<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
  <channel>
    <title>BTC Signal: Research</title>
    <link>https://btcsignal.io/research.html</link>
    <description>Bitcoin research from thirteen indicators, ML-weighted into a composite. Methodology, backtests, and the parts that do not work, published with the data.</description>
    <language>en</language>
    <lastBuildDate>Wed, 08 Jul 2026 08:52:55 +0200</lastBuildDate>
    <atom:link href="https://btcsignal.io/feed.xml" rel="self" type="application/rss+xml"/>
    <item>
      <title>The Dashboard Takes the Calendar Test</title>
      <link>https://btcsignal.io/articles/calendar-test.html</link>
      <guid isPermaLink="true">https://btcsignal.io/articles/calendar-test.html</guid>
      <pubDate>Wed, 08 Jul 2026 08:52:55 +0200</pubDate>
      <description>One indicator failed, seven got stronger. The full results of turning the oil regime-marker test on this website's own dashboard, and the honest answer to 'why keep the one that failed?'</description>
    </item>
    <item>
      <title>The Oil Anomaly Is a Calendar</title>
      <link>https://btcsignal.io/articles/oil-anomaly.html</link>
      <guid isPermaLink="true">https://btcsignal.io/articles/oil-anomaly.html</guid>
      <pubDate>Tue, 07 Jul 2026 23:17:21 +0200</pubDate>
      <description>The strangest feature in this website's ML rankings, resolved: XGBoost used the oil price to memorize which macro era a row belonged to. A new caveat for every SHAP ranking, here and elsewhere.</description>
    </item>
    <item>
      <title>Gradual, So Far</title>
      <link>https://btcsignal.io/articles/gradual-so-far.html</link>
      <guid isPermaLink="true">https://btcsignal.io/articles/gradual-so-far.html</guid>
      <pubDate>Tue, 07 Jul 2026 22:47:58 +0200</pubDate>
      <description>Six months of Fed balance-sheet growth under a technical name: $189 billion since December, against $3 trillion in 2020. The gradual phase of 'gradual, then big' is on script. The big one has not arrived, and bitcoin fell anyway.</description>
    </item>
    <item>
      <title>Then It Sells the Underlying</title>
      <link>https://btcsignal.io/articles/sells-the-underlying.html</link>
      <guid isPermaLink="true">https://btcsignal.io/articles/sells-the-underlying.html</guid>
      <pubDate>Tue, 07 Jul 2026 22:12:35 +0200</pubDate>
      <description>Strategy's largest bitcoin sale ever was not profit-taking. It was principal, sold at a loss, to pay a coupon the company had just raised. The sequence was published here in June. What remains is arithmetic.</description>
    </item>
    <item>
      <title>The Dry Powder Myth</title>
      <link>https://btcsignal.io/articles/dry-powder.html</link>
      <guid isPermaLink="true">https://btcsignal.io/articles/dry-powder.html</guid>
      <pubDate>Tue, 07 Jul 2026 21:50:10 +0200</pubDate>
      <description>Does rising stablecoin liquidity predict bitcoin? Tested as a level, a flow, and a valuation ratio. On all three it fails the bar that any real indicator has to clear.</description>
    </item>
    <item>
      <title>Bitcoin Has No Power Law</title>
      <link>https://btcsignal.io/articles/power-law-verdict.html</link>
      <guid isPermaLink="true">https://btcsignal.io/articles/power-law-verdict.html</guid>
      <pubDate>Sun, 21 Jun 2026 10:40:35 +0200</pubDate>
      <description>A rigorous academic test confirmed this website's power-law critique, and corrected one claim it advanced without finishing the test. On confirmation, correction, and the discipline of being accurate.</description>
    </item>
    <item>
      <title>A Coupon Is Not a Cash Flow</title>
      <link>https://btcsignal.io/articles/bitcoin-yield.html</link>
      <guid isPermaLink="true">https://btcsignal.io/articles/bitcoin-yield.html</guid>
      <pubDate>Sat, 20 Jun 2026 22:11:50 +0200</pubDate>
      <description>Bitcoin has no cash flows, so every 'yield' on it is a risk premium in disguise. Where the yield comes from, and why the maturity story is half right and half dangerous.</description>
    </item>
    <item>
      <title>A Foundation Model Can't Beat a Coin Flip on Bitcoin</title>
      <link>https://btcsignal.io/articles/foundation-model.html</link>
      <guid isPermaLink="true">https://btcsignal.io/articles/foundation-model.html</guid>
      <pubDate>Fri, 19 Jun 2026 15:14:58 +0200</pubDate>
      <description>Can AI predict bitcoin? A state-of-the-art tabular foundation model, run on the same walk-forward protocol, landed at a coin flip on daily direction. Why the limit is the market, not the model.</description>
    </item>
    <item>
      <title>Three Reflexive Machines</title>
      <link>https://btcsignal.io/articles/reflexive-machines.html</link>
      <guid isPermaLink="true">https://btcsignal.io/articles/reflexive-machines.html</guid>
      <pubDate>Fri, 19 Jun 2026 10:05:25 +0200</pubDate>
      <description>Strategy, STRC, and Terra-Luna compared at the level of mechanism. Why a reflexive, leveraged bitcoin structure with a discretionary brake is not a Terra-style death spiral.</description>
    </item>
    <item>
      <title>The Power Law, Taken Seriously</title>
      <link>https://btcsignal.io/articles/power-law-taken-seriously.html</link>
      <guid isPermaLink="true">https://btcsignal.io/articles/power-law-taken-seriously.html</guid>
      <pubDate>Fri, 19 Jun 2026 09:49:53 +0200</pubDate>
      <description>The strongest case for the Bitcoin Power Law is a mechanism, not a curve fit. It earns real credit, and it rests on one fitted exponent it declines to test.</description>
    </item>
    <item>
      <title>Gradual, Then Big</title>
      <link>https://btcsignal.io/articles/gradual-then-big.html</link>
      <guid isPermaLink="true">https://btcsignal.io/articles/gradual-then-big.html</guid>
      <pubDate>Sun, 07 Jun 2026 10:15:52 +0200</pubDate>
      <description>When debt service crowds out the budget, the central bank stops setting the terms. What that gravity well does to non-sovereign assets.</description>
    </item>
    <item>
      <title>Liquidity, Already Counted</title>
      <link>https://btcsignal.io/articles/liquidity-counted.html</link>
      <guid isPermaLink="true">https://btcsignal.io/articles/liquidity-counted.html</guid>
      <pubDate>Wed, 03 Jun 2026 11:38:32 +0200</pubDate>
      <description>Why liquidity barely registers in the model when the macro consensus says it should rank #1.</description>
    </item>
    <item>
      <title>The Tech-Beta Mirage</title>
      <link>https://btcsignal.io/articles/tech-beta-mirage.html</link>
      <guid isPermaLink="true">https://btcsignal.io/articles/tech-beta-mirage.html</guid>
      <pubDate>Tue, 02 Jun 2026 14:43:16 +0200</pubDate>
      <description>The machine looked for what predicts bitcoin. The stock market wasn't there.</description>
    </item>
    <item>
      <title>It's a Power Trendline, Not a Power Law</title>
      <link>https://btcsignal.io/articles/power-trendline.html</link>
      <guid isPermaLink="true">https://btcsignal.io/articles/power-trendline.html</guid>
      <pubDate>Tue, 26 May 2026 15:08:41 +0200</pubDate>
      <description>Bitcoin's price-vs-time chart is a power-function regression. The real statistical Power Laws are in the volatility and the return tails.</description>
    </item>
    <item>
      <title>Sticky States, Brief Transitions</title>
      <link>https://btcsignal.io/articles/regime-transitions.html</link>
      <guid isPermaLink="true">https://btcsignal.io/articles/regime-transitions.html</guid>
      <pubDate>Tue, 26 May 2026 15:08:35 +0200</pubDate>
      <description>A Markov map of the composite score, with honest caveats about what 16 years of data can and cannot say, and a visible retraction where regeneration killed a finding.</description>
    </item>
    <item>
      <title>Update on the Dam</title>
      <link>https://btcsignal.io/articles/dam-about-to-break.html</link>
      <guid isPermaLink="true">https://btcsignal.io/articles/dam-about-to-break.html</guid>
      <pubDate>Tue, 26 May 2026 15:08:31 +0200</pubDate>
      <description>Jane Street cut its Bitcoin ETF exposure by 71% in Q1 2026. The price suppression mechanism is unwinding.</description>
    </item>
    <item>
      <title>What If? Forward Simulation</title>
      <link>https://btcsignal.io/articles/forward-simulation.html</link>
      <guid isPermaLink="true">https://btcsignal.io/articles/forward-simulation.html</guid>
      <pubDate>Tue, 07 Apr 2026 17:58:54 +0200</pubDate>
      <description>6 simulated Bitcoin scenarios. 3 DCA strategies. 3 years. Honest results.</description>
    </item>
    <item>
      <title>Does the Signal Actually Work?</title>
      <link>https://btcsignal.io/articles/signal-dca.html</link>
      <guid isPermaLink="true">https://btcsignal.io/articles/signal-dca.html</guid>
      <pubDate>Sun, 05 Apr 2026 22:28:50 +0200</pubDate>
      <description>Three DCA strategies tested across 8 years. What works, what doesn't, and why.</description>
    </item>
    <item>
      <title>The Illusion of Diversification</title>
      <link>https://btcsignal.io/articles/correlation-analysis.html</link>
      <guid isPermaLink="true">https://btcsignal.io/articles/correlation-analysis.html</guid>
      <pubDate>Sat, 04 Apr 2026 14:23:36 +0200</pubDate>
      <description>Why 13 indicators are really 7 dimensions. Correlation analysis and what changed in response.</description>
    </item>
    <item>
      <title>How the Composite Score Works</title>
      <link>https://btcsignal.io/articles/composite-score.html</link>
      <guid isPermaLink="true">https://btcsignal.io/articles/composite-score.html</guid>
      <pubDate>Sat, 04 Apr 2026 12:29:04 +0200</pubDate>
      <description>13 indicators, ML-weighted, combined into a single buy/sell score.</description>
    </item>
    <item>
      <title>SHAP Analysis: Which Indicators Actually Matter</title>
      <link>https://btcsignal.io/articles/shap-analysis.html</link>
      <guid isPermaLink="true">https://btcsignal.io/articles/shap-analysis.html</guid>
      <pubDate>Sat, 04 Apr 2026 12:29:04 +0200</pubDate>
      <description>ML feature importance across three Bitcoin cycles. The prediction model failed. The rankings were more interesting, including the inconvenient ones.</description>
    </item>
    <item>
      <title>The Dam and the River</title>
      <link>https://btcsignal.io/articles/etf-plumbing.html</link>
      <guid isPermaLink="true">https://btcsignal.io/articles/etf-plumbing.html</guid>
      <pubDate>Fri, 03 Apr 2026 17:11:24 +0200</pubDate>
      <description>Why Bitcoin's ETF plumbing is delaying an inevitable repricing.</description>
    </item>
  </channel>
</rss>
