This section is about risk: how much the portfolio can lose in one bad day, what a known crisis does to it, and how to tell what state the market is in right now. Four templates answer these questions with a check on past data, not promises.
Every template is a live walkthrough on demo data: open it in the Templates section of Analytics Agent and press “Use this template” to run it on your own data.
VaR Backtest (How Much Can a Day Take?)
Shows: how much the portfolio can lose in a day when things go badly — computed three ways — and, most importantly, a check against a decade of past days: how many times actual losses exceeded the promised figure, and whether the failures come in bursts.
You get: you learn whether the number — usually shown without a check — can be trusted, and how much worse the worst day was than the calculated one.
Stress Test (What 2008 Does to This Book)
Shows: the portfolio revalued under five known shocks — 2008, covid, the 2022 rate hikes, and others — with a severity verdict and two recovery counters: how much the market regrew on its own, and how much of the hole ordinary returns would have to close.
You get: you can see which exact position sinks you in a named crisis, and what a cushion of bonds and cash would cost — along with the price of that insurance.
Volatility Forecast (How Rough Is the Market Now?)
Shows: how rough the next two weeks will be according to the chosen model, with a range built on the model’s own past errors, and a comparison of several models — including a simple one that is hard to beat.
You get: position size becomes a budget divided by the swing range; knowing it with an honest interval, you are no longer surprised by the week that breaks every calculation.
Market Regimes (Calm or Crisis — and How Sure?)
Shows: the market’s state right now — calm, tense, or crisis — with a probability, the history of these states, their typical durations, and a measured reaction lag: how many days pass before the switch notices a crash.
You get: the phrase “it is a crisis now” becomes measurable — with an honestly shown lag, not a pretty picture drawn after the fact.