---
title: "What Changed in KISS 1.9.0, and Why the History Was Rebuilt"
canonical: "https://themacrodashboard.com/blog/what-changed-in-kiss-1-9-0/"
pubDate: "2026-06-01T00:00:00.000Z"
updatedDate: "2026-06-01T00:00:00.000Z"
author: The Macro Dashboard
description: "A reader-facing methodology changelog covering corrected inputs, uncertainty handling, implementation-aware performance, and the full historical rebuild."
categories: [Field Notes]
---

KISS 1.9.0 changed enough of the machinery that the old history had to go. Keeping the old rows would have made the chart smoother to maintain and harder to defend.

The release changed input measurement, uncertain-regime handling, and portfolio simulation. Every active history row was rebuilt under the new rules.

## The main calculation changes

Investable ETFs now use adjusted total-return series where distributions matter. Indexes, yields, spreads, volatility indexes, currencies, commodity-level proxies, bitcoin, and individual stocks retain raw economic levels where that is the intended measurement.

Rates, credit spreads, breakevens, funding spreads, and the yield curve now use arithmetic basis-point changes rather than generic percentage changes. A 25-basis-point move should remain 25 basis points whether the starting level is positive, near zero, or negative.

The U.S. reserve-liquidity calculation now normalizes Federal Reserve assets, Treasury General Account balances, and overnight reverse repo into the same dollar units before subtraction.

## The regime became more honest about uncertainty

Market evidence is assigned to four independent domains. Each domain receives 25% of the combined score, so a large collection of related equity ETFs cannot dominate credit, rates, or funding evidence merely because more tickers are available.

A tie, a winner margin below two points, coverage below 70%, or stale required data now produces Mixed. The model holds the last confirmed top-down target instead of selecting a regime through object order or inventing a halfway portfolio.

The public output also reports methodology version, winner margin, coverage, domain agreement, confidence, and ambiguity reasons.

## Context and allocation were separated

GRID, U.S. reserve liquidity, fast financial conditions, quarterly global liquidity, NFCI, and Gavekal remain useful. They do not change KISS allocation directly.

The BIS [global liquidity indicators](https://www.bis.org/statistics/dataportal/gli.htm) are quarterly measures of foreign-currency credit. That cadence makes them valuable context and a poor daily trading trigger. The same job separation applies to revised macro history.

FRED's [ALFRED documentation](https://fred.stlouisfed.org/docs/api/fred/alfred.html) explains why vintage data matter. Today's historical value may differ from what investors knew at the time. Until the dashboard stores point-in-time macro vintages, GRID stays outside allocation history. The FRED guide to [real-time periods](https://fred.stlouisfed.org/docs/api/fred/realtime_period.html) provides the technical mechanism needed for a future point-in-time implementation.

## VAMS was tested and retained

The existing VAMS formula was compared with a simpler trend candidate across deterministic multi-regime fixtures, walk-forward windows, parameter neighborhoods, turnover, and drawdown.

The simpler candidate traded less but produced worse drawdowns and a lower ending value in the declared test. The current formula was retained. RSI remains display-only, and the dependence among the five trend votes is now documented rather than ignored.

The decision was not based on maximizing the displayed 2025-present result. AQR's [trend-following research](https://www.aqr.com/Insights/Research/Journal-Article/A-Century-of-Evidence-on-Trend-Following-Investing) shows why evidence across markets and periods deserves more weight than one attractive backtest.

## Performance became implementation aware

The display now says archived-signal model simulation. It uses adjusted total returns, next-session execution, transaction costs, slippage, and a declared cash-return proxy. Taxes remain excluded.

The CFA Institute's [GIPS standards](https://www.gipsstandards.org/) are built around clear performance definitions and presentation. The dashboard does not claim compliance, but it follows the same basic instinct: simulated output should not be described as audited live performance.

## Why every history row had to be recalculated

Regime classification affects top-down targets. VAMS state memory affects sleeve multipliers. Two-close confirmation affects rebalance dates. Execution timing affects portfolio returns. Changing any one of those while preserving old rows would create a mixed-method series.

The rebuild therefore started from warmup data and recalculated each date in chronological order. It regenerated regime history, VAMS memory, allocations, rebalance events, performance, compact endpoints, snapshots, and generated assets under one methodology identifier.

## Version first, performance second

Check the methodology version before comparing dashboard results. KISS 1.9.0 history belongs with KISS 1.9.0 performance.

Previously published figures may still be useful as an audit record. They are not directly comparable without accounting for the rules that produced them.