👏 Stop paying for beta! 👏 Why We Entered the CrowdCent Hyperliquid Challenge Our security ranking tool is built on expected partial moments (lnkd.in/ek7zjRU). Clients have given it strong marks, but it's hard for anyone outside that group to verify. Now they can. The CrowdCent Hyperliquid Ranking challenge asks contestants to submit relative performance ranks on roughly 200 crypto securities, not code or model internals. We like that structure. Too many "quant contests" are IP grabs dressed up as competitions. Many of the entrants are familiar names from other data-science and crowdsourcing competitions, and the leaderboard shows why that matters: contestant ranks correlate heavily with the ensembled meta-model. Groupthink, fully quantified. Our entries are the exception. As of this week, #ovvo-nns holds three of the top four leaderboard slots, including #1, all with negative correlation to the meta-model. The next-highest finisher sits at +0.50 correlation to the crowd. We're on the other side of that number and still ranking ahead of it. It gets more interesting under the hood. Isolating the orthogonal portion of our 30-day rank signal produces a 100% hit rate across observed periods, versus 75% for the raw signal. The orthogonal component isn't just uncorrelated, it's the more predictive part of the model. Portfolio theory says uncorrelated return sources are valuable. Negatively correlated sources that still outperform are rarer, and worth more. If you run a long/short book or need to rank securities, stop replicating the crowd and paying for more beta. Add real alpha to the stack instead. 👉 ovvolabs.com