<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:content="http://purl.org/rss/1.0/modules/content/"><channel><title>Craft on The Public Ledger</title><link>https://blog.jbledger.com/categories/craft/</link><description>Recent content in Craft on The Public Ledger</description><generator>Hugo</generator><language>en-us</language><lastBuildDate>Sat, 12 Sep 2026 12:30:00 +0000</lastBuildDate><atom:link href="https://blog.jbledger.com/categories/craft/index.xml" rel="self" type="application/rss+xml"/><item><title>An Honest Backtest: Projecting a Fiscal Year-End Two Ways</title><link>https://blog.jbledger.com/posts/an-honest-backtest-projecting-a-fiscal-year-end-two-ways/</link><pubDate>Sat, 12 Sep 2026 12:30:00 +0000</pubDate><guid>https://blog.jbledger.com/posts/an-honest-backtest-projecting-a-fiscal-year-end-two-ways/</guid><description>Linear continuation vs. damped Holt-Winters, walked forward over 36 held-out periods. The winner&amp;rsquo;s margin is thin enough that a library convention flips it — so the page shows both methods and says so out loud.</description></item></channel></rss>