signal, traceable threshold. The AI comes only afterwards and phrases what the rules have found. ### Can the AI chat give investment recommendations? No. The chat explains data, metrics and relationships in the portfolio. Advice, order suggestions and invented figures are constrained by guardrails.