PROJECT III — FROM INCEPTION TO THE FINAL TEST MIND BEND THEORY | 20/20 OUT-OF-SAMPLE SESSIONS CAPTURED What began as a question about stock-market mechanics has become one of Mind Bend Theory's most extensive forward-testing experiments. Can stock-borrowing fees and share availability reveal a measurable relationship with future price movement and can that relationship survive independent testing? WHERE IT ALL BEGAN On September 11, 2026, Project III began its documented forward-capture experiment using Sound Hound AI ($SOUN). The objective was simple: capture borrowing conditions before the market outcome, preserve the original evidence, and determine whether those measurements had any meaningful relationship with subsequent stock-price changes. We weren't interested in explaining yesterday's price using information discovered afterward. We wanted evidence collected before the outcome existed. PHASE I — BUILDING THE EVIDENCE We began collecting Brokers securities-lending data, including: Shares available to borrow. Stock-borrowing fee rates. Rebate rates. Original provider timestamps. Complete source files preserved with SHA-256 verification. The first six documented sessions established our forward-capture process. We continued collecting observations, improving the evidence pipeline, and testing whether the records could be independently reconstructed. PHASE II — THE FIRST 20 SESSIONS Our exploratory research examined whether changes in borrowing fees correlated with subsequent 60-minute stock returns. The initial results were intriguing: Pearson correlation: +0.5533 (p = 0.0140) Spearman correlation: +0.5070 (p = 0.0267) These calculations used 19 paired observations from the initial 20-session sequence. They suggested a positive statistical relationship worth investigating. But there was a problem: a relationship discovered in historical observations might disappear when tested against completely new data. We needed independent replication. PHASE III — THE FROZEN OOS EXPERIMENT On October 14, 2026, we began a new 20-session out-of-sample test. The measurement rules were frozen in advance. Every qualifying session would compare the change in SOUN's borrowing fee against its subsequent 60-minute SIP price return, measured using precisely defined minute-open prices. No changing the formula after seeing results. No selecting only favorable trading days. No manufacturing a stronger correlation. On November 10, we reached 20/20 scheduled OOS captures. The final SIP price observation remains pending, along with outstanding evidence and timing-eligibility reviews. WHAT HAVE WE OBSERVED? The latest sessions delivered some of the most dramatic borrowing changes in the experiment. November 6: Borrowing fees surged 185%, available shares dropped 47%, and SOUN subsequently declined 2.05% during our frozen measurement window. November 9: Borrowing fees fell 64.8%, available shares increased 55.6%, and SOUN subsequently gained 2.56%. These two observations moved opposite to the direction suggested by our original positive correlation. That doesn't settle the experiment. It demonstrates why independent testing matters. PHASE IV — WHAT HAPPENS AFTER THE FINAL SIP? Once the twentieth SIP outcome becomes available, we begin the final scientific audit. 1. Reconstruct the complete experiment. We'll organize the original September research, the exploratory 20-session dataset, and the independent OOS observations into a documented research record. 2. Verify every piece of evidence. Original borrowing files, SIP price data, SHA-256 hashes, timestamps, eligibility rules, and calculated results must reconcile. If an observation fails the original frozen requirements, we will document it and determine whether additional forward sessions are required. 3. Run the frozen statistical tests. We'll calculate Pearson and Spearman correlations across the qualifying OOS sample and compare them with the original exploratory results. 4. Determine whether the hypothesis survived. Did the positive relationship replicate? Did it weaken, disappear, or reverse? Was it driven by a handful of extreme observations? We will report the actual results, including uncertainty and limitations, rather than forcing a favorable conclusion. 5. Publish the research. Our objective is a reproducible report explaining the methodology, evidence, statistical findings, failures, and what the experiment can and cannot establish. PHASE V — THE REAL MISSION SOUN was never intended to be the final destination. It was our controlled research environment. The longer-term objective is to determine whether stock-borrowing data can become a useful, independently validated market-structure measurement framework. If the $SOUN findings justify additional research, we intend to develop separately frozen experiments involving GameStop ($GME) and AMC Entertainment ($AMC). We would test whether the relationships observed in SOUN generalize to other stocks, different market conditions, and eventually broader market datasets. Only repeated, independent evidence could justify considering a practical analytical signal. A trading strategy would require additional validation, including transaction costs, execution constraints, and prospective performance testing. THE END GOAL Build a transparent, evidence-driven research system capable of distinguishing genuine market relationships from statistical coincidences. Not another prediction engine built on hype. A framework where every observation is preserved, every calculation can be challenged, and every claimed signal must prove itself against new data. 20/20 CAPTURED. THE FINAL SIP IS PENDING. THE VERDICT IS NEXT. $SOUN IS THE SANDBOX. $GME AND $AMC ARE THE MISSION. THE MARKET IS THE END GAME. MIND BEND THEORY | PROJECT III FROZEN BEFORE THE OUTCOME. THE SIGNAL HAS TO EARN ITS EXISTENCE. #ProjectIII #MindBendTheory #SOUN #GME #AMC #MarketStructure #QuantitativeResearch