Stats

A news scan found the following information about funding and coin open interest. The first premise is based on 8-hour settlements on the OKX exchange from August 11 to September 13 at 00UTC, focusing on the latest Bitcoin and Solana prices. The Bitcoin price is at the 35.35 percentile with a funding rate of 0.497035 basis points, and the Solana price is at the 55.56 percentile with a funding rate of 0.276108 basis points. The Ethereum funding rate is 1 basis point, which is the same as the prior 95th percentile, and 23 out of 99 prior instances also had a funding rate of 1 basis point. The strict percentile for Ethereum is 76.77, compared to a weak 100. The second premise looks at the relative open interest of coins on September 12 at 08UTC, with Bitcoin at 27,484.289, an increase of 0.120866%, Ethereum at 639,751.978, an increase of 0.919325%, and Solana at 2,741,317.07, a decrease of 1.54566%. These quantities do not reveal the direction of long or short positions. The third premise examines the funding sums over a period, which are 61.9642, 62.3439, and 22.0552 basis points. A positive funding rate is a cost to swap longs, not a profit earned by cash spots. There is a strong bear case for Ethereum, as higher open interest and positive funding could indicate leverage crowding that amplifies downturns. However, this could also be due to arbitrage hedges, and the tied 1 basis point is common, with no indication of a global extreme or squeeze probability. The conclusion is to reject funding-only entry and instead report the tie interval and preserve the actual price trigger. The sources of this information are the OKX API for funding rate history and open interest. The raw observations are from numbers 26,531 to 26,535.

The fund manager has made some observations about the stock market. Firstly, they have looked at how different stocks have moved together recently, which is called correlation. They found that the correlation between EEM and QQQ is 0.830609, between EEM and IWM is 0.692751, and between QQQ and IWM is 0.737515, based on the last 252 days of daily returns. This means that these stocks have been moving together a lot lately. The manager also looked at the gaps in stock prices at the opening of trading, which is a risk because it can be hard to execute trades at the desired price. They found that the worst gap in EEM was a drop of 5.739837%, and the worst gap that would affect the fund's one share position was a drop of 3.893906, which is 0.457574% of the account. This is worse than the nominal stop of 1.91. The manager also looked at the returns of the stocks over the last five years and found that the top 10% of returns were over 10.717963%, the median return was 2.401325%, and the bottom 10% of returns were -8.174912%. Based on this, the manager thinks that a target of 75.11 is reasonable, but notes that this does not mean that it will be hit 90% of the time. The manager also notes that overnight shocks can be worse than the worst gap in the sample, and that exiting trades during the day can be different from the opening gaps. The manager has decided to keep one share and not to validate a new strategy or add any new positions.

The premises are as follows: the equity is 850.99 and the fund is 289.34, which is higher than the previous value of 221.26. The sizes are 96.44, 96.45, and 96.45. The retained split-adjusted worst gaps from prior to September 11, 2016, are 9.457211%, 9.301937%, and 11.623512%, all of which occurred on March 16, 2020. The loss is calculated using the formula loss = N * (r + (1 - r) * 2g + 0.005), where r is 2 times the average true range divided by the close. A news scan found the following data: the symbol QQQ has a size of 96.44, a normal value of 2.1229602366830784, a gain of 15.593948094606814, an rr of 7.3453792610693736, a stress of 20.44468315542464, a breach gap of 0.2430583044256341, and an sma distance pct of 8.437307785361249. The symbol IWM has a size of 96.45, a normal value of 2.0502761345148643, a gain of 12.368598158700099, an rr of 6.032649919922496, a stress of 20.094531818797652, a breach gap of 0.243614842541859, and an sma distance pct of 5.756535927075612. The symbol EFA has a size of 96.45, a normal value of 1.5925006841611973, a gain of 9.288407846549198, an rr of 5.832592688298649, a stress of 24.12629631591408, a breach gap of 0.24726510275948646, and an sma distance pct of 5.986340973057924. The joint stress is 64.6655, which is 7.5989% of the equity, and the ordinary stress is 5.7657. The gross all-target is 37.251 over 1-3 months, which is not the expectation. The new 252 daily return correlations are as follows: QQQ has a correlation of 1 with itself, 0.7375146484818859 with IWM, and 0.7068026873334806 with EFA. IWM has a correlation of 1 with itself, 0.7375146484818859 with QQQ, and 0.7494056074136088 with EFA. EFA has a correlation of 1 with itself, 0.7068026873334806 with QQQ, and 0.7494056074136088 with IWM. A larger cap increase of 30.77% does not increase the evidence. The per lot stressed losses are below 25.5297, but the EFA margin is only 1.4034. A gap beyond the stop is needed for a 3% loss, which would be between 24.31 and 24.73. A full loss of 96.45 exceeds 3%, so there is no hard guarantee. The hard fund and 25% position caps pass. The source of this data is https://runway.adityaperswal.workers.dev/data/stock-bars, which retained the prior risk history and has a new adjusted correlation window.

The automated trading fund manager reported the following results: 1. The mean return of 31 trades was 1.4456%, with a 95% confidence interval of -2.3253% to 5.6540% based on 20,000 bootstrap samples. The holdout set of 16 trades had a mean return of 3.50445%, with a confidence interval of -1.93285% to 9.08670%. Although the confidence intervals cross zero, the positive geometric results suggest that the strategy is worth watching. 2. A comparison of daily and intraday results over 1429 sessions found a median absolute close difference of 2.2852 basis points, with the 90th percentile at 11.2613 and the 99th percentile at 18.5117. The largest difference of 379.10 basis points occurred on March 8, 2022, which had incomplete data. Three training sessions were missing 25 samples each, but the actual next-day stop remained in place. 3. The fund's own record shows only two winning trades, with a total return of $4.977287 and a mean return of 2.278710%. The source of the data is https://runway.adityaperswal.workers.dev, and the bar data is from https://runway.adityaperswal.workers.dev/data/stock-bars. A news scan found the headline was not available, but the prior moves were adjusted from a daily watch to a positive raw sampled watch. New implementation losses or changes to paid net rates could lower confidence, and dividend accounting and event-specific economics are still unresolved.

A fund manager made changes to the C91208 ETF boundary control to preserve a weak relationship between flow and return. The manager corrected the one-row train boundary by excluding the response that crossed from December 31, 2025, to January 1. The correlation was -0.026123675 for N496. The mean of the next NAV was 0.184456816% for positive flow and 0.186261395% for nonpositive flow. The correlation for 2026 N173 was 0.038952704, which remained unchanged. The manager fixed counterexample dates, but this only affects sample boundaries, not real-time publication availability or executable prices and costs. As a result, no new directional forecast or trading rule was adopted, and no further tuning was done. The source of the data is the iShares Historical workbook, available at https://www.ishares.com/us/products/333011/ishares-bitcoin-trust-etf, specifically in the Data Download section, observations 26467-68.

A test was done on the stock TCOM, looking at nine times when the company AMC released its earnings from February 2024 to June 2026. The test checked the daily prices from the close of the release day to the close of the next day. The median absolute change was 5.543964% and the mean change was -0.908716%. On September 11, the stock price was $39.02, and on October 16, the call option price was $45.30 and the put option price was $35.40. The test also assumed a $0.05 per share reserve for buying and selling. The results showed that the put option was not a good choice, with only 1 out of 9 times being successful, which is 11.11%. This is below the 40% threshold. Without the reserve, the put option was successful 2 out of 9 times. The test also looked at the expected payout and found that there was no documented case where the actual payout was more than three times the premium. The test used data from option contracts and stock bars. The sources used were option_contracts and https://runway.adityaperswal.workers.dev/data/stock-bars. The test also corrected for a missing quarter and included May 19, 2025, in the results. The old forecast was retained separately. The premises of the test were to reproduce the results exactly in the new SIP bars, and the test confirmed the May 19, 2025, AMC earnings with a next-day return of -5.543964%. The accessible-date sample is now 9, not the full 10 quarter population. The test also looked at the schedule for the second quarter and first half of 2024 and found that the actual PDF confirmed the date but not the first hour. The test kept the +8.564650% next-day sensitivity and did not silently assign AMC. The test also used NumPy to do a bootstrap analysis with a seed of 91208 and found a mean 95% CI of [-6.041939, 4.510271]%. The test found that the price change versus the sample change was separated, and the current costed put was 1/8 on the old population, and adding the May quarter made it 1/9. The cost reserve alone reduced the current put 2/9 gross to 1/9 costed. The test also found that the February 2025 quarter missed the 11.455664% hurdle by -11.387292%. The collection of data used Python requests, which failed twice, and then used Node fetch, which returned 200/926 bars with no pagination. The test also used Yahoo 5y OHLC to provide a separate check, and found that the raw historical levels differed from the SIP adjusted prices, with a maximum overlapping return difference of 0.10619 pp. The final study used SIP throughout, and no event return was selected from Yahoo. The validity of the test was limited by the timing, population, and costs, and none of these established an entry point.

A news scan found the headline that the C91208 IBIT share flow does not exhibit a stable one-day predictive relation. The analysis used a type of statistical model called DescriptiveOLS to examine the relationship between the change in shares and the next day's return on investment, also known as NAV return, which is the return on the net asset value of the investment. The model looked at data from 2024 to 2025, which included 497 observations, and found a correlation of -0.026228 and a beta of -0.001625. This means that when the change in shares was positive, the next day's return was on average 0.192300% higher than when the change was not positive, which had an average return of 0.186261%. However, when the model looked at data from 2026, which included 173 observations, it found a correlation of 0.038953 and a beta of 0.328987. In this case, when the change in shares was positive, the next day's return was on average -0.025226% lower than when the change was not positive, which had an average return of -0.065636%. The analysis also noted that there were two missing historical share values, which were recorded as not available, or NA, rather than being treated as zero. The results suggest that the relationship between share flow and next day's return is not stable and that acting on aggregate IBIT flow alone or selecting a threshold from the known 2026 data may not be reliable. The data for this analysis came from the iShares Bitcoin Trust ETF, which can be found at https://www.ishares.com/us/products/333011/ishares-bitcoin-trust-etf.

A news scan found the headline 'C91208 hourly weekend risk and dependence: lower typical volatility is not a stop guarantee'. This is a report about an analysis of cryptocurrency data from January 1, 2026, to September 12, 2026. The analysis used hourly data from a venue, which is a place where financial transactions are made, and removed one hour of data per coin, resulting in 6101 valid returns for each coin. The actual closing prices at 23UTC, which is a time of day in Coordinated Universal Time, were used to calculate daily returns, without filling in any missing data. The standard deviation, which is a measure of how much something varies, of the hourly log returns, which is a way of measuring the change in price, for Bitcoin, Ethereum, and Solana were.468478%,.622116%, and.671698%, respectively. On weekends, these numbers were lower, at.331830%,.490827%, and.589622%, compared to weekdays, which were.512833%,.667232%, and.701739%. Even though the typical variance, which is a measure of how much something varies, is lower on weekends, it can still happen with poor quote depth, which is a measure of how many buyers and sellers there are, and extreme gaps in prices. The simple hourly returns for Bitcoin, Ethereum, and Solana were -1.323930%, -1.918972%, and -1.940170%, respectively. The daily standard deviation was 2.434355%, 3.323474%, and 3.337775%, respectively. The correlation, which is a measure of how much two things are related, between Bitcoin and Ethereum was.888000, between Bitcoin and Solana was.837910, and between Ethereum and Solana was.849084. An analysis using a statistical method called ordinary least squares, or OLS, found that the beta, which is a measure of how much one thing is affected by another, of Ethereum and Solana with respect to Bitcoin was 1.179221 and 1.201385, respectively, with an intercept, which is a starting point, of -.000204% and -.000914% per hour, respectively. However, this analysis did not take into account any uncertainty, and did not claim to have found any alpha, which is a measure of how well an investment performs. The report concludes that major cryptocurrencies are substantially one risk exposure, meaning that they are all affected by the same things, and that neither lower weekend variance nor cash pressure, which is the need to have cash on hand, warrants a tight stop, which is a rule to sell an investment when it reaches a certain price. The report also states that there is no current entry, meaning that there is no current investment or trade being made. The source of the data is https://runway.adityaperswal.workers.dev/data/crypto-bars.

A news scan found the headline is not applicable here, instead we have a report from an automated trading fund manager. The manager looked at the funding settlements for Bitcoin, Ethereum, and Solana on the OKX exchange from August 10 to September 12. The funding settlement is the interest paid or received by traders, a concept also known as the cost of carrying a position, which is a fee for holding a trade overnight. The manager found that the latest funding settlements were not extreme compared to the previous 99 settlements. For Bitcoin, the latest settlement was 0.855375 basis points, which is a unit of measurement equal to 0.01%, with a median of 0.646500 basis points and a 95th percentile of 1.0 basis point, meaning 69.69697 percent of the settlements were below this level. For Ethereum, the numbers were 0.463076, 0.679862, and 1.0 basis points, with 36.363636 percent of settlements below this level. For Solana, the numbers were 0.871952, 0.121511, and 1.0 basis points, with 75.757576 percent of settlements below this level. The total funding settlements for the period were 62.640955, 62.307298, and 23.329550 basis points, which is 0.626410, 0.623073, and 0.233296 percent of the total notional value, a measure of the total value of the assets held. The manager also looked at the open interest, which is the total number of outstanding contracts, for each coin and found that it was 27451.11 BTC, 633924.129 ETH, and 2784353.63 SOL. The manager concluded that there is no signal to reverse the funding, and will continue to collect data. The data came from the OKX exchange's public APIs for funding rate history and open interest.

The EEM target distribution audit is based on three premises: it uses total returns over 60 sessions, the endpoint window covers the last 5 years, and overlapping observations are not independent trades. A Systematic Investment Plan adjustment was made, and data from June 1, 2021, to September 11, 2026, shows 1255 endpoint returns since September 12, 2021. The returns are as follows: the 90th percentile is 10.717963%, the median is 2.401325%, and the 10th percentile is -8.174912%. The returns were negative 38.247012% of the time, and the range was from -20.291480 to 26.007905%. The target was calculated as 67.84 * (1 + 0.107179635), which equals 75.111066, rounding to 75.11. The initial warmup period in July missed 10 endpoint returns, resulting in a 10.744101% return, but this was corrected before the final signal recording, which agrees with 2837. Note that the 90th percentile, or P90, represents the target magnitude, not a 90% chance of gain or hitting the target. There is no new strategy backtest or edge gate, and empirical fund trade odds are not available due to zero closed lots.

The fund has increased its capital by 18.18%. The current equity is $850.99 with a capital of $221.26, which replaces the previous capital of $187.22. The fund is divided into three equal parts of $73.75, $73.75, and $73.76. The worst split-adjusted gaps from 2016 to September 11 are retained for the stocks QQQ, IWM, and EFA, with percentages of 9.457211%, 9.301937%, and 11.623512% respectively, all as of March 16, 2020. A scenario is used to calculate the stress, where L equals N times the current rate plus 0.005, and the current rate is the current average true range divided by the close price. The assets in the fund are QQQ, IWM, and EFA, with sizes of $73.75, $73.75, and $73.76 respectively. The joint stress is 5.81% of the equity, and the gap beyond the stop needed to lose 3% of the account is $32.63, $32.69, and $33.01. The correlations between the assets range from 0.779 to 0.817, which means they are not independent. The gross target is $28.49, and the ordinary stop is $4.41. The source of the data is https://runway.adityaperswal.workers.dev/data/stock-bars. A prior simulation from 2905 is retained, but it is not a live win odds or Kelly, and no new backtest is required due to the fixed mandate.

A study of major coins found that their dependence on each other strengthened in 2026. The study, which used daily returns from 2024 and 2025 as a training period and returns up to September 11, 2026, as a holdout period, found the following correlations: ETH/BTC correlation increased from 0.790720 to 0.911085, SOL/BTC correlation increased from 0.769173 to 0.878901, and ETH/SOL correlation increased from 0.767210 to 0.879675. The study used a statistical method called ordinary least squares, or OLS, which is a way to find the best fit line for a set of data. The results showed that the return on ETH and SOL could be predicted by the return on BTC, with the following betas: ETH train beta 1.156279, holdout beta 1.252137; SOL train beta 1.452620, holdout beta 1.205607. A bootstrap analysis, which is a way to test the accuracy of the results, found the following confidence intervals for the betas: ETH [1.160094, 1.380073], SOL [1.117952, 1.328144]. The daily intercept, which is a measure of the average return, was 0.011985% for ETH and 0.000076% for SOL, but these results were not statistically significant. The annualized volatility, which is a measure of risk, was 63.913068% for ETH and 63.791445% for SOL. The study notes that the correlations and betas found are recent associations and not stable crash hedges or causal predictions. The study concludes that BTC, ETH, and SOL must share a systemic risk budget, rather than being independent investments, which supports a joint simulation from 2029-2030, but is not a signal to enter the market. The data for this study came from https://runway.adityaperswal.workers.dev/data/crypto-bars.

The automated trading fund manager has analyzed the outcome uncertainty and own-record calibration. The own supply record shows two operator-liquidation winners with a combined stock profit and loss of 4.977287 and a mean of 2.278710%. However, with 100% wins, it is not possible to estimate average losses or execution advantage. A naive exact 95% binomial lower bound is already 15.8114%, and correlated shipping lots further weaken independence, according to the source venue query trades fund, which can be found at https://runway.adityaperswal.workers.dev. The fund manager also analyzed the FRO fixed training, which had 33 closed trades with a mean of 1.87277%, and a 95% confidence interval of -1.97206 to 6.13580. The test period had 17 trades with a mean of 4.65443% and a confidence interval of -1.35232 to 10.72225. The managed holdout had 18 trades with a mean of 2.97157% and a confidence interval of -2.24030 to 8.46648. The STNG fixed test had 11 trades with a mean of -0.34261% and a confidence interval of -5.19766 to 5.84822. The collection was completed with daily SIP, including prior 2020 warmup, 2021-24 training, and 2025 to September 11 held out for this rule. The test period was already seen for other strategies, and the selected current tankers create a selection bias, not a universe-wide edge. The unfinished trades mark equity but do not enter the closed outcome confidence interval. These distributions and the current correlation of 0.777079 are descriptive, not causal, according to the source https://runway.adityaperswal.workers.dev/data/stock-bars. The prior was shifted from an old 2R inconclusive watch to a provisional 3R positive FRO watch, but not to statistical certainty. A costed prospective 15-minute raw cash implementation passing unchanged rules would strengthen it, but weak cash or normalizing net rates or prospective execution losses would weaken it. No precise Bayesian posterior is claimed.

An automated trading fund manager performed an edge test on ten issuer-dated scheduled AMC earnings from March 2024 to June 2026. The test found that the daily pre-release close to next-session close had a median absolute percentage change of 4.700378% and a signed mean of -3.027709%. On September 11, the price of S2.53 and the cost of an October 16 call option of $0.21 and a put option of $0.16, plus assumed roundtrip costs of $0.05 per share, required a percentage change of +9.090909% or -9.486166% to be profitable. However, both of these percentages were below 40%. The manager also reviewed old frozen data from N9, which had a median percentage change of 5.494505%, and decided to preserve it separately. The manager found that the issuer, High Tide Inc, announced that it would release its third fiscal quarter 2026 financial results after financial markets close on Monday, September 14, 2026. The manager also reviewed prior observations from N9, which skipped September 2025 without a recovered exclusion rationale. The manager added a new observation, N10, and retained the frozen N9 forecast. The manager also reviewed training data from 2024-25 and holdout data from 2026, and found that the results did not meet the mandatory gate, resulting in a rejection. The manager noted that the sample size was small and non-stationary, and that the data was limited to US listing data and stale put marks. The manager also noted that there was no population-wide calendar claim, and that the premise of the test was that residual time value could make options profitable, but there was no priced or executable evidence to quantify this exception.

The fund allocation has increased by 29.41% from 144.67 to 187.22, with equity remaining unchanged at 850.99. A proposed allocation of 62.40/62.41/62.41 is being considered. The crisis-inclusive period from 2016 to September 11, 2026, has been analyzed, with 2687 split-adjusted daily opens and prior closes. A loss scenario has been modeled, including a 2ATR stop, multiplicative adverse gap, and assumed 50bp friction. The formula used is L=N*[r+(1-r)*g+.005], where r is 2ATR/close. The worst gaps were found to be QQQ 9.457211%, IWM 9.301937%, and EFA 11.623512%, all on March 16, 2020. Gap correlations were QQQ/IWM 0.811520, QQQ/EFA 0.778810, and IWM/EFA 0.816941. Using DOUBLE, each worst gap gives 13.228414/13.002589/15.611427 dollars, which is 1.554474/1.527937/1.834502% equity each below the 25.5297 risk threshold. The joint total is 41.842430, which is 4.916912% equity. A common 20% extra beyond stop+50bp loses 41.364711, which is 4.860775% equity, and a common 10% loses 23.015787. Ordinary stops sum to 3.730764, and all initial targets yield 24.103442 gross, which is not the expected payoff. To reach a 3% per lot requires 39.071750/39.112107/39.405948% extra gap beyond stop under this equation. A total loss of 62.41 lot is 7.333811% equity, so there is no guaranteed 3% bound. An unconditional per-lot total-loss cap would need to be <=25.52 before fees. The conclusion is that the current equal lots pass explicit severe historical scenarios, but diversification does not imply a 3% bound. This is stress sizing, not a trade-edge backtest or probability forecast, and full-loss possibility and pre-2016/sample omissions remain. The data source is https://runway.adityaperswal.workers.dev/data/stock-bars?timeframe=1Day&feed=sip&adjustment=split&start=2016-01-01 and user venue/account context on September 12, 06:01 UTC.

A news scan found the headline about C91204 uncertainty. The automated trading fund manager analyzed ETH hourly training with 13 closed trades, resulting in 7 wins, with a mean return of 4.123471%. The manager used a method called bootstrap resampling 10,000 times to estimate the range of possible mean returns, which was between -3.768357% and 12.803958%. Another method, called Wilson confidence interval, estimated the range of possible win rates to be between 29.1434% and 76.7942%. The manager also tested the system on a holdout set of 3 trades, which all won, with a mean return of 6.157475%, and a Wilson confidence interval of 43.8494% to 100%. However, the manager noted that even with these wins, it is not 100% certain that future trades will win. The manager also found that the average win was 16.101921% and the average loss was 9.851387%. The manager used a formula called the Kelly criterion to determine the optimal amount of capital to allocate, which was 1.299744 times the capital, but noted that this implies an impossible level of precision. The manager also analyzed the daily ETH training data, which had a mean return of -2.732972%, but the holdout set had a return of 0.024738% per trade, which compounded to a marked return of -0.292829%. The manager attributed this to something called volatility drag, which is not a contradiction. The manager also analyzed the correlation between BTC and SOL, and found that they have mixed and correlated market exposures, and are not independent samples. The manager used a statistical method called OLS to analyze the relationship between open interest and price, and found a correlation of 0.113948, and a beta of 0.069650, but did not claim to have found a causal relationship or significant prediction. The manager based the analysis on three premises: that the returned edge comes from a small few tails, that the same 2026 history is already known and many cells have been observed, and that frequencies are not replicates. The manager concluded that the system is a candidate for watching, but not for precise probability or blanket late-entry permission. The manager preserved all 200 cells, including coherence 8, and did not optimize on the holdout set. The sources of the data were https://runway.adityaperswal.workers.dev/data/crypto-bars and https://www.okx.com/api/v5/rubik/stat/contracts/open-interest-history?instId=ETH-USDT-SWAP&period=1H&limit=100.

A test of Nokia's earnings from January 2024 to July 2026 found that the median absolute change in the stock price on the day of the earnings release was 5.864627%. The range of changes was from -8.474903% to 11.444479%. For example, on September 11, the stock price was $11.13, and to reach the target price of $13.45 on October 30, the stock would need to increase by 21.293801%. However, this did not happen, and the option to buy the stock at $13.45 on October 30 was not profitable. A news scan found the headline 'Nokia Corporation lowers 2025 operating profit guidance due to currency' on the Nokia website. The test also found that the stock price did not move as expected after the earnings release, and the options to buy or sell the stock were not profitable. The test used data from the Nokia website, option contracts, and stock prices. The test was done on 11 earnings releases from 2024 to 2026, and the results were compared to the median change in the stock price. The test found that the stock price did not move as expected after the earnings release, and the options to buy or sell the stock were not profitable. The sources used for the test include the Nokia website, option contracts, and stock prices.

A fund manager analyzed the stock FTK. The manager used daily stock prices from September 11 and earlier, and also used prices for the stocks SPY and ACDC. The manager found that on September 9, FTK's price changed by 14.402954% while SPY's price changed by -0.464776%. This difference of 15.659472 percentage points is a residual, which is a measure of how much a stock's price changed compared to what was expected. The manager also looked at how closely the prices of FTK and other stocks were related over the last 60 days, and found correlations of 0.406972 with ACDC and 0.320440 with SPY. The manager then looked at the results of some trades, but these trades were not reliable for estimating losses. The manager also tested a strategy using FTK's prices over the last 20 days, and found that it would have lost an average of 6.461891% per trade. The manager decided not to use this strategy because it did not seem to work well.

A question is raised about whether a fixed allocation of roughly $48.22 is still acceptable when partial exits are delayed until regular trading hours. This is based on several premises: the current equity is $850.99, with 3% being $25.53; losses can increase rapidly when stop-loss orders are triggered; the selected exchange-traded funds (ETFs), which are a type of investment fund that trades on a stock exchange like a stock, share risks associated with the overall stock market; and historical data may not fully capture potential future shocks. Data from a source, which is a website providing financial information, shows the daily performance of three ETFs, QQQ, IWM, and EFA, over a one-day period with adjustments made for stock splits. The data vendor uses a specific method, known as SIP daily open/prior close-1, split-only adjustments, to preserve price gaps that occur when a stock goes ex-dividend, meaning it no longer includes the right to the next dividend payment. Over a five-year period from September 2021 to September 2026, the worst losses were -5.3571% for QQQ, -5.2747% for IWM, and -4.5015% for EFA. The worst losses for QQQ and IWM occurred on August 5, 2024, and for EFA on March 3, 2026. Analyzing Monday data, the worst losses were -5.3571%, -5.2747%, and -3.5974%. The correlations between the daily gaps in prices were.8245,.7278, and.7597, and the total return correlations were.7784,.7349, and.7545. The worst joint opening loss at the proposed weights was $6.8617 on August 5, 2024. Expanding the data to January 2016 to September 2026, the worst losses for all three ETFs occurred on March 16, 2020, with QQQ losing -9.457211%, IWM losing -9.301937%, and EFA losing -11.623512%. The joint loss was $14.651681, with correlations of.81152,.77881, and.81694. This expanded data changes which ETF had the largest loss in the tail of the distribution, indicating that relying only on the five-year data may be overly optimistic. A chronological check of the tail risks from 2016 to 2024 and from 2025 to 2026 shows different cutoffs and breaches. This analysis was not frozen in time and does not claim to have a pristine holdout sample. The selection of today's ETFs is conditioned on their past performance and survival, and the historical data lacks extremes like those seen in 2008. No inference is made about the probability of future compliance with these risk bounds. The fact that all three ETFs had their worst loss on the same date is a counterexample to the assumption of independence, and the earlier 2020 observations exceeded the five-year maximum for EFA, showing that historical maximums may not be limits. The conclusion is to retain fixed ranks, use risk estimates that include crisis scenarios, and explicitly disclose the potential for common shocks, without re-testing momentum.

The analysis of the EEM overnight performance is based on four premises: the momentum signal remains fixed, the daily opening gap is measured, the prior 20-day true range is known before opening, and historical gap frequency does not determine stopped-trade loss probability. According to data collected from September 12, 2021, to September 11, 2026, the raw gaps were -2.730322% on the first percentile, -1.652163% on the fifth percentile, and a median of +0.034083%, with a minimum of -5.739837% on March 3, 2026. The data shows that 11 out of 1255 times, the open was below the prior close by more than 2 prior raw Average True Ranges (ATR), which is a measure of price volatility. The analysis also found that Mondays did not have a significant effect, with only 3 out of 234 events showing a specific pattern. The preferred mandate alignment, which is a set of rules for making investment decisions, was met in 6 out of 812 cases, with a success rate of 0.738916%. Two additional price breaches on December 20, 2023, and December 17, 2024, matched the issuer's ex-distributions, which refers to the distribution of profits or dividends. The analysis concludes that a counterexample to a hard loss cap, which is a maximum amount of loss allowed, was observed, and there is no reason to cancel a valid long position, which is a type of investment where the investor expects the price to rise. The observations from September 23 to September 41, out of a total of 25966, preserved the full distributions of the data.

A news scan found the following analysis: The analysis of daily log returns from January 1, 2026, to September 11, 2026, for eight liquid volatile coins, including BTC, ETH, SOL, HYPE, and UNI, gave 254 observations per major coin, with no imputation for moment and correlation estimates, which is a statistical method to handle missing data. The annualized volatility, a measure of how much the price of a coin fluctuates, was 46.46% for BTC, 63.38% for ETH, 63.68% for SOL, 86.10% for HYPE, and 82.30% for UNI. The correlation, which measures how much two coins move together, between ETH and BTC was 0.91299, between SOL and BTC was 0.88227, and between XRP and BTC was 0.87507. The correlation between HYPE and BTC was 0.52600, and between UNI and BTC was 0.67039, which reduces but does not remove common risk. The daily 1st percentile loss, a measure of the worst daily loss, was -5.82% for BTC and -8.59% for ETH. The excess kurtosis, a measure of how fat the tails of the distribution are, was 7.55 for BTC and 5.17 for ETH, which warns that normal-distribution tail models are inadequate. The historical forward 20-day maximum high gain 90th percentile, a measure of the best 20-day gain, was 20.51% for BTC, 35.06% for ETH, 32.06% for SOL, 63.48% for HYPE, and 56.44% for UNI. The conclusion is that the three major coins are mostly one bet, and it is recommended to budget aggregate risk, use wide measured stops, and accept low target touch odds over a two-day weekend. These are descriptive ranges, not fair value or probabilities conditional on a signal. Different listing history and regimes limit comparison. The source of the data is https://runway.adityaperswal.workers.dev/data/crypto-bars, which is a website that provides crypto data. The full 8 rows of data were saved as observations, but no orders were placed.

The automated trading fund manager has reviewed the performance of the CAG23 actual issuer earnings from 2021 to July 2026. The Primary T5/ATR1/3 model shows a training win rate of 52.941%, with an average win of 1.325795% and an average loss of 1.995905%. However, the held-out data shows only 6 wins and 4 losses, with an average win of 0.559145% and an average loss of 3.051251%, resulting in a mean loss of 0.644320%. The bootstrap 95% mean confidence interval is -2.094866% to 0.587018%. The model also used a simple indicator OLS, which stands for Ordinary Least Squares, a statistical method, to analyze the return on investment. The results show a correlation between momentum and SMA, or Simple Moving Average, of 0.781325, and between momentum and return of 0.069265, with an R-squared value of 0.063501. The coefficients for the intercept, momentum, and SMA distance are -0.00418587, -0.07371741, and 0.26760662, respectively. However, the related signals are not independent votes, and the 17 data points with 3 parameters are very unstable. The fund's actual trades show losses for MIRM, IONS, and RARE, with a mean loss of 16.380601% and no winning trades. Two operator liquidations have censored the intended horizon, and the Wilson 95% upper win bound is 56.1506%. The RARE fill price is 13.98976, which is 41.7093% below the stop price, resulting in an actual loss of 6.44035 relative to the initial risk. The sources of the data include the query_trades fund catalyst and the stock bars data from the SIP feed. Prior evidence suggests that the calendar alone is unlikely to be enough, and new independent costed cohort mean positive with defensible win/loss and dated entry would be needed to change the allocation. The financial shortfall is not an evidence update.

A test of Edge found that the current contracts for CAG and LW are not acceptable. For CAG, the test looked at the median absolute change in the stock price on mornings when the company released its earnings, which was 2.064410% from July 2024 to July 2026. The test also looked at specific prices, such as 14.60 on September 11 and 15.5 on October 9, and found that the current trade of 0.20 on September 11 would require a change of 8.904110% or 7.876712% including a cost of 0.05. For LW, the test found a median change of 13.153864% from July 2024 to July 2026, with specific prices such as 47.85 on October 16 and 57.5 on the same day. The test also found that the current premium was more than 51.0594 times the absolute cap. The mean mapped intrinsic value of LW was 1.498726 USD per share, which is 2.724957 times the current premium and 2.497877 times the assumed all-in cost. However, this is not the actual option expectancy. The test did not find any 40% pass or documented 3x expected payoff. The test omitted October 1, 2024, due to ambiguity about the release hour, but adding the reaction on October 2 still yielded a failure. The dates for the events were: CAG on August 31, with a morning Q&A on September 30, 2026, before 9:30 ET, and LW on September 8, with a release approximately at 8 ET and a call at 9 ET on October 6, 2026. The expiries were 9 or 10 days after the events, with default exits on September 29 or October 5 RTH. The historical originals were recorded as 25518+ and 25563+, with sources including option contracts and stock bars. The closes were not executable asks. The test looked at two things: daily signed stock reactions mapped to current strike and premium, and price and turnover screens. However, the price and turnover screen is only for discovery and is biased towards current listings, with no independent probability estimates. The test found that the most recent 2026 N2 for CAG was both negative and less than 1.26%, with no put cost coverage, and that the historical tail in October 2024 does not guarantee a repeat. For LW, the test found that the N8 vs 9 sensitivity both failed 40%, with a correlation of 0.985243 between four shared release dates. The validity of the test requires a gate, but the observed gates failed, so the test rejects the current contracts. The soundness of the test is limited by the closes, corporate adjustments, small samples, and LW timing. A counterexample is that residual time could make real options worth more than their intrinsic value, but without actual net option history, this does not document mispricing.

A news scan found the source https://runway.adityaperswal.workers.dev/data/stock-bars with a system for adjusting stock investments, known as SIP, from June 2021 to September 11, 2026. The system uses premises, or assumptions, that certain signals, 2781 of them, are accepted, and that correlation, which measures how much things move together, describes concentration, not a new selection gate, which is a rule for choosing investments. The daily return correlations, which measure how much the daily returns of two things move together, for EEM, or emerging markets, and QQQ, or US technology stocks, is 0.830609, for EEM and IWM, or US small-cap stocks, is 0.692751, and for QQQ and IWM is 0.737515. This means that investing in these three types of stocks would not be like making three separate bets. The rolling 60-session returns, which measure the returns over a 60-session period, for EEM in a five-year window, known as N1255, were negative 8.174912 percent at the 10th percentile, 2.401325 percent at the median, and 10.717963 percent at the 90th percentile, with 61.593625 percent of the returns being positive. The windows overlap, meaning that some sessions are counted multiple times, and the positivity rate is not the same as the win rate for stopped trades. Using the chosen 90th percentile return, the yield would be 7.271066 dollars from 67.84 dollars, with two ATR, or average true range, which measures the range of an investment's price, risk of 1.90528, and a ratio of 3.816272 before rounding and costs. A 10 percent gap, or sudden change, in the price of one share would result in a loss of 6.784 dollars, or 0.797189 percent of the account equity, compared to a nominal stop, which is a stop-loss order, of 0.223890 percent equity. The conclusion is to keep fixed equal slots and hard caps, and not to use discretionary leverage, which is borrowing money to invest, or extra correlated positions, which are investments that move together. The trend has completed zero trades, so empirical fund expectancy, which is the expected return of the fund, and slippage, which is the difference between the expected and actual returns, are not available.

A news scan found the headline is not available, but a data analysis was performed. The analysis used a source from https://runway.adityaperswal.workers.dev/data/stock-bars. It applied a statistical method called Ordinary Least Squares, or OLS, to daily returns of HWM and SPY over 60 sessions ending on September 4 before September 8. The results showed a beta of 0.867538 and an alpha of -0.005349% per day. On September 8, HWM had a return of -10.69927%, SPY had a return of -0.549215%, and the residual was -10.21746 percentage points. This residual does not seem to be caused by a specific event called GEdeal or a forecast of a subsequent reversal. Additionally, the analysis found that the returns of FRO and STNG over the last 60 days had a correlation of 0.778, and the returns of LPG and FRO had a correlation of 0.690, indicating that shipping positions share similar risks. A query of trades found that the actual returns of FRO and STNG were 3.80516% and 0.75226%, respectively, with a mean return of 2.27871% and a win rate of 100%. However, this assumes that these positions are independent, which they are not. Since there are no losing trades in the sample, it is not possible to identify the average loss or edge. The original targets were 2.61446R and 2.15126R, but these are below the current 3R policy, and the positions have already been closed. In conclusion, the analysis did not find any evidence of a causal relationship between the event and the returns, and the 100% win rate is likely due to chance rather than any underlying edge. The analysis is based on records 25267 and two queried lots.