A question was asked about whether corporate-action basis could change the selection on Monday. The analysis was based on two premises. First, the same 33 series of 11 ETFs, including raw, split, and all, were fetched from a website. Second, the provider of the data defines all as split, dividend, and spin-off, and raw as unadjusted. The rule requires adjusted returns. The findings showed that the eligible ranking was XLK with 20.034955%, XLE with 19.704288%, XLV with 12.444763%, and XLF with 5.419049% when all adjustments were included. The same ordering was seen when only split adjustments were included, with percentages of 19.737064%, 18.000057%, 11.537557%, and 4.692775% respectively. However, when only raw prices were used, the eligible ranking changed to XLV with 11.537557%, XLF with 4.692775%, XLK with 2.275285%, and XLE with 0.138153%, replacing XLE with XLF. A comparison of 12-month returns showed that XLK had a return of -30.147736% and XLE had a return of -28.536504% when raw prices were used, versus returns of 40.437523% and 47.372061% when all adjustments were included, indicating that splits can create false losses. The eligibility of the ETFs remained unchanged across different bases for all 11 ETFs. The difference between dividend-inclusive and split-only adjustments was calculated, with mean differences of 0.297891% for XLK, 1.704231% for XLE, and 0.907206% for XLV, attributed to non-split adjustments, chiefly distributions. However, the exact dividend amount was not verified. A counterexample to the claim that dividend-driven selection is important was found, as the same top 3 ETFs were selected without dividend adjustments. The valid inference is that splits are material to membership, but dividends are material to levels, but not to today's membership. The soundness of the analysis is limited to the same vendor and current vintage, and not independent market verification or point-in-time backtest. The collection of additional generic adjustment sources has been stopped, as the current data answers the question, although the exact dividend amount remains a separate unknown.
The automated trading fund manager analyzed the equity at 02UTC with a price of 999.38. The analysis was based on three premises: first, 3768 unchanged adjusted series and 1331 synchronized returns were refreshed; second, fixed equal slots were constrained by a 260 fund and 249.845 position cap; third, stress assumptions included a maximum gap of 10% or 2 times the absolute historical worst gap, plus 0.5% friction. The common number of slots, N, was calculated to be 86.66, and the risk ceiling of 125.52531 was found to be nonbinding. The source of the data was https://runway.adityaperswal.workers.dev/data/stock-bars. The historical worst gaps were -6.256806%, -5.793196%, and -2.639964%, and the stress gaps were 12.513612%, 11.586393%, and 10%. The per lot stress was 13.792109%, 13.489460%, and 11.106454%, which is below 19.9876. The joint stress was 38.388023%, which is 3.841% of the equity, but this is not a guaranteed loss maximum. The total instrument loss of 86.66 exceeds 2%, so the gaps are unbounded. The basket volatility was 16.089109%, the worst day was -7.087094%, and the correlations were 0.238532, 0.389867, and 0.241683. The selected sample for today's hindsight excludes extinct instruments and is not an unbiased strategy performance. There is no new information to justify repeated simulation, and the prior regime and block sensitivity remains 19.76-50.20% loss frequencies with static brackets. The specific equal-weight mandate supersedes generic Kelly guidance, and the target gross profit is 43.002165 if all targets are reached, but this is not an expectation.
The premises of this update are as follows: the adjusted synchronous returns from June 2021 to September 2026 were obtained from https://runway.adityaperswal.workers.dev/data/stock-bars. The basket was selected today, so the historical equal-weight diagnostics are conditioned on this selection and do not represent a strategy backtest or an untouched holdout. The equity 1000 cap 260 dictates that each position is 86.66, assuming no tighter risk. A Python sample of 1331 showed an annualized daily volatility of 16.0891%, a worst basket day of -7.0871%, and pair correlations of 0.238532, 0.389867, and 0.241683. The selected ETF had worst negative open/prior close gaps of -6.256806%, -5.793196%, and -2.639964%. The gap size is calculated as g = max(0.10, 2 * absolute worst gap), and the risk is calculated as r = 2 * average true range / equity. The loss is calculated as N * (r + (1 - r) * g + 0.005). The common N is calculated as the floor of the minimum of 260/3, 250, and 19.99 / max loss fraction, which equals 86.66. The stress levels are 13.7921, 13.4895, and 11.1065, each less than 20, and the joint stress is 38.3880. Assuming a 0.5% friction and gaps of 12.5136%, 11.5864%, and 10%, arbitrary worse gaps can exceed 2% equity, so this is not guaranteed to be the maximum loss. The full premium is not at risk, as it would be for options, and the total loss of each ETF would cost 86.66. The positive sample correlation refutes the hedge guarantee. The target scenario basket gross profit of 43.0022 is neither an expectancy nor a probability forecast. A recorded 3650 Monte Carlo sensitivity analysis, which used 60,000 paths, static brackets, and recent versus older samples, was reused, as there was no new market day and the rounding/adjustment vintage did not supply independent evidence. The loss frequencies of 19.76% to 50.20% are not live odds or Kelly inputs, and there is no mandate for revalidation.
The analysis is based on three premises: using fresh split-normalized adjusted SIP daily bars from https://runway.adityaperswal.workers.dev/data/stock-bars, evaluating the current XLK/XLE/XLV equal basket, and estimating 1% gap thresholds from 2021-06-02 to 2025-12-31. The analysis found pair correlations before 2026 of 0.315635 for XLK/XLE, 0.500392 for XLK/XLV, and 0.293787 for XLE/XLV, and in 2026 of -0.247776, -0.158572, and -0.078192. The equal daily-return basket had an annualized volatility of 16.7368% versus 11.0104%, and the worst synchronized basket gap was -3.322667% versus -0.954476%. The analysis also found the following per-symbol threshold and test results: for XLK, the pre-2026 gap threshold was -2.5947445082801974%, with 2 test breaches out of 179, and a test breach rate of 1.1173184357541899%; for XLE, the pre-2026 gap threshold was -3.1902366689284984%, with 4 test breaches out of 179, and a test breach rate of 2.2346368715083798%; and for XLV, the pre-2026 gap threshold was -1.3698510638297818%, with 3 test breaches out of 179, and a test breach rate of 1.675977653631285%. The conclusion is that recent negative correlation reduces measured basket variance, but the XLE gap tail worsened, with the worst pre-2026 gap being -4.952927% versus -5.792734% in 2026. The decision is not to lower stress sizing from recent diversification. The validity of the analysis is based on direct historical comparison, but its soundness is limited by a small 179-day recent sample, current-winner selection, adjusted feed, and observed data vintage. This analysis is not an untouched strategy holdout and does not claim returns or win rates. All figures are stored in records 30442-43/30450.
A review of dividend data found 231 issuer events that reconcile with historical records. The review aimed to determine if missing dividends could change the top 3 or P90 targets. It used three premises: matching 1332 SIP bars per symbol through September 18, requesting all versus split data, and using primary SSGA historical distributions data. The data showed that cash payments before December 5 were halved only for five split ETFs. The review compared daily changes in the all/split ratio against the prior split close and found a maximum absolute error of 2.284555 basis points and a median error of 0.312859 basis points. It also compounded the close and dividend payments for explicit ex-date close reinvestment. The results showed small differences between the calculated means and the feed data for XLK, XLE, XLV, and XLF. Omitting dividends did not change the basket or SMA eligibility. The conclusion is that dividend omission is not a current issue, and all-adjusted signals and targets should be retained. The validity of the review is limited by the use of same feed prices, issuer revision risk, rounding, and backward adjustment versus close-reinvestment convention. The review is a dated reconciliation, not a permanent certification, and a strong counterexample could still corrupt the next run. The data collection includes all fixed 11 assets, excludes out-of-universe assets, and does not include dividends before the fetched price start.
The relationship between different sectors of the stock market has changed significantly over time. This is based on two premises: first, daily returns from September 18 were used, with some adjustments made to ensure the data was consistent. Second, the current selection of stocks was chosen after the fact, which may introduce some bias into the results. The analysis compared two time periods: a five-year sample with 1255 data points and a more recent sample with 179 data points. The results showed that the correlation between different sectors, such as technology (XLK), energy (XLE), and healthcare (XLV), was different in the two time periods. For example, in the five-year sample, the correlation between XLK and XLE was 0.241529, while in the recent sample it was -0.247776. The analysis also looked at the volatility of each sector, with results including an annual volatility of 0.2597661336235835 for XLK in the five-year sample and 0.2835588873548983 in the recent sample. The conclusion is that the recent negative correlations between sectors cannot be relied upon to continue, and that the volatility of a basket of stocks with equal daily returns is not the same as the actual profit and loss of an investment strategy. The analysis did not claim to have found a way to predict future results, and the 2026 data was not used as a separate test of the strategy after it had already been researched.
A question was asked about the stability of a specific financial measurement, called the 5-year rolling 60-session P90, which is a historical return measure. This measure is based on certain premises, including the use of adjusted data from a source called SIP, which is normalized, meaning it is adjusted to a standard scale. The data is split into sections, with 1,255 endpoints per ETF, and the normalization details are specified. The P90 measure is calculated for three ETFs, XLK, XLE, and XLV, and the results are 18.3622%, 20.6279%, and 10.6286% respectively. The correlations between adjacent windows of data are also calculated, with results of 0.97755, 0.97481, and 0.97676. The data is then divided into 60 non-overlapping sections, each containing 20-21 observations, and the P90 measure is calculated for each section. The results show that the P90 measure varies widely, from 12.983% to 27.164% for XLK, 12.393% to 24.969% for XLE, and 7.040% to 12.423% for XLV. The data is not independent, meaning that the sections are not completely separate from each other. When the data is divided into two chronological halves, the P90 measure is 17.105% and 22.554% for XLK, 22.269% and 18.233% for XLE, and 9.515% and 11.698% for XLV. It was found that 73 out of 126 upper-decile endpoints for XLE fell in 2022, and removing that year from the data lowers the P90 measure to 17.5194%. The conclusion is that the historical quantile, a measure of the range of possible returns, is reproducible, but the estimated precision is weak and dependent on the specific regime, or market conditions. One counterargument is that the width of the interval is due to the deliberate overlapping of the estimator and the small disjoint samples, but this does not necessarily invalidate the fixed rotation premium, a measure of the excess return of a portfolio. The validity of the results is limited by the dependence of the data, which reduces the unique information, but does not automatically introduce systematic bias. The soundness of the results is also limited by the single 5-year vintage, the ex-post current leaders, the split/dividend methodology, and the unobserved future regimes. A news scan found the headline that no new strategy backtest, out-of-sample edge claim, Kelly odds, or target override is made. The sources stop changing this descriptive answer after chronology, offsets, leave-year-out, and block-length checks. A new independent total-return index or corrected corporate-action feed could alter the exact numbers, but the targets would be retained by a fixed rule.
The relationship between selected stocks in a basket has not changed. The analysis is based on adjusted stock prices through September 18, and the current basket was chosen after the fact. The daily return statistics are used to describe the stocks, not as a strategy for investing. A correlation, which measures how closely stocks move together, was calculated over 60 and 252-day periods. The results show that the stocks tend to move in opposite directions, with correlations ranging from -0.46584 to 1. However, even with this negative correlation, the worst simultaneous daily return for the basket was -7.0913% and the 1st percentile was -2.7520%. This means that having stocks with negative correlations does not necessarily reduce the risk of large losses, known as gap stress. The analysis is limited because it only looks at past data, which may not reflect future market conditions. Additionally, the 1331 observations used in the analysis are not independent of each other, but rather are dependent on the specific market conditions at the time. Since there are no live trades, it is not possible to estimate how much the actual performance of the strategy would differ from the expected performance, known as slippage, or the percentage of winning trades, known as the win rate.
The manager of an automated trading fund made some observations about the recent performance of a selected basket of stocks. The manager looked at the daily returns of the stocks in the basket, with each stock having an equal weight. The manager also calculated the correlations between the stocks, which is a measure of how much they tend to move together. The correlations were negative over the past 60 and 252 days, meaning that when one stock went down, the others tended to go up. However, the manager found that on the worst historical day for the basket, the correlations turned positive, meaning that all the stocks went down together. This challenges the idea that diversification can always protect against losses. The manager also found that the worst historical day for the basket was a loss of 7.0913%, which would translate to a loss of about $22.70 at the current exposure level of $320.10. The manager concludes that increasing the risk size of the portfolio in an attempt to diversify it would not be a good idea. The manager also notes that the leadership of the stocks in the basket has been confirmed, but the breadth of the market has deteriorated, and would consider changing the initial candidates if the eligibility criteria changed, but not because of any change in the overall market narrative.
A news scan found the report from https://runway.adityaperswal.workers.dev/data/stock-bars. This report looks at how well a trading model works by checking its performance over two time periods: June 2021 to December 2024 with 902 gaps per ETF, and January 2025 to September 17, 2026, with 428 gaps per ETF. The report uses something called a retrospective chronological split, which means it looks back at how the model performed over these two time periods. It also uses a freeze pre-2025 stress, which is a way to measure the maximum possible loss, set at 10% for XLE and XLV, and 12.51854% for XLK. The report found that in the later time period, there were no breaches, or times when the model's performance was worse than the set stress level, for any of the ETFs. However, the worst performance for XLE increased from 4.952927% to 5.792734%, for XLV it increased from 1.975372% to 2.644903%, and for XLK it decreased from 6.259270% to 4.795771%. The full figures are: XLE had 902 training samples and 428 test samples, with a minimum loss of -0.049529267294310286 in the training period and -0.0579273396952954 in the test period, and no breaches; XLK had 902 training samples and 428 test samples, with a minimum loss of -0.06259270246217741 in the training period and -0.04795771263815474 in the test period, and no breaches; XLV had 902 training samples and 428 test samples, with a minimum loss of -0.01975371985633656 in the training period and -0.02644902894279455 in the test period, and no breaches. The report notes that the model only looks at gaps between the open and previous adjusted close, so it does not account for intraday price movements or unexpected events. The ETFs were selected based on current data, and the report concludes that just because there were no observed breaches, it does not mean that the model will always perform well. This diagnostic is not a guarantee of future performance, and the model should maintain its established stress levels and equal weights.
A source from https://runway.adityaperswal.workers.dev/data/stock-bars found the following daily correlations between sectors from June 2021 to September 17, 2026: energy and technology had a correlation of 0.238174, energy and health had a correlation of 0.241805, and technology and health had a correlation of 0.389758. All of these correlations are positive, which means that sector diversification does not remove exposure to the overall equity market. P90 targets refer to the historical 60-session quantiles with overlapping samples, and do not represent a 90% success probability. The gross target to risk ratios at the last close were approximately 5.17, 5.51, and 3.96. Since there is no trade history, the realized win rate, slippage, and conditional expected payoff are unknown. The published premium is accepted as per the mandate, and the correlations and bootstrap only describe exposure.