The Difference Between a Signal and a Forecast

A signal does not owe you a prediction

A forecast tells a story about what should happen next. A signal measures what the current evidence supports.

Confusing the two is a reliable way to get frustrated with the dashboard. A signal can reduce exposure before a decline, add exposure before the data improves, or look wrong for weeks while the market keeps moving the other way.

The point is not to know the next tick. The point is to keep exposure matched to the quality of the evidence.

Howard Marks makes this kind of distinction in The Most Important Thing: the future is uncertain, but risk can still be judged. The dashboard is trying to judge risk, not eliminate uncertainty.

Forecast Tries to predict the next market move.

Signal Measures whether current evidence supports exposure.

Forecast error Can become an argument with the market.

Signal review Asks whether the inputs were interpreted consistently.

Why this matters after a wrong-looking move

Every risk process has moments that look wrong.

If the dashboard cuts exposure and the market keeps rising, the immediate temptation is to call the signal a failure. Sometimes that will be fair. Sometimes the evidence weakened and the market simply kept climbing on narrower support.

The review question should be more precise: did the signal follow the rules, and were the inputs actually deteriorating? If yes, the dashboard did its job even if the next week was annoying.

This is why the earlier post on reducing risk while the market is still going up matters. Risk management often looks early before it looks useful.

Rules make the signal reviewable

A discretionary forecast is hard to audit. A rule-driven signal is easier.

The dashboard takes inputs, classifies regimes, reads VAMS states, and adjusts the model allocation. A subscriber can disagree with the model, but the process is visible enough to review.

That is different from a market call. The dashboard does not need to say, “The S&P 500 will fall next month.” It can say, “The evidence no longer supports full risk exposure.”

1. Market domains update Participation, credit, rates, funding, and broad-dollar evidence change.
2. Rules classify evidence The dashboard converts the four domains into a regime and each sleeve into a VAMS state.
3. Risk budget adjusts The top-down target and sleeve multipliers determine exposure.
4. Subscriber maps it The signal is adapted to personal constraints.

GRID, the separate liquidity dashboard, and Gavekal still matter. They explain the backdrop and provide cross-checks, but they do not directly alter the KISS allocation.

Forecasts still have a place

None of this means forecasts are useless. Investors need expectations. Valuation, growth, inflation, policy, and liquidity all require judgment.

The problem starts when a forecast becomes the whole process. A strong view can be early. It can be right for the wrong reason. It can be right but sized too aggressively.

Signals add discipline. They force the portfolio to respond to evidence instead of confidence.

Judge the rule on the rule’s job

Judge the dashboard as a signal process.

Do not ask whether it predicted the next move. Ask whether it adjusted exposure consistently when the evidence changed. That standard is less exciting and much more honest.