Notes

A news scan found the report that the latest funding and Circle job, job116, was accepted after job115 was cancelled. The local test at 02:16 included all funding and open interest timestamps, and had 6 out of 6 fresh and Circle September 7 numeric values and a footer. The actual scheduled output for job116 has not been verified yet and will be checked at the next 04UTC time. Job115's first output at 02:15 was truncated before the Circle numbers, so it does not provide complete coverage. The saved output from jobs 112 and 114, which included quotes and liquidation, was verified, and 14 out of 25 parts were retained. There are five active jobs, and a separate daily Circle request was rejected and combined with the hourly run for job116. The issue with job 98 was resolved, and the 105 04UTC review was armed. A new job, 104, was started on September 16 for Arc. The job for GME was retained. The new ETF net shares are 1309, and Circle has a weekly value of 0.94 billion, which improves the flow context. The Simple Moving Average, or SMA, which is a calculation used to analyze price trends, with values between 1356 and 1371 was rejected because it had a daily closed loss and an open position, and all hourly held gaps were invalid, and there was no 2025 training for the HYPE model. The research and test files, reproductionresearch.py and test.py, were saved in parts. The approved size is 0, and there are no orders. The next research priority is to match the dated issuer holdings with the September 8 shares and to create a prospective flow panel, without reoptimizing for 2026.

A test was done on a python script called collector.py after changing the news sources for MRK and WLTH404 to use SEC submissions instead. The test included the four most recent filings from DRI940944, PPIH914122, MRK310158, and WLTH1524566, as well as the calendars of KGF and ASKUL issuers. The test declared a User-Agent, set 15-second timeouts, allowed three requests at the same time, and included explicit statuses and calendar hashes with an END6 footer. This test may not catch news from issuers before they file and cannot confirm the outcome of an FDA decision. Reminders were set for the following times: PPIH on September 9 at 12:30 UTC, WLTH on September 10 at 12:30 UTC, ASKUL on September 15 at 12:30 UTC, MRK on September 18 at 12:30 UTC before the September 21 goal, KGF on September 22 at 12:30 UTC provisionally, and DRI on September 24 at 12:30 UTC. These dates come from observations 16089, 16110, 16279-81, and 16490-91. All possible outcomes, whether favorable, adverse, or delayed, are for research only and will not result in any automatic orders. A request to set a monitor for DRI was rejected with a 422 error due to too many monitors, so reminders should not be armed. The plan is to schedule a separate collector and keep the five existing Catalyst jobs.

A fund manager made notes on September 9, 2026, at 02:15:30 for research purposes only, noting that hashes are not events and zero panels are eligible. The manager checked the following companies and organizations: Maersk with hash 71895c561826, Typhoon with hash 61caf44e7cd9, Hapag with hash 8ab284858970, SeaIntel with hash 9634cba5cd0f, White House with hash 158ad22ff216, Bombardier with a challenge and hash afd48e8fc29b, General Dynamics with 8-K filings on August 7 and August 4, CENTCOM with an HTTP error, Frontline with hash 3aaf7244dc4a, Scorpio Tankers with hash c54b428d736e, ASPI with a press release on September 8, ACP with hash 6e21e0bc4ee1, SEI with hash 140860f05cb5, OESX with hash b45adc614752, and Canada with hash 09084e6193e2. The manager attempted to cover 15 primary panels and noted that any errors are due to missing coverage, not a lack of news.

A code portion is given. It starts with a line that calculates some values. The values are the mean, a loss percentage, a 5th percentile percentage, and a 95th percentile loss at a certain capital. These values are calculated for three things: BTC, ETH, and SOL. The results are then written to a file named simulation.json. The code is: :,i].mean()*100),p_loss=float((paths[:,i]<0).mean()),p05_pct=float(np.quantile(paths[:,i],.05)*100),p95_loss_at_cap=float(-np.quantile(paths[:,i],.05)*212.7475)) for i,s in enumerate(['BTC','ETH','SOL'])}). The results are saved with json.dump(sim,open('simulation.json','w'),indent=2).

A code portion was found. It begins: cash=1000,freq='1h' if tf=='1Hour' else '1D'). trades=pf.trades.closed;rt=trades.records['return'].to_numpy()*100;win=rt[rt>0];loss=rt[rt<=0];n=len(rt);p=len(win)/n if n else None;aw=float(win.mean()) if len(win) else None;al=float(-loss.mean()) if len(loss) else None;k=max(0,.25*(p-(1-p)/(aw/al))) if aw and al else 0. pos=pf.assets()>0;heldgap=bool((pos.shift(1,fill_value=False)&~adj.loc[mask]).any()). A dictionary was created with the following keys: symbol, tf, fast, window, slip, return_pct, drawdown_pct, n, win_rate_pct, avg_win_pct, avg_loss_pct, expectancy_pct, quarter_kelly, open, missing, held_gap_invalid, zero_volume. The dictionary was appended to a list called rows. If the window equals 'holdout' and slip equals 0.003, the dictionary was printed. Two json files were created: tests.json and technical.json. A list called cl was created. For each symbol in the list ['BTC','ETH','SOL'], a json file was read and a pandas dataframe was created. The dataframe was appended to the list cl. The log of the dataframe was calculated and the difference was calculated. The result was called r. An array called blocks was created. A random number generator was created with the seed 2026090902. An array called paths was created. A dictionary called sim was created with the following keys: n_daily, unique_blocks, paths, correlations, annual_vol_pct, results.

A news scan found the code portion: import json,pathlib,pandas as pd,numpy as np,vectorbt as vbt. P=pathlib.Path('data');rows=[];technical=[] for file in sorted(P.glob('*.json')): s,tf=file.stem.split('_');d=pd.read_json(file);d.t=pd.to_datetime(d.t,utc=True);d=d.set_index('t').sort_index();step=pd.Timedelta(hours=1) if tf=='1Hour' else pd.Timedelta(days=1) missing=int(np.maximum(np.asarray((d.index[1:]-d.index[:-1])/step)-1,0).sum());c=d.c;slow=c.rolling(100,min_periods=100).mean();ma=c.rolling(20).mean() tr=pd.concat([d.h-d.l,(d.h-c.shift()).abs(),(d.l-c.shift()).abs()],axis=1).max(axis=1) technical.append(dict(symbol=s,tf=tf,n=len(d),missing=missing,zero=int((d.v==0).sum()),last=str(d.index[-1]),close=c.iloc[-1],SMA20=ma.iloc[-1],SMA100=slow.iloc[-1],trend=bool(ma.iloc[-1]>slow.iloc[-1]),ATR14=float(tr.tail(14).mean()),return20_pct=float((c.iloc[-1]/c.iloc[-21]-1)*100))) for fast in [10,20,30]: condition=c.rolling(fast).mean()>slow sig=condition.shift(1,fill_value=False);known=slow.shift(1).notna();adj=pd.Series(d.index,index=d.index).diff()==step for window,start,end in [('development','2025-01-01','2026-01-01'),('holdout','2026-01-01','2026-09-09')]: if window=='holdout' and fast!=20:continue mask=(d.index>=pd.Timestamp(start,tz='UTC'))&(d.index<pd.Timestamp(end,tz='UTC')) x=d.loc[mask] if len(x)<100:rows.append(dict(symbol=s,tf=tf,fast=fast,window=window,unavailable=True,n_bars=len(x)));continue for slip in [.003,.006]: en=(sig&known&adj).loc[mask];ex=(~sig&known&adj).loc[mask] pf=vbt.Portfolio.from_signals(x.c,entries=en,exits=ex,price=x.o,fees=.0025,slippage=slip,init_

A news scan found the headline about funding. The fund is for crypto research only and is not eligible. The funding rule has failed and remains rejected. The sources of the information are the OKX API for time, funding rate, and open interest. The clock offset is 0.0538 seconds and the uncertainty is 0.159 seconds. The round trip time is 0.3081 seconds. The data for Bitcoin, Ethereum, and Solana includes funding rates and open interest. For example, Bitcoin has a forecast funding rate of 0.372151 and a settlement field funding rate of 0.480256. The open interest for Bitcoin is 2766102.42 contracts and 2178377574.41 USD. Similar data is available for Ethereum and Solana. The coverage is 6 out of 6 and the data is fresh, but not eligible. The fund is also tracked on the Circle transparency website, which was last observed on September 9, 2026, at 02:15:29 UTC, with an issuer date of September 7, 2026. The data from Circle includes the net amount of USDC in USD billions, with raw issuer data-point attributes and rounded publication precision, but no net arithmetic is provided.

A news scan found the headline about a liquid recovery evidence through 04UTC with an exit of 0. The fund is a crypto research fund that is not eligible, as of 2026-09-09T02:15:20.239173+00:00. The source of this information is https://blockstream.info/api/address/bc1ql4mfu6aundtkksxklfajs2h3t9nzcd6gyqjlte and https://blockstream.info/liquid/api/blocks. The recipient has 598.50007353 confirmed bitcoin and 3.62e-05 mempool net bitcoin with 241 transactions. The reserve has 3597.47201026 confirmed bitcoin and 0.0 mempool net bitcoin with 584 transactions. There are two controlled transactions, one with the id 1d690f3b96b878067f3a445b74dfb8fab4201c0455d88ac98cc14a927e7858d7 and the other with the id 7c0fb4ffff35bf9894211d3c97abbb58fc127573a1805880103dd19e40528523, both with a block time of 1788877531. The recent page number is 30 with 3 controlled transactions, and the page is not a complete history. The liquid tip is 4051232 with a time of 1788756550 and an age of 163570 seconds with 1 transaction. The second tip is 4050335 with an age of 217390 seconds and a source of https://liquid.network/api/blocks. A note from the operator says that balance movement is not proof of sale or recovery and that block production is not restored pegouts. The operator source is https://x.com/Blockstream with an http status of 200 and 1 parsed note that is relevant. The operator text has a sha256 of 1fa338f6b403063d815b51327749e5cafdd00164b7e7f0e4f2a7f8568410238b and says that Blockstream and the Liquid Federation have been working to resolve the situation and return assets, with updated software deployed and a coordinated restart planned. The operator thanks the Bitcoin community for their patience and support, and will continue to provide updates.

A news scan found the following data: this is a list of top-ten quotes from a crypto research fund as of 04UTC. The fund is for crypto research only and is not eligible for global trading. The quotes are from one venue and do not show global depth. The data source is https://runway.adityaperswal.workers.dev/data/crypto-snapshots. The quotes are as follows: BTC at 2026-09-09T02:15:09.329904624Z with a price of 78784.055 to buy and 78802 to sell, ETH at 2026-09-09T02:13:32.504698973Z with a price of 2494.4 to buy and 2495.15 to sell, SOL at 2026-09-09T02:14:07.508337859Z with a price of 103.6 to buy and 103.73 to sell, USDC at 2026-09-09T00:12:48.996942215Z with a price of 0.9995 to buy and 1.00001 to sell, USDG at 2026-09-08T11:48:27.749222185Z with a price of 0.99999 to buy and 1.00001 to sell, XRP at 2026-09-09T02:15:08.925016696Z with a price of 1.41375 to buy and 1.419 to sell, UNI at 2026-09-09T02:15:02.07492377Z with a price of 6.77166 to buy and 6.7832 to sell, LTC at 2026-09-09T02:13:41.928858872Z with a price of 53.93 to buy and 54.2344 to sell, HYPE at 2026-09-09T02:14:30.971399988Z with a price of 85.24 to buy and 85.622 to sell, and LINK at 2026-09-09T02:14:04.924464865Z with a price of 12.46 to buy and 12.48 to sell. The age of each quote is given in seconds, with fresh=1 indicating the quote is up to date. The spread_bp is the difference between the buy and sell prices, loss_pct is the percentage loss, and recovery_pct is the percentage recovery. The fund has 8 out of 10 fresh and valid quotes, and funding and trading rules remain independently gated, with no orders currently placed.

A news scan found the script output: this is a python script that runs in a fresh directory with specific libraries. It does not place orders and is based on a frozen note and output analyses. The script imports necessary libraries and sets up a directory for data. It uses environment variables for site URL and data token. The script defines a function to fetch data for given symbols and timeframes, handling pagination and errors. It uses a thread pool to execute the function for multiple symbols and timeframes, printing results and errors as they occur. The script uses specific parameters, including symbols 'BTC ETH SOL XRP LINK LTC UNI HYPE', timeframes '1Day' and '1Hour', and a specific date range.

A news scan found the following information about Darden Restaurants, which is referred to by its stock ticker DRI. The company had 18 actual SEC earnings releases from 2022 to June 2026, with 16251 to 74 individual observations confirmed. There were 12 training notice vintages that were incomplete, resulting in a full history test retrospective eligibility gap. Six holdout advance notices were verified, including notices from February 20 to March 20, 2025, May 22 to June 20, 2025, August 21 to September 18, 2025, November 20 to December 18, 2025, February 19 to March 19, 2026, and May 28 to June 25, 2026. All of these notices were for conference calls at 8:30 ET on the specified dates. The current notice is for August 27, with a paper window starting on September 17 and closing on September 23. The primary rule and six holdout outcomes were frozen, with no alternative picked using the holdout. There was no entry today, even if the edge passed, because the window starts on September 17. A monitor attempt was rejected due to a 20 armed cap, and the review is un armed, pending collector fallback. There was no guessed date or hold through.

A local collector script that runs at 02:10:19 produces 1705 characters across all 15 panels and adds the ticker symbols SEI, which is Solaris Energy, and OESX, which is Orion Energy Systems, to the existing 13 sources. The script shortens URLs from WH and Canada to listing pages, but still fetches the original pages. Changes in hash values are not considered news, and source coverage is only partial, with ongoing issues including a CENTCOM HTTP error and a challenge with Bombardier. The script will preserve the existing settings for December 18, 36, and 73, and replace the existing 111 hourly runs with 24 runs and a 60-second timeout. The next run will verify the actual saved footer, but will not automate any orders.

The venue rejected a separate Circle daily job because there were too many jobs, which means more jobs than allowed, with the error message too_many_jobs: crypto already had five active jobs, which were numbered 14, 25, 112, 113, and 114. The job numbered 113 was cancelled and a new combined funding and Circle hourly collector job was submitted to run twice, while the existing jobs numbered 14 and 25 remained. A test of the Circle numeric parser, which is a tool that interprets numbers, was successful at 02:08 UTC. At 04:00 UTC, the actual outputs will be reviewed to decide whether Circle needs a daily time slot, and currently there is no independent sixth job. The jobs numbered 112 and 114 were renewed and have an internal cutoff time of 04:01 UTC, which means that the venue accepting them does not necessarily mean they will be executed.

A test was done on September 9 at 02:08 UTC on the website https://www.circle.com/transparency, which loaded successfully with a status code of HTTP 200. A Python script using the requests library with a timeout of 20 seconds and the standard library HTMLParser was used to scrape the website. The script looks at the start tags and their attributes, and keeps the id if it starts with

The first request for SPY data started in 2008, but the first available data point is from 2016-01-04. As a result, the planned training period from 2010 to 2020 is unavailable. The holdout period from 2021 to 2025 and the model parameters will be preserved. The actual training will start in 2017, after a 252-session warm-up period, and in 2018 for 378 lookbacks, which have sufficient history. There will be no imputation of missing data from 2010 to 2015. This disclosure of shorter coverage is an amendment to be made before any portfolio results are shared. A snapshot census initial parse failed because the ticker NA was interpreted as NaN, so the parse was rerun with the option keep_default_na=False to prevent silently dropping the listed symbol.

This is a preregistered independent earnings-window test for DRI. Before looking at DRI's price history, the test will verify the issuer's earnings dates from 2022 to 2026, using dates with original advance notices known at least 5 sessions before the event. The test will follow a fixed rule: buy at the open 5 days before the event and sell at the close 1 day before the event, never holding through the event. The test will be trained on data from 2022 to 2024 and will use untouched data from 2025 to 2026 as a holdout. The baseline assumptions include 10 basis point fees and 10 basis point slippage on each side, with a stress test of 10 basis point fees and 25 basis point slippage on each side. The test will also include a sensitivity sweep of different training periods, including T-3, T-5, and T-10, and exit periods, including T-1 and T-2, but the primary test will use T-5 and T-1 and will not use the holdout data to choose the best approach. The test requires an event sample of at least 20 and will separate the results if the advance notice history is missing. The test will use daily bars and will not validate 15-minute stops. The test requires a positive expectancy and uncertainty in the holdout period, as well as affordability of the actual window and whole-share, before making any proposal. The test will also keep the Merck label expansion distinct from earnings and will not use any potential edge from DRI.

A news scan found a description of the QTR9090200 strategy. It freezes eight established asset proxies: SPY, EFA, EEM, IEF, TLT, GLD, DBC, and UUP. The strategy is based on the Moskowitz/Ooi/Pedersen 2012 paper, which uses a concept called time series momentum, meaning the tendency of assets to continue performing well if they have done so in the past. The QTR9090200 strategy adapts this concept to cash-only ETFs. It works as follows: at the end of each month, if the return of an ETF over the past 252 sessions is greater than 0, the strategy enters a position in that ETF at the next session's open, allocating 1/8 of the capital to each ETF, without using leverage. The position is closed at the last session of the month. The baseline strategy does not exit positions during the month, but a managed candidate uses a stop-loss, which is a rule to sell a security when it falls to a certain price, set at one times the average high-low range of the prior 20 sessions, and a target of 1.5 times the risk, with a horizon of less than or equal to one month. If the high of the previous day reaches the target, the stop-loss is raised to the entry point the next day. The strategy assumes fees of 1 basis point plus slippage of 5 basis points per side, and stress tests with fees of 1 basis point plus 15 basis points. The strategy is tested using a diagnostic sweep with lookbacks of 126, 252, and 378 sessions, and targets of 1.5 and 2, at both cost levels, without selecting the best configuration. The training period is 2010-2020, with a chronological holdout period of 2021-2025, and an extension in 2026. The strategy is compared to an equally weighted buy-and-hold strategy, a bootstrap of month-blocks across assets preserving correlation, a regression benchmark, and 100 monthly paths scenarios using a method called Monte Carlo. The default cash return is 0, and adjusted ETF prices include adjustment assumptions and fixed operating costs. The monthly unconditional close is a scheduled price approximation, and the stop semantics do not prove 15-minute fills. A daily cross-section screen is separate from historical validation. The strategy is adopted only if the cost stress is positive, uncertainty is credible, and exits and lot constraints are independently validated; otherwise, the result is zero. The assets are frozen before downloading history.

A fund manager has set aside some rules, including momentum and funding rules, to check for primary changes. The manager uses a three-level decision tree. The first level looks at recovery and flows, including liquid operations, authorized payouts, and Bitcoin sales. The second level examines the universe and structure, including eligibility, execution, and fees. The third level considers distinct edges, including weekend effects, robustness, and risk. There are at least 18 questions to consider, with primary endpoints being closest to the event. The manager has completed research using daily and hourly bars, sub-five-minute quotes, and dated evidence, discarding undated headlines. The starting cash is $235.86, with a goal of repeatable net payoff. The manager will examine a trend diagnostic on eligible non-stable top 10 stocks, using a slower SMA20/100 trend, and sweep fast 10/20/30 in 2025 only. Trading will only occur if both frequency and execution gates pass, and any positive result will require prospective validation due to existing 2026 exposure.

A trading plan was set at 02:00:50 UTC with a budget of 41 minutes and a hard stop at 45 minutes. The reserve time is 5 minutes. The plan considers three ideas: tanker destruction can increase owner profits, component scarcity can benefit producers, and momentum can provide an edge. The plan has a 12-part question tree: (1) delivered scarcity and missing physical units, which involves IMO, cargo, and loadings; (2) substitution and spare capacity, which involves restart dates; (3) rent capture and contract repricing, which involves accepted net fixtures; (4) margins and cost pass-through, which involves fuel, insurance, and labor; (5) capital and plant completion, which involves financing conditions and dilution; (6) customer constraint and single-source inputs, which involves switch qualification time; (7) order quality and cancellations, which involves deposits and cash conversion; (8) global flows and regional arbitrage, which involves FX and venue; (9) trade policy and exact classification, which involves plant and customer exposure; (10) crypto and power/chip constraint, which involves miner and hash price cost capture; (11) market response and breakout regime, which involves untouched costed holdout; (12) execution and whole-share affordability, which involves 15-minute and overnight exit risk. The plan will resolve with dated quantities and issuer cash/margins, and will discard repeated headlines as independent evidence. The candidate screen is for discovery, not an unbiased predictive test. A new rule will be frozen before testing, which includes a prior 20-day close breakout, next open, ten-session holding, pre-2026 development and 2026 holdout, and 10, 25, and 50 basis points each-side cost sensitivity, with no retuning after holdout.

A fund manager reviewed past results and found that short-term rules did not work, but this does not mean a longer-term trend with diverse investments would not work. The manager set aside 349.01 USD. The manager looked at three main areas of research, called roots. Root A is about data integrity, which means making sure data is accurate and complete, and includes three parts: A1, checking if US census data is complete and up to date; A2, checking global coverage, including which exchanges are included, foreign exchange rates, and which venues are eligible; and A3, checking cryptocurrency data, including which pairs are listed, stablecoins, and costs. Root B is about documented edge, which refers to a proven advantage in investing, and includes three parts: B1, trend mechanism, which includes underreaction, crisis convexity, and crowded reversals; B2, implementation, which includes using only cash or also futures, and how to execute and exit trades; and B3, validity, which includes testing the edge over time, testing multiple times, and selecting the best performers. Root C is about candidates, which includes three parts: C1, leaders, which includes momentum, range, and volume; C2, primary evidence, which includes index methodology, company filings, and issuer incentives; and C3, fundamentals, which includes cash generation, leverage, and valuation. Root D is about decision, which includes four parts: D1, statistics, which includes win/loss, mean confidence interval, and benchmark regression; D2, feasibility, which includes whole lots, a maximum of 212.7475 USD per position, a loss cap of 8.5099 USD, and 15-minute stops; D3, consequences, which includes first price effect, second funding and competition, and third crowding and execution; and D4, watch, which includes monitoring INBX full data, TYRA and IRD news, and existing collectors. The manager will use daily data to find a medium-term trend and discard intraday noise for screening, but will require intraday data to confirm stop execution. The manager starts with 245.69 USD in cash and no positions, with the goal of finding a positive costed rule that works independently, with a validated edge that takes into account execution costs. The hard deadline is 02:45:50 UTC, and the finish reserve starts at 02:39:50. There will be no reoptimization of the frozen reversal family.

The trading plan started on September 9 at 02:00:48 UTC and had a hard stop at 02:45:48, with a planned duration of 43 minutes, including a final 5 minutes. The plan began with $49.14 in cash, aiming for a positive and repeatable pre-event expectancy, with a binding constraint based on independent actual-window evidence. The plan considered daily sessions and original release dates, discarding intraminute noise except for execution validation. A question tree was used to guide the plan, which included the following questions: 1. Was the universe of potential investments missed? This included checking the US census for listing completeness and dated issuer event intersection, as well as global original calendars and venue/FX eligibility, and crypto earnings/FDA linkage versus mandate creep. 2. What makes a date investable? This included considering timing, such as notice known before entry and early competing pipeline news, business factors like organic cash bridge and funding and dilution, and market factors like range/liquidity and costed exits. 3. Can evidence defeat zero size? This included statistical considerations like chronological holdout and selected sample/multiple tests, risk factors like gap stress and quarterKelly, and off-run factors like verified collector output and review reminders versus first news. The plan applied lessons 454-456 and did not revive failed rules through retuning. There were no holdings in the queried ledger, and all losses were closed, with the operator censoring 2.

A news scan found the headline about a crypto fund. The fund is for research only and is not eligible. On 2026-09-09T02:00:45.847978+00:00, the fund's data showed the following information: the recipient had 598.50007353 confirmed bitcoin and 3.62e-05 mempool net bitcoin, with 241 transactions. The reserve had 3597.47201026 confirmed bitcoin and 0.0 mempool net bitcoin, with 584 transactions. There were two controlled transactions, one with the id 1d690f3b96b878067f3a445b74dfb8fab4201c0455d88ac98cc14a927e7858d7 and the other with the id 7c0fb4ffff35bf9894211d3c97abbb58fc127573a1805880103dd19e40528523, both occurring at block time 1788877531. The recent page number was 30, with 3 controlled transactions, and the page is not a complete history. The liquid tip was 4051232 at time 1788756550, with an age of 162696 seconds and 1 transaction. The second tip was 4050335, with an age of 216516 seconds. A source from https://x.com/Blockstream provided a statement, which is an attributed claim, not proof of payout. The statement said that Blockstream and the Liquid Federation have been working to resolve the situation and return assets. Updated software has been deployed and the teams are focused on hardening the network. No automatic trade will be made based on this information.

A news scan found the following headlines: for INBX, a link to the news release about INBRX-106 nearly doubling response rate and achieving interim median PFS of 9.6 months in Phase 2 HNSCC study, a link to the webcast presentation for primary endpoint results for INBRX-106 Phase 2 first-line HNSCC study, and a link to the second quarter 2026 financial results. For TYRA, a link to the conference call on initial results of oral dabogratinib from the SURF302 Phase 2 study for IR-NMIBC, a link to the second quarter 2026 financial results and recent highlights, and a link to the first quarter 2026 financial results and recent highlights. For IRD, a link to the presentation of OPGX-BEST1 cohort 1 clinical data in investor webcast and at upcoming leading ophthalmology conferences, a link to the financial results for second quarter 2026 and corporate update, and a link to the BEST1 program update with detailed timeline for OPGX-BEST1 results.

The fund manager made notes on September 9, 2026, at 01:45:26 for research purposes only, noting that hashes are not events and 0 are eligible. The manager checked the following companies and organizations: Maersk, which is okay, with a hash of 67595b405bcb and a link to their news page at https://www.maersk.com/news; Typhoon, also okay, with a hash of def05fb81eab and a link to an article about the east Asia typhoon season at https://www.maersk.com/news/articles/2026/09/04/east-asia-typhoon-season-impacting-logistics; Hapag, which is okay, with a hash of 8ab284858970 and a link to their operational updates page at https://www.hapag-lloyd.com/en/services-information/operational-updates/updates/2026/09/ops-update-middle-east-week36.html; SeaIntel, which is okay, with a hash of 9634cba5cd0f and a link to their press room at https://sea-intelligence.com/press-room; WH, which is okay, with a hash of 158ad22ff216 and a link to a presidential action at https://www.whitehouse.gov/presidential-actions/2026/09/modifying-the-scope-of-products-of-canada-subject-to-the-additional-duties-imposed-to-offset-canadian-discrimination-against-the-united-states-with-respect-to-motor-vehicles/. The manager also checked Bombardier, which is a challenge, with a hash of 9b4b7b595334 and a link to their news page at https://bombardier.com/en/media/news; GD, with links to their SEC filings at https://data.sec.gov/submissions/CIK0000040533.json and https://www.sec.gov/Archives/edgar/data/40533/0001193125-26-340490/d129075d8k.htm and https://www.sec.gov/Archives/edgar/data/40533/0001193125-26-332899/d122083d8k.htm; CENTCOM, which had an error with an HTTP error; FRO, which is okay, with a hash of 3aaf7244dc4a and a link to their second-quarter results at https://www.frontlineplc.cy/fro-second-quarter-and-six-months-2026-results/; STNG, which is okay, with a hash of c54b428d736e and a link to their updates on third-quarter TCE rates at https://www.scorpiotankers.com/scorpio-tankers-inc-announces-updates-on-third-quarter-2026-tce-rates-and-time-charter-out-agreements/; ASPI, with a link to their press releases at https://ir.aspisotopes.com/news-events/press-releases/rss and a specific release at https://ir.aspisotopes.com/news-events/press-releases/detail/119/asp-isotopes-hosts-inaugural-capital-markets-day-and; ACP, which is okay, with a hash of 6e21e0bc4ee1 and a link to a PDF about the postponement of the maximum authorized draft adjustment at https://pancanal.com/wp-content/uploads/2026/09/ADV-33-2026-Postponement-of-Maximum-Authorized-Draft-Adjustment-in-the-Neopanamax-Locks.pdf; and Canada, which is okay, with a hash of 09084e6193e2 and a link to a news page about counter-tariffs at https://www.canada.ca/en/department-finance/news/2026/08/list-of-products-from-the-united-states-subject-to-counter-tariffs-effective-september-8-2026.html. The manager attempted to check 13 panels and noted that errors are due to missing coverage, not a lack of news.

The fund is called Catalyst Research and its size is 0 as of September 9, 2026, at 01:45:29 UTC. A news scan found the headline that Telix completed enrollment in Phase 3 BiPASS on September 1, 2026. Another news scan found the headline that Mission Produce announced its fiscal 2026 third quarter financial results on September 8, 2026. ABM reported its fiscal third quarter 2026 results on September 8, 2026. Roivant announced positive results from the PHocus study on September 8, 2026. The SEC website has a filing from UNFI on September 8, 2026. Roivant's earnings were reviewed on September 8 at 10:30 UTC, with a call at 12:00 UTC and an earnings report at 10:15 UTC. The PHocus study had primary and secondary endpoints, including a 6-minute walk test. The earnings schedule includes ABM and UNFI in the morning and AVO after close on September 8, with reviews at 12:30 UTC on September 8 for ABM and UNFI, and on September 9 for AVO. There are also macro events scheduled, including a QFR at 14:00 UTC and an SCE at 15:00 UTC on September 8, and an ECEC and QSS on September 9. All branches of the decision tree require recording primary data and verifying the date if the outcome is favorable, and cashing out if it is adverse or delayed. The decision tree does not include orders. The update ends with 4 RSS feeds and 1 SEC filing, and includes static reminders that are not exhaustive.