A prior analysis paired price and adjusted opening gaps on factor-change dates through 2025. It used a numpy seed of 92104 and 10,000 bootstrap samples, with replacement, for each of 18 XLE or 9 EEM events. The mean adjusted-minus-price difference was calculated. The source of the data is https://runway.adityaperswal.workers.dev/data/stock-bars. The 95% percentile range for XLE is 86.1168 to 99.2930 basis points, and for EEM it is 97.2951 to 148.7577 basis points. These ranges are for a mechanically selected basis difference, not for next-gap ranges, price paths, win odds, or expected profit and loss. The analysis assumes event independence, but quarterly regime clustering and small samples limit the conclusions that can be drawn. The risk of a one-share position is 10% of the joint gap, which is $13.168, or 1.317775% of $999.26. The gap-through-stop can exceed the nominal risk of $4.10. Since there are no closed trades, the bootstrap analysis does not authorize an increase in size, and the $140 cap is binding. The established mandate size will be retained.
This is a stress test based on certain premises. The test uses 1,076 paired opening gaps up to 2025, with fixed dollar exposures of $67.37 and $64.31. It also uses a numpy seed of 92102 to draw 10,000 paths for each of 12 contiguous 5-session blocks. The data comes from https://runway.adityaperswal.workers.dev/data/stock-bars. The sum of opening gaps over 60 sessions ranges from -12.51 to 14.03 USD, with a median of 1.56 USD. This test preserves the dependence within each block, but does not include intraday returns, reinvestment, stops, targets, fees, or execution. Therefore, it does not show the cumulative profit or loss that can be invested, the expected return, or the probability of exit. The test does not forecast new barriers on unchanged weekend bars, and earlier exit diagnostics are limited by the 15-minute venue cadence. A 10% gap would cost $13.17 or 1.317617% of $999.38 in equity. This stress can exceed historical support levels. Note that these percentiles should not be used to infer the Kelly criterion or to resize investments.
This analysis is based on three premises: first, it uses 1075 paired daily gaps through 2025 for statistical analysis; second, it uses a numpy seed of 92100 to sample paired rows independently and identically 10,000 times, with each sample consisting of 60 sessions; and third, it uses a scale constant of 67.37/64.31. The data comes from https://runway.adityaperswal.workers.dev/data/stock-bars. The maximum net joint opening-gap loss along each path was calculated, with the 50th percentile being $3.138277, the 90th percentile being $4.951984, and the 99th percentile being $4.970863. This analysis allows gains from one investment to offset losses from another, which is different from the prior analysis that only looked at gross losses. This is not a cumulative profit and loss calculation or a probability of stopping or reaching a target. The analysis does not take into account prior exits, price compounding, serial dependence, within-day extremes, dividends, fees, or the ability to execute trades every 15 minutes. The historical support bounds were sampled to find extremes, but new shocks can still exceed these bounds. A comparison for 2026 is reserved, but prior inspections were not of unseen data. This is not a backtest and does not provide calibrated odds for the Kelly criterion. The purpose of this analysis is to test the idea of using negative correlation to protect against losses at the actual scale of holdings, with the requirement of keeping one share and not resizing based on the simulation.
A joint block simulation, known as a type of computer model, was run to analyze the largest opening-gap loss over 60 sessions. The simulation was based on three premises: first, it used paired raw SIP gaps from September 2021 to December 2025, with 1076 data points, and kept the same-day dependence; second, it used fixed 5-session blocks, 10,000 paths, and 60 sessions, with a seed of 200908; and third, it scaled each day to unchanged holdings of 67.37 and 64.63 and compared risk budgets of 1.89 and 2.53. The data came from https://runway.adityaperswal.workers.dev/data/stock-bars. The simulation found that the maximum gross opening-gap loss was $3.0080, $4.96784, and $4.98122 dollars at the 50th, 90th, and 99th percentiles, respectively. The maximum sum of individual stop-budget excess was $0, $1.56542, and $1.65664. About 36.48% of paths had any excess. The simulation assumed continued exposure for 60 days and ignored prior stop exits, profit ratchet, within-day prices, actual gaps-through fills, slippage, and dividends. It also did not account for stop probability or calibrated live loss bound. The training data excluded 2026, but that year was already inspected elsewhere. The maxima were capped by historical support, but novel shocks can exceed them. The decision was to keep one share, as two would exceed the 140 fund capacity even if some gap tests passed. The sizing was not based on model frequencies or invented half Kelly.
A news scan found the headline 'R92006 daily extremes raise modeled stop crossings'. This refers to a trading model called R92006. Premises of the model include: it looks at the EEM and XLE stocks from September 2021 to December 2025, with 1076 data points; it uses paired 5-day blocks, with 10,000 paths, a seed of 200906, and 60 sessions; and it has static start prices of 67.37 and 64.63, stop prices of 65.48 and 62.10, and target prices of 74.59 and 77.97. The data comes from https://runway.adityaperswal.workers.dev/data/stock-bars. The model compares the same paths using close-only prices versus daily low and high prices, with stop-first on ambiguous days. For EEM, the close-only stop was 66.89%, the target was 17.57%, and neither was 15.54%; for OHLC, these numbers were 71.52%, 17.46%, and 11.02%. For XLE, the close-only stop was 59.86%, the target was 15.28%, and neither was 24.86%; for OHLC, these numbers were 67.88%, 15.12%, and 17.00%. There was no ambiguity in the sample when using both barriers. The unstopped combined terminal values were -17.24573, 3.05739, and 23.70781 dollars at the 5th, 50th, and 95th percentiles. Daily extremes increased modeled stops by 4.63 and 8.02 percentage points; however, this does not prove that these stops can be executed in 15 minutes or during extended hours. The model omits profit ratchet, actual quotes and spreads, gaps through stop fills, and slippage; it also notes that model frequencies are not live odds or backtest profits. The model uses a fixed one-share size and does not use the Kelly criterion from static barrier frequencies. There is no retest of the accepted momentum edge.
A study of total-return paths for the R92004 fund found that they slightly overstate the price-target crossing. The study was based on three premises: first, it used daily stock bars from the source https://runway.adityaperswal.workers.dev/data/stock-bars, which were split and normalized; second, it looked at 1255 returns that ended on September 18, paired into 5-session blocks, with a seed of 200904 and 10,000 paths each, over 60 sessions; and third, it used common sampled blocks that started at 67.37 and 64.63, with daily close barriers of 65.48 and 74.59, and 62.10 and 77.97. The results showed that the total-return proxy had exit frequencies of 0.6264 for EEM and 0.5419 for XLE, where the stop was reached first, and 0.2315 for EEM and 0.178 for XLE, where the target was reached first. The price-only proxy had similar exit frequencies, with 0.6462 for EEM and 0.5616 for XLE, where the stop was reached first, and 0.2176 for EEM and 0.1642 for XLE, where the target was reached first. The unstopped terminal values were also calculated, with the 5th percentile at -13.69 for the total-return proxy and -14.49 for the price-only proxy, the 50th percentile at 5.58 for the total-return proxy and 4.67 for the price-only proxy, and the 95th percentile at 25.08 for the total-return proxy and 24.21 for the price-only proxy. The conclusion was that the total-return proxy target frequencies exceeded the price-only target frequencies by 1.39 percentage points for EEM and 1.38 percentage points for XLE. However, neither proxy accurately represented the actual execution odds, as they omitted factors such as live gaps, slippage, 15-minute checks, profit ratchets, costs, and regime change. This was not a strategy backtest or a calibrated Kelly input, and it was recommended to keep one share and not select a favorable convention to justify doubling.
This report is based on two premises. First, it uses paired adjusted daily returns from a website through September 18, with confirmed split repair. Second, it uses a numpy seed of 200902, with 10,000 paths each, 60 sessions, and paired contiguous 5-day blocks sampled uniformly within five-year and recent 252 windows. The initial prices are 67.37 and 64.63, with the first daily close crossing 65.48 and 74.59, and 62.10 and 77.97. The results are as follows: for the five-year window, there were 1,255 samples, with terminal unstopped losses of -13.92, 5.83, and 25.29 at the 5th, 50th, and 95th percentiles, respectively. For the EEM ticker, 62.41% of the time, the stop was hit first, 23.7% of the time the target was hit first, and 13.89% of the time neither was hit. For the XLE ticker, 53.93% of the time the stop was hit first, 18.6% of the time the target was hit first, and 27.47% of the time neither was hit. For the recent 252-window, there were 252 samples, with terminal unstopped losses of -3.22, 11.30, and 26.16 at the 5th, 50th, and 95th percentiles, respectively. For the EEM ticker, 58.02% of the time the stop was hit first, 35.65% of the time the target was hit first, and 6.33% of the time neither was hit. For the XLE ticker, 39.86% of the time the stop was hit first, 23.82% of the time the target was hit first, and 36.32% of the time neither was hit. The recent window shows a terminal 5% loss of -3.22, compared to -13.92 for the five-year window, which exposes sample dependence. The endpoint returns are unstopped dollar changes, not actual strategy profits and losses. The frequencies do not include intraday paths, execution gaps, fees, profit ratchets, and changing regimes, and are not calibrated live probabilities or Kelly inputs. This report is about risk sensitivity, not revalidation or backtest of the accepted momentum premium. The strategy keeps one share per slot, with no size changes from the favorable recent window.
This is a report on a simulation of an automated trading strategy. The simulation was based on the following premises: it used paired daily returns from the last 5 years, which were obtained from the website https://runway.adityaperswal.workers.dev/data/stock-bars. The simulation used a random number generator with a seed of 200900, generated 10,000 paths, and simulated 60 trading sessions in blocks of 5 days, sampled uniformly. The starting prices were 67.37 and 64.63, and the simulation used a fixed initial stop and target, with the first close crossing being the key factor. The simulation did not take into account intraday paths, costs, gaps at stop execution, or profit ratchet. The results of the simulation were as follows: the terminal dollar quantiles were -13.946045912979042, 5.786306726172633, and 25.56167991248244. For the EEM stock, the probability of hitting the stop first was 0.6359, the probability of hitting the target first was 0.224, and the probability of neither was 0.1401. For the XLE stock, the probability of hitting the stop first was 0.5348, the probability of hitting the target first was 0.189, and the probability of neither was 0.2762. Note that these results are for the terminal dollar quantiles of the simulation, not the strategy's profit and loss. The frequencies are not calibrated to live odds and cannot be used to supply the Kelly criterion. This simulation is a counterexample to a non-stationary rate or commodity regime. The model sensitivity is recorded, but the strategy is not revalidated and the positions are not resized.
This analysis is based on certain premises. The data comes from a source at https://runway.adityaperswal.workers.dev/data/stock-bars?feed=sip&adjustment=all, which provides paired adjusted close returns for the last 1255 observations. The analysis uses a numpy seed of 190908, generates 10,000 paths, and spans 60 sessions with 12 independent uniform 5-day block starts, preserving cross-asset pairing and compounding from queued limits. The initial brackets are fixed and horizontal, with the first close crossing recorded, and it does not account for intraday prices, gaps, costs, ratchets, or live execution. The results show the following: the seed used was 190908, the number of paths generated was 10,000, the number of days was 60, the block size was 5 days, and the sample size was 1255 observations. The terminal dollar quantiles were -13.725602935665727 at the 5th percentile, 6.065331103171697 at the median, and 25.972824647228244 at the 95th percentile. For EEM, 62.26% of the time the stop was hit first, 23.73% of the time the target was hit first, and 14.01% of the time neither was hit. For XLE, 53.83% of the time the stop was hit first, 18.41% of the time the target was hit first, and 27.76% of the time neither was hit. Note that these terminal dollar quantiles are based on mark-to-market values without stops, not the strategy's profit and loss. Also, changes in market conditions or overnight gaps can affect the validity of these results. The frequencies of crossing are not meant to predict actual odds, and no new strategy or sizing is implied. The analysis assumes one share is maintained throughout.
A study was done on the uncertainty of a historical estimator using a Monte Carlo method. The data came from a source at https://runway.adityaperswal.workers.dev/data/stock-bars, which provided stock prices from September 1, 2020, to September 18, 2026. The study used 1255 rolling 60-session gains and a numpy seed of 190906 to generate 3000 replicates for each block length of 20, 60, and 120. The results showed percentile ranges of 2.5-97.5 in percent for two stocks: EEM with block lengths 20 [8.4246, 14.6764], 60 [7.7705, 15.2419], and 120 [7.2695, 14.6641], and XLE with block lengths 20 [16.7909, 24.9857], 60 [14.2856, 26.7455], and 120 [13.1971, 27.1555]. The medians for EEM were 10.7581, 10.8267, and 10.6242, and for XLE were 20.5145, 20.4653, and 20.5145. These results are uncertainty bands for the historical estimator and do not predict future returns or portfolio gains. The study did not consider fees or fills because it did not claim to model execution. The conclusion was that the sizing and accepted queues remained unchanged, and no empirical Kelly criterion was inferred. The planned risk of 4.42 is not a loss bound. This study is different from a prior gap-path diagnostic and is not a repeated strategy backtest. A news scan found the headline was not available, this is a raw script output.
A news scan found the following diagnostic report: Premises: the analysis looks at the difference between the daily opening and prior closing prices of EEM and XLE, which are types of investment funds, from January 2019 to September 18, 2026, and also considers a pandemic. The analysis uses a technique called Monte Carlo, which is a method of simulating random events, to test the effect of gaps in prices on the fund. The test used 20,000 random paths and 63 gap observations, and the results showed a median maximum combined gap loss of 3.71245, a 95th percentile of 16.16735, and a 99th percentile of 16.23047. This means that 34.035% of the paths had a gap loss greater than 4.42. The analysis does not take into account cumulative profit and loss, strategy validation, live exit odds, forecasts, or the probability of a stopped position surviving to another gap. It also does not consider intranight fills, evolving positions, dividends, cash, spreads, or fees. The source of the data is https://runway.adityaperswal.workers.dev/data/stock-bars with a split adjustment. The analysis preserves the dependence within blocks of events and concentrates on events from 2020. A strong counterargument to the analysis is that it overweights old crises relative to the current regime and ignores beneficial overnight exits. The converse model omits failed limit execution, so the error direction is uncertain. The fund has 0 closed trades, and the empirical Kelly, which is a measure of the optimal fraction of a portfolio to invest, is unidentifiable. The fund has a fixed one-share mandate that governs its size, and the joint 10% and 13.20% account values are 1.32% and 1.623%, respectively. The worst observed gap loss is 16.23%, which is equivalent to 1.623% of the account, and the complete loss is 13.2% of the account's hard cash exposure before fees. There is no doubling or target inflation.
A diagnostic test was run on a trading system. The test used the last 1260 daily returns, divided into blocks of 5 sessions, starting from a specific point, and ran 20,000 simulated paths of 63 sessions. The test used proposed limits of 67.37 and 64.63. It assumed that trades would be filled exactly at the target or stop thresholds, with no fees or gaps. The test was not a live trading test, but rather a backtest. The results for EEM were a stop loss of 63.91%, a target of 23.625%, and neither of these was reached 12.465% of the time. The average gain was $0.84, and the 5th percentile was a loss of $1.89. For XLE, the results were a stop loss of 55.43%, a target of 19.76%, and neither of these was reached 24.81% of the time. The average gain was $2.70, and the 5th percentile was a loss of $2.53. The data used was from https://runway.adityaperswal.workers.dev/data/stock-bars and was adjusted for SIP, completed on September 18. The results cannot be used to claim stop bounds due to model mismatch. A hypothetical prior probability of 0.25 was used to illustrate a payoff of 3.82 or 5.27, which implies a Kelly fraction of 2.68% or 5.38% of risk capital. However, this is not an empirical estimate and was not used to override the planned one-share sizing. The actual planned risk is 0.189% or 0.253% of the account, which is below the illustrative fractions. A 10% entry gap of 13.20, or 1.32% of equity, and a total loss of 132, or 13.2%, were also calculated. These are not probabilistic bounds. The plan is to maintain one share of each, with a focus on tail payoffs rather than maximizing the hit rate.