<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"><channel><title>ArkNode Labs</title><description>Quant systems, AI agents, and a GPU homelab - documented honestly, including what failed.</description><link>https://arknodelabs.com/</link><language>en-us</language><item><title>GPU passthrough into LXC, without the pain</title><link>https://arknodelabs.com/experiments/gpu-passthrough-lxc/</link><guid isPermaLink="true">https://arknodelabs.com/experiments/gpu-passthrough-lxc/</guid><description>Four identical RTX 5060 Ti cards, one Proxmox container, and a PCIe lane map that quietly decides which card is allowed to talk to which. The passthrough was the easy part.</description><pubDate>Tue, 11 Aug 2026 00:00:00 GMT</pubDate><category>homelab</category></item><item><title>The look-ahead bug that made our backtest look worse</title><link>https://arknodelabs.com/experiments/backtest-lookahead-bug/</link><guid isPermaLink="true">https://arknodelabs.com/experiments/backtest-lookahead-bug/</guid><description>Our Chandelier exit peeked one bar ahead, cutting winners short instead of inflating the result. A fixed backtest is not a found edge: ETH 4h went from -16.2% to +185.0%.</description><pubDate>Fri, 03 Jul 2026 00:00:00 GMT</pubDate><category>trading</category></item><item><title>Our LLM Polymarket bot was just reading the market price back to us</title><link>https://arknodelabs.com/experiments/polymarket-postmortem/</link><guid isPermaLink="true">https://arknodelabs.com/experiments/polymarket-postmortem/</guid><description>Aggregate Brier score 0.12. And it lost to the market in every single edge bucket. A postmortem of an LLM forecaster that turned out to be an expensive mirror.</description><pubDate>Sun, 07 Jun 2026 00:00:00 GMT</pubDate><category>trading</category></item><item><title>Everything I got wrong planning a $5.5k AI homelab</title><link>https://arknodelabs.com/experiments/ai-homelab-mistakes/</link><guid isPermaLink="true">https://arknodelabs.com/experiments/ai-homelab-mistakes/</guid><description>Three planning mistakes, one invisible GPU, and seven months of running a 64GB-VRAM box that ended up doing a completely different job than the one I built it for.</description><pubDate>Mon, 01 Dec 2025 00:00:00 GMT</pubDate><category>homelab</category></item></channel></rss>