mykola-quant/seven-hour-profile-falsification

This project is a statistical study, not a piece of software you install and run for daily use. It is one entry in a public series of tests where the author writes down, in advance, exactly how a trading idea will be tested and what result would count as support or as failure, before looking at any data. This particular study tests a popular social media trading theory that says price movement in three recurring 7 hour trading sessions, tied to Asia, London, and New York market hours, falls into one of six repeating patterns, and that spotting which pattern is happening gives traders an edge during the New York session. The author tested this idea, unmodified from how it is usually described online, across four assets: Bitcoin, Solana, gold, and silver, using three different levels of strictness for defining each pattern. The pre-registered rule required a pattern to clear a statistically strict bar on most of these combinations to count as real. None of the six patterns cleared that bar on any asset. Two patterns showed a consistent direction across all three strictness levels for one asset each, but neither came close to being statistically meaningful, so the author reports them plainly without treating them as discoveries. The repository also notes that a large share of trading days, sometimes the majority, do not match any of the six patterns at all, and that silver behaved differently from the other three assets throughout, in a way that echoes an earlier study by the same author. An illustrative account simulation is included too, but the author is careful to explain that a few of those simulated accounts running out of money early is a side effect of using the same fixed trade size across very different assets, not new evidence against or for the idea. The project includes the full written methodology, Python scripts for collecting price data and running the statistical tests, and the raw results as CSV files. It is best suited to people interested in whether a specific trading theory holds up under a careful, honest test, rather than developers looking for a reusable tool. It is released under the MIT license.