Notes

A calculation was done using the number of days available in the second quarter as a hypothetical proxy for the third quarter. This showed that 8% of the unfixed amount, multiplied by $10,000 per day, would result in an incremental $2.9704 million. For the full fleet, this would be $37.13 million, ignoring any bookings. The operating cash flow for the second quarter was $205.9 million, and after subtracting capital expenditures of $13.3 million, the result was $192.6 million. The first half of the year had $371.4 million, and after subtracting $24.7 million, the result was $346.7 million, before considering non-cash items or claims, and this is not the normalized owner free cash flow. No new valuation multiple was selected. https://www.sec.gov/Archives/edgar/data/1649313/000110465926102581/bwlp-20260630xex99d2.htm

A Python scenario, which is not a measured forecast, calculates a constant 25 fleet times 92 quarter days times 99 percent times 88,000, which equals 200.376 million TCE proxy. This is a measure of shipping revenue, where TCE stands for time charter equivalent, a way to calculate the average daily revenue. The remaining 1 percent times 10,000 per day equals 0.23 million uplift. It is assumed that there are 40 days of net freight, which is 3.52 million, diluted by 5 out of 15 unpaid extra days, resulting in 78,222.22 divided by 64,000 daily. To preserve 88,000, it would require 440,000 divided by 1,320,000 more net freight, excluding bunker effects, which refer to fuel costs. The 40-day and 5-day periods are assumptions, and the 15-day Cape differential, which is a measure of shipping route profitability, is based on the issuer's 10K report, a financial document. No probability of stock performance is inferred from this scenario. https://www.dorianlpg.com/newsroom/news-and-press-releases/news-details/2026/Dorian-LPG-Ltd--Announces-VLGC-Newbuilding-Contracts-Forward-Chartering-Estimates-and-New-Credit-Facility/default.aspx

A news scan found the headline 'Federal Reserve calendar retrieved Sep 20'. According to this, on September 22, an event is scheduled at 13 ET, and on Monday, September 21, events are scheduled at 10 and 15, with a specific event at 16:15 ET. These are context points and not validated triggers. https://www.federalreserve.gov/newsevents/2026-september.htm

A calendar retrieved on September 20 shows the following events: on Monday, September 21, the SCE Household Spending report is scheduled for 11:00 ET; on Tuesday, September 22, the Philadelphia Non-manufacturing report is scheduled for 8:30 ET and the Richmond Manufacturing report is scheduled for 10:00 ET. Strong orders and prices may lead to a test of cash capture; weak orders may lead to a demand destruction countercase; there are no surprise or automatic orders expected from the calendar. https://www.newyorkfed.org/research/calendars/i-sep26.html

The company MLKN will release its earnings on August 25, before the market opens, and will have a call at 08:30 ET on September 22. The fund manager needs to recheck orders, input costs, and cash, considering the context of industrial demand, and notes that there is no supply position. https://news.millerknoll.com/2026-08-25-MillerKnoll-Schedules-First-Quarter-Fiscal-Year-2027-Conference-Call-and-Webcast

A news scan found the headline 'IONS R92008 September 22 catalyst superseded'. On September 4, the issuer filed an 8K report that mentioned the September 3 approval of ZANVASTRO. This means the former September 22 PDUFA, which is a Food and Drug Administration decision date, is no longer an upcoming binary event, meaning it is no longer a yes or no decision. Third-party calendars still list the event as pending, but the original filing takes precedence. https://www.sec.gov/Archives/edgar/data/874015/000114036126035657/ef20081649_8k.htm

A news scan found the headline about MRK R92008. On June 30, 2026, a 10Q report, which is a quarterly financial report, said that there will be a US label update for WINREVAIR HYPERION-based products on September 21, 2026, at the Food and Drug Administration, or FDA, as part of a PDUFA, which stands for Prescription Drug User Fee Act, a process for reviewing new drugs. The exact announcement time is not given, only the date. A search of targeted issuers on September 20 found no disposition, which means no decision or action was taken, and this does not prove that something is pending. The approval of ZENITH in October 2025 should not be used as an outcome for HYPERION. https://www.sec.gov/Archives/edgar/data/310158/000031015826000212/mrk-20260630.htm

A news scan found the headline about BWLP R92008 quarter cash and charter cost offsets. On August 28, the Q2 and H1 report showed that the operating cash flow was 205.9 million in Q2 and 371.4 million in H1, with capital expenditures of 13.3 million in Q2 and 24.7 million in H1. The index-linked charter hire increased by 7.7 million in Q2. The Product Services segment had a realized gain of 127.4 million and an unrealized loss of 145.5 million, resulting in a net loss of 30.9 million in Q2. Note that the realized trading gain is not the same as the total owner free cash flow. The available days in Q2 were 3713, and the reconciliation for H1 was 7278, which is different from the narrative of 7300. This discrepancy should be preserved and not silently combined. https://www.sec.gov/Archives/edgar/data/1649313/000110465926102581/bwlp-20260630xex99d2.htm

The issuer reported on August 28 that for the third quarter, 92% of available days were booked, which is approximately 88,000 days per year, excluding IFRS 15 and FFA, which stands for Financial Framework Agreement, a type of contract. The fixed time charter, or TC, for the second half of the year is 41% at 44,100 dollars per day, plus a 4% FFA of 48,000 dollars. In the second quarter, the time charter was 53% at 64,000 dollars, and the total charter equivalent, or TCE, was 74,000 dollars, with 71,600 available days out of 71,600 calendar days, after subtracting 16.4 million dollars for IFRS 15 and 12 million dollars for FFA. The fleet order book has 155 vessels, which is 35% of the total, with 27 vessels to be delivered in 2026 and 13 remaining, posing a risk to next year's supply. The issuer claims that terminal capacity is the first constraint, followed by vessels, as US export capacity has expanded. This claim is based on information from an operator and not from an independent shipping census, which is a survey of ships. https://www.sec.gov/Archives/edgar/data/1649313/000110465926102581/bwlp-20260630xex99d1.htm

A May 27 filing for the fiscal year that ended on March 31 showed that Helios had 99% of the revenue, with no individual charterer having more than 10% of the net pool revenue. The net pool receivable was $123.4 million, including $28.6 million from WC, which refers to working capital, a company's available money. A risk factor stated that waiting times for the Panama Canal are typically unpaid and higher tolls may not be reflected in charter rates, which are the prices paid for renting a ship. Using the canal saves approximately 15 days compared to going around the Cape. The costs of a voyage are borne by the voyage owner or pool, while the time charter customer bears substantially all voyage expenses. The distribution of the pool's profits comes after costs, working capital, and collection of payments. There was no single component that relied on one source for its revenue. https://www.sec.gov/Archives/edgar/data/1596993/000159699326000025/lpg-20260331x10k.htm

The issuer of the LPG R92008 fund fixed the September quarter calendar days at 99% of days, with a rate of over 88,000 USD per day, excluding potential demurrage costs that may occur in September. This refers to a fee paid when a ship stays in a port longer than expected. The fund has a fleet of 25 ships, including three 2030 Panamax dual-fuel ships with a capacity of 90,000 cubic meters, which were newly built for 345 million USD. On September 2, the fund secured a seven-year facility of 368.4 million USD, with an interest rate of 140 basis points over the Secured Overnight Financing Rate, or SOFR, which is a measure of the cost of borrowing money overnight. The facility is split into a term loan of 213.4 million USD and a revolver of 155.1 million USD, totaling approximately 368.5 million USD. The forecast does not expect the fund to collect cash, but the borrower has a favorable bias, meaning they are likely to make payments on time. https://www.dorianlpg.com/newsroom/news-and-press-releases/news-details/2026/Dorian-LPG-Ltd--Announces-VLGC-Newbuilding-Contracts-Forward-Chartering-Estimates-and-New-Credit-Facility/default.aspx

A news scan found the headline about an operator announcement made on September 20. The announcement said that on September 3, there were 9 NeoPanamax and 25 Panamax daily slots, and on September 15, there were 23 Panamax slots. This means that the maintenance scheduled for September 22 will not reduce the number of slots from 25 on that date. Note that the number of slots is not the same as the number of actual transits. Additionally, the announcement mentioned that A33 will retain a 48ft draft, A35 will allow consecutive Panamax bookings across periods, and the latest index is A35. https://pancanal.com/en/panama-canal-adopts-additional-measures-to-address-reduced-precipitation-in-the-canal-watershed/

A news scan found the headline about LPG R92008 ACP activity. The operator reported that in August, there were 33.1 arrivals per day and 33.19 transits per day. The water time was 34.90 hours and the in-transit time was 11.27 hours. A total of 175 to 178 reserved slots were used by Neo, out of 227 slots available at auction, which is a type of bidding process. For September, the plan is to have a 5-hour estimated maintenance on September 21, and a 10-day tentative maintenance from September 22 to October 3 for Panamax, a type of large ship, culvert, which is a pipe that allows water to flow under something. These are just plans and not the actual outages or wait times for September. https://pancanal.com/wp-content/uploads/2026/09/adv-34-2026-monthly-canal-operations-summary-august-2026.pdf

A direct scheduler output journal note, numbered 3890, with a timestamp of 2026-09-20T06:31:08 UTC, shows that job 284 ended successfully with an exit code of 0, completing 11 out of 11 tasks. On September 18, the CSTM price was 25.55, the Simple Moving Average, a way to smooth out price data, over 200 periods was 27.067, the Relative Strength Index, a measure of price change, over 2 periods was 26.88, and the Average True Range, a measure of price volatility, was 0.921, compared to the Average Daily Range, a measure of daily price movement, of 0.838. There was no new market data since the previous run. Monitor 440 was resolved. A query of trades, called query_trades, made at 08:01, returned no results, with a live sample size of 0. https://runway.adityaperswal.workers.dev/data/stock-bars

A news scan found the headline 'KALU R92006 mandated CLI RSI diagnostic'. The symbol KALU was analyzed with 1375 bars of data. The return percentage was 64.80387605281646, compared to a buy and hold percentage of 35.89183283255941. The maximum drawdown percentage was 23.135230420011744. The win rate was 72.09302325581395, with an average trade percentage of 1.3086354870343437, based on a sample of 43 trades. This analysis used a strategy called CLI with a diagnostic tool called RSI, which stands for Relative Strength Index, a way to measure stock performance. The analysis was done over 2000 days, using default RSI settings of 14, 30, and 70, with no fees or slippage, and same close execution, using data from IEX, with no explicit adjustment, and no held apart window. This analysis is not comparable to other models, such as SIP/RSI2/next open costed model, and does not include trade validation. The source code was audited in a file called engine/bars.js. https://runway.adityaperswal.workers.dev/data/stock-bars?feed=iex

A news scan found the headline 'CSTM R92006 mandated CLI RSI diagnostic'. The symbol CSTM was analyzed with the following results: it had 1372 bars, a return percentage of 77.12, and a buy and hold percentage of 73.73. The maximum drawdown percentage was 25.22, the win rate was 68.29, and the average trade percentage was 1.53. This analysis was based on a sample of 41 trades. The analysis used a strategy called Relative Strength Index, or RSI, which is a way to measure stock performance, with settings of 14, 30, and 70 over 2000 days. The analysis did not include fees or slippage, which are costs of buying and selling, and it used the same closing price for execution. It also used data from IEX, a stock exchange, and did not adjust for any held apart window, which is a period of time when trades are not made. This analysis is not comparable to other models, such as SIP/RSI2/next open costed model, and did not include trade validation, which is a way to check if trades are correct. The source code for the analysis was audited in a file called engine/bars.js. https://runway.adityaperswal.workers.dev/data/stock-bars?feed=iex

A news scan found the headline 'CENX R92006 mandated CLI RSI diagnostic'. The symbol CENX was analyzed with the following results: it had 1375 bars of data, a return percentage of -37.26664061837659, and a buy and hold percentage of 126.00637127135825. The maximum drawdown percentage was 63.73420657949548, the win rate was 62.857142857142854, and the average trade percentage was -0.8042348342885773, based on a sample of 35 trades. This analysis used a strategy called CLI, which is a type of trading strategy, with an RSI, or Relative Strength Index, diagnostic. The RSI is a tool used to measure the strength of a stock. The analysis was done over 2000 days, using a default RSI setting of 14, 30, and 70, with no fees or slippage, and same close execution, using data from IEX, with no explicit adjustment, and no held apart window. This analysis is not comparable to other models, such as SIP/RSI2/next open costed model, and does not include trade validation. The source code for the analysis was audited using the engine/bars.js code. https://runway.adityaperswal.workers.dev/data/stock-bars?feed=iex

A news scan found the headline 'AA R92006 mandated CLI RSI diagnostic'. The results of a trading strategy for the stock symbol AA are as follows: the strategy had 1373 trades, with a return percentage of 7.770770983978892, compared to a buy and hold return percentage of 34.71927162367221. The maximum drawdown percentage was 31.55825010044823. The win rate was 66.66666666666, and the average trade percentage was 0.5041038314435492, based on a sample of 33 trades. The strategy used a Relative Strength Index, or RSI, which is a measure of stock price movement, with settings of 14, 30, and 70, over 2000 days, with no fees or slippage, and same close execution, using data from IEX, with no explicit adjustment, and no held apart window. This strategy is not comparable to other models, such as SIP/RSI2/next open costed model, and does not include trade validation. The source code was audited in the engine/bars.js file. https://runway.adityaperswal.workers.dev/data/stock-bars?feed=iex

A news scan found the headline about CSTM R92006 valuation sensitivity using July 29 guidance and September 18 price. The company has 135.527728 million shares, with a price of 25.55 dollars per share, resulting in a market capitalization of 3462.7334504 million dollars and an enterprise value of 5222.7334504 million dollars. The enterprise value to EBITDA guide midpoint is 5.2227334504. The floor yield percentage for free cash flow is 8.66367580113177 percent. The free cash flow conversion percentage for the first half of the year is 16.60839160839161 percent. The guidance for the second half of the year segment EBITDA is between 408 and 448 million dollars. The free cash flow needed for the second half of the year is 205 million dollars, with a ratio of free cash flow in the second half to the first half of 2.1578947368421053. The hurdle free cash flow at different yields is as follows: at a 7 percent yield, the hurdle free cash flow is 298.730671627351 million dollars, at an 8 percent yield, it is 341.4064818598297 million dollars, and at a 10 percent yield, it is 426.7581023247871 million dollars. A grid of possible free cash flows and yields shows the following prices: if the free cash flow is 200 million dollars, the price at a 7 percent yield is 21.08161111608731 dollars, at an 8 percent yield is 18.446409726576395 dollars, and at a 10 percent yield is 14.757127781261117 dollars, if the free cash flow is 300 million dollars, the price at a 7 percent yield is 31.62241667413096 dollars, at an 8 percent yield is 27.669614589864594 dollars, and at a 10 percent yield is 22.135691671891674 dollars, and if the free cash flow is 400 million dollars, the price at a 7 percent yield is 42.16322223217462 dollars, at an 8 percent yield is 36.89281945315279 dollars, and at a 10 percent yield is 29.514255562522234 dollars. Another grid of possible EBITDA and multiples shows the following prices: if the EBITDA is 720 million dollars, the price at a 5 times multiple is 13.576557558760227 dollars, at a 6 times multiple is 18.88912356001423 dollars, and at a 7 times multiple is 24.201689561268232 dollars, if the EBITDA is 1000 million dollars, the price at a 5 times multiple is 23.90654700564301 dollars, at a 6 times multiple is 31.285110896273565 dollars, and at a 7 times multiple is 38.663674786904124 dollars. The market capitalization to enterprise value ratio is in millions of dollars. The guide floor free cash flow is 300 million dollars, but this is not guaranteed cash. The yields of 7, 8, and 10 percent and the multiples of 5, 6, and 7 times are assumptions and not approved targets. The previous fiscal year segment EBITDA was 720 million dollars per the first quarter deck, and the guide midpoint is 1000 million dollars. The debt as of June 30 is not the same as the current debt as of September. The hurdle of 31.4886 is an arithmetic number, not a fair value. https://www.sec.gov/Archives/edgar/data/1563411/000156341126000190/a2026-q2xearningspressre.htm

A simulation of CENX, which is a stock symbol, was run for the recent quarter using a policy called quarter_at2R_BE. This policy is a set of rules for buying and selling stocks. The simulation used 20,000 paths, which are like different possible outcomes, and had a stop loss of 66.33%, a target profit of 30.245%, and a timeout of 3.425%. The simulation also used a trim of 41.735%, which means that some trades were stopped early to limit losses. The average profit per share was 1.606397255146701, and the maximum affordable quantity was 3 shares. The simulation was observed on 2026-09-20. A regression analysis was done to compare the daily simple returns of CENX to the SPY, which is a stock market index, using a statistical method called OLS, or ordinary least squares, with an intercept. The simulation used a random seed of 92006, and had 20,000 paths per cell, 60 sessions, and 5-day blocks of open, high, low, and close prices. The simulation also included a 20 basis point slip, which is like a small fee, and an overnight gap stop, which is a rule to stop trading if the price changes too much overnight. The results of the simulation are based on a beta distribution, which is a way of modeling probabilities, with parameters of 1 and 1, and a win posterior, which is a measure of the probability of winning. The simulation compared the results of the quarter_at2R_BE policy to a policy called BE, or buy and hold, on the next day. https://runway.adityaperswal.workers.dev/data/stock-bars?timeframe=1Day&feed=sip&adjustment=all

A simulation of the stock symbol CENX was recently run with a fixed policy. The simulation used 20,000 paths and had a stop percentage of 60.39%, a target percentage of 32.76%, a timeout percentage of 6.85%, and a trim percentage of 0%. The profit percentage was 38.56%. The mean profit and loss per share was $1.96, with a 1% chance of a loss of $6.07 per share and a 5% chance of a loss of $4.98 per share. There was a 95% chance of a profit of $13.66 per share. The maximum affordable quantity was 3. A news scan found the observation date to be 2026-09-20. A regression analysis was done using daily simple returns, comparing CENX to the SPY stock index, but no claims of causality or significance are made. The simulation used a beta of 1,1 and a win posterior. It also used 20,000 paths per cell, 60 sessions, and 5-day blocks of opening, high, low, and closing prices, with a 20 basis point cost per side and overnight gap stops. The simulation stopped on the first ambiguous intraday trade and used a 2R quarter then BE next day comparator. The unconditional model was not validated with live odds, and fractional trims would have exited the regular trading hours. A 15-minute venue path was not observed. https://runway.adityaperswal.workers.dev/data/stock-bars?timeframe=1Day&feed=sip&adjustment=all

A simulation for the stock symbol CENX was observed on 2026-09-20. The simulation used a long regime, meaning it was looking to buy and hold the stock, with a policy called quarter_at2R_BE, which means it aimed to make a certain amount of money over a quarter, then switch to a different strategy the next day. The simulation ran 20,000 paths, or possible scenarios, and used a stop loss, which is a rule to sell a stock when it falls by a certain percentage, in this case 71.8%. It also used a target percentage, which is a rule to sell a stock when it rises by a certain percentage, in this case 26.195%. The simulation found that the average profit per share was 0.7839162555960636, and the median number of days to complete a trade was 14. The simulation also found that the maximum number of shares that could be bought without exceeding a certain risk level was 3. A news scan found the headline was not available, but the simulation used a type of analysis called regression, which is a way of looking at how two things are related, in this case the daily returns of CENX and the SPY stock index. The simulation did not claim that one caused the other, just that they were related. The simulation also used something called bootstrap, which is a way of testing how well a model works by trying it many times, and something called beta, which is a measure of how much a stock's price moves compared to the overall market. https://runway.adityaperswal.workers.dev/data/stock-bars?timeframe=1Day&feed=sip&adjustment=all

A simulation for the stock symbol CENX was run on September 20, 2026, with a long fixed policy. The simulation used 20,000 paths and had a stop percentage of 66.82%, a target percentage of 28.57%, and a timeout percentage of 4.61%. The results showed a profit percentage of 32.25% and a mean profit and loss share of 0.97. The maximum affordable quantity was 3. A regression analysis was also done, comparing the daily simple returns of CENX to the SPY stock symbol using a statistical method called ordinary least squares, or OLS, which is a way to find the best fit line. The simulation used a random number generator with a seed of 92006 and assumed 20,000 possible paths for each scenario, with 60 trading sessions and 5-day blocks of opening, high, low, and closing prices, or OHLC blocks. The simulation also accounted for overnight gaps and stop losses, with a 20 basis point, or 0.2%, cost per trade. The results were compared to a benchmark, with 2 times the return on the second day, and the performance was measured against a buy and hold strategy, or BE, the next day. The model used independent and identically distributed, or IID, data with 20,000 means and a beta distribution with parameters 1 and 1, which is a type of statistical distribution. The simulation did not claim to predict the future or show causality, and the results were not validated with live trading data. https://runway.adityaperswal.workers.dev/data/stock-bars?timeframe=1Day&feed=sip&adjustment=all

A news scan found the headline about a simulation for the symbol CSTM, which is a stock ticker symbol, on 2026-09-20. The simulation used a policy called quarter_at2R_BE, which means it followed a certain set of rules for a quarter, then compared to a benchmark the next day. The simulation had 20000 paths, or possible outcomes, and used a stop loss, which is a rule to sell a stock when it falls by a certain percentage, of 64.725 percent. It also used a target percentage of 33.415 percent and a timeout percentage of 1.8599999999999999 percent. The trim percentage, which is the percentage of trades that are stopped, was 44.945 percent. The profit percentage was 46.915 percent. The mean profit and loss per share was 0.8587949744897898. The simulation also found that the maximum affordable quantity, or the most of the stock that could be bought, was 4. The results were observed on 2026-09-20. The simulation used a type of analysis called regression, which is a way of understanding the relationship between two things, to look at daily simple returns, which is a measure of how much a stock's price changes, compared to the SPY, which is a stock market index. The simulation did not claim that one thing caused another, and did not make any claims about significance, which is a measure of whether a result is due to chance. The simulation used a type of statistical analysis called bootstrap, which is a way of estimating how likely something is, and a type of probability distribution called beta, which is a way of understanding the likelihood of different outcomes. The simulation was run 20,000 times, and used a random number generator with a seed of 92006. The simulation used 60 sessions, which is a period of time, and 5-day blocks of open, high, low, and close prices, which are the prices of a stock at the start, highest point, lowest point, and end of a period of time. The simulation also used a 20 basis point, which is a unit of measurement for interest rates, per side, and accounted for overnight gaps, which is when a stock's price changes suddenly, and stop slip, which is when a stop loss is triggered. The simulation used a 2R quarter, which means it followed a certain set of rules for a quarter, and then compared to a benchmark the next day. The simulation did not validate live odds, which is a measure of how likely something is to happen, and did not use fractional trims, which is a way of stopping trades, that would exit during regular trading hours, and did not observe a 15-minute venue path, which is a period of time. https://runway.adityaperswal.workers.dev/data/stock-bars?timeframe=1Day&feed=sip&adjustment=all