Notes

The fund, which is for crypto research only, was observed on 2026-09-10T16:32:40.067891+00:00. The data source is https://runway.adityaperswal.workers.dev/data/crypto-snapshots and https://blockstream.info/liquid/api/blocks. The fund's status is as follows: BTC had a bid price of 76911.5 and an ask price of 76932.1, with an age of 40, and was considered okay with a depth in USD of 77.1/77.7 and a hurdle percentage of 0.529. ETH had a bid price of 2437.19 and an ask price of 2437.875, with an age of 137, and was considered okay with a depth in USD of 243.3/80.7 and a hurdle percentage of 0.530. SOL had a bid price of 99.06 and an ask price of 99.1672, with an age of 92, and was considered okay with a depth in USD of 93.8/94.3 and a hurdle percentage of 0.611. USDC had a bid price of 0.9995 and an ask price of 1.00001, with an age of 3678, and was not considered okay with a depth in USD of 9995000.0/10000100.0 and a hurdle percentage of 0.553. USDG had a bid price of 0.99999 and an ask price of 1.00001, with an age of 45638, and was not considered okay with a depth in USD of 9999900.0/10000100.0 and a hurdle percentage of 0.504. XRP had a bid price of 1.3469102 and an ask price of 1.3493, with an age of 92, and was considered okay with a depth in USD of 131.1/5626.5 and a hurdle percentage of 0.680. UNI had a bid price of 5.935 and an ask price of 5.94809, with an age of 53, and was considered okay with a depth in USD of 4189.5/4185.7 and a hurdle percentage of 0.724. LTC had a bid price of 51.718 and an ask price of 52.02, with an age of 69, and was considered okay with a depth in USD of 3532.3/3523.4 and a hurdle percentage of 1.089. HYPE had a bid price of 79.08 and an ask price of 79.48, with an age of 47, and was considered okay with a depth in USD of 61381.1/61104.2 and a hurdle percentage of 1.010. LINK had a bid price of 11.53 and an ask price of 11.55, with an age of 54, and was considered okay with a depth in USD of 13359.5/13398.0 and a hurdle percentage of 0.676. There were 8 out of 10 fresh quotes. The recipient BTC was 598.50044115, with a mempool net of 4.62e-05 and a transaction number of 292. The reserve BTC was 3601.47203772, with a mempool net of 0.0 and a transaction number of 590. The recent page was 31, controlled was 0, and not full history was 1. The tip was 4051484, with an age in seconds of 90 and a time of 1789057870. There was an error with a connect timeout. The operator notes were 2, and the operator hash was 1fa338f6b403063d815b51327749e5cafdd00164b7e7f0e4f2a7f8568410238b, with a claim about the recent incident affecting the Liquid Network and the movement of funds. The fund had 7 out of 7 panels attempted, with 0 eligible, and the balances were not sales or recovery. https://runway.adityaperswal.workers.dev/data/crypto-snapshots ; https://blockstream.info/liquid/api/blocks ; https://blockstream.info/api/ ; https://x.com/Blockstream

A Python audit was conducted on frozen cash-basis payoffs. The results for Bitcoin (BTC) showed a profit of 2.7120158% and a loss of 4.1207999%, with an overall performance of -1.2737933% and a Kelly criterion of -0.11742127. For Ethereum (ETH), the results were a profit of 2.5091341% and a loss of 5.1951198%, with an overall performance of -1.7710069% and a Kelly criterion of -0.17645598. For Solana (SOL), the results were a profit of 3.8428040% and a loss of 5.3172454%, with an overall performance of -0.7372207% and a Kelly criterion of -0.04796112. The audit used data from January 1, 2026, to September 9, 2026, with 25 basis point fees and 10 basis point slippage. There were no fresh tests, and only one forced liquidation of a Bitcoin position was observed. Six quote receipts were collected between 14:01 and 16:01, with lengths of 1436, 1445, 1443, 1442, 1454, and 1454, and gaps of 29.90325, 15.05653, 15.08778, 30.00755, and 30.02392 minutes. The job was completed at 16:31:30, with a tip of 4051483 and no changes to the operator hash or balances. A total of 185 tasks were done after 16:01, but 184 were active and need replacement due to a hard cutoff at 16:31. The scheduled 15-minute coverage was not observed. https://runway.adityaperswal.workers.dev/data/crypto-bars ; backtests2183–2185 ; crypto notes1972,1973,1976,1978,1981,1985 ; query_trades fund=crypto

A news scan found the following data: the August final-demand Producer Price Index, which is a measure of the average change in prices of goods and services, increased by 0.4% from the previous month and 5.4% from the same month last year. Excluding food and energy, the increase was 0.3% from the previous month and 4.7% from the same month last year. The price of energy increased by 4.2% and services by 0.1%. On September 5, the number of initial claims for unemployment benefits was 206,000, compared to a revised 207,000 the previous week. The number of insured unemployed people on August 29 was 1,774,000. This data was retrieved between 16:31 and 16:33 UTC and did not change from the morning's verified baseline. The data is preliminary and may be revised. A headline from a news scan found the data from agency releases dated September 10 at 08:30 ET. https://www.bls.gov/news.release/ppi.nr0.htm ; https://www.dol.gov/ui/data.pdf

A news scan found the following data: from 12:45 to 13:45, the OKX coin open interest changed as follows: BTC from 28335.215 to 29439.974, an increase of 3.8988905 percent, ETH from 613651.653 to 636463.563, an increase of 3.7174038 percent, and SOL from 2712042.38 to 2727481.03, an increase of 0.5692629 percent. This data refers to the total amount of outstanding contracts, known as open interest, in the market. The Onevenue gross contracts, which represent the total number of contracts, did not show any long net flows, meaning there were no significant increases in buying or selling, or global leverage, which refers to the use of borrowed money to invest. There were no extreme percentile changes, which would indicate unusual market activity. A Circle issue from September 7 showed a 7-day redemption of 8.80-7.86, totaling 0.94 billion, and a 30-day redemption of 36.59-34.37, totaling 2.22 billion. There were no changes to the dated stock and no new buying on September 10. The saved quote notes had 1949, 1950, and 1955 characters, with 1433, 1432, and 1450 of them being numeric. The receipt from 12:46:27 to 13:01:44 to 13:31:36 had gaps of around 15 and 30 minutes. The funding from 1947 to 1957 was complete at 12:45 and 13:45. A job, which is a scheduled task, numbered 181, was due to run at 13:45, but was overdue at 13:51, and there was no actual quote receipt at 13:45 yet. Local replacement quotes, which are used to update existing quotes, were around 1452 characters, and the funding was 1281 characters, both complete. The daily volume of 14 pairs, which are combinations of currencies or assets, includes USDG, a type of stablecoin, over 3 and 7 days, but the ranking is not based on global liquidity, which refers to the ability to buy or sell an asset quickly. https://www.okx.com/api/v5/public/open-interest ; https://www.circle.com/transparency ; crypto notes1949/1950/1955/1947/1957

A Python program recalculated two values: the payoff and the quarter Kelly criterion. The results for Bitcoin, or BTC, were: 12 trades, a 41.6667% probability of winning, a 2.7120158% win size, a 4.1207999% loss size, and a mean return of -1.2737933%. The raw quarter Kelly value was -0.1174213. For Ethereum, or ETH, the results were: 9 trades, a 44.4444% probability of winning, a 2.5091341% win size, a 5.1951198% loss size, and a mean return of -1.7710069%. The Kelly value was -0.1764560. For Solana, or SOL, the results were: 8 trades, a 50% probability of winning, a 3.8428040% win size, a 5.3172454% loss size, and a mean return of -0.7372207%. The Kelly value was -0.0479611. All values were clamped to 0. This audit reviewed a frozen reused 2026 holdout, not a new experiment. The only trade in the history of Bitcoin was a trade that was closed by the operator, with a price return of -0.0809902% and a hold time of 47.9278 minutes. It was not possible to estimate the win or loss from this single discretionary exit. There were no open, pending, or closing trades from the query. https://runway.adityaperswal.workers.dev/data/crypto-bars ; recorded backtests2183–85 ; query_trades fund=crypto

The fund is for crypto research only and was observed on 2026-09-10T13:52:35.448492+00:00. The source of the information is https://www.okx.com/api/v5/public/time, which has an offset of 0.1162 seconds and an uncertainty of 0.1903 seconds. Another source, https://www.okx.com/api/v5/public/funding-rate and https://www.okx.com/api/v5/public/open-interest, provided the following data: for Bitcoin, the funding rate is 1.0 basis points and the settled rate is 0.7581 basis points, with an open interest of 29510.278 coins. For Ethereum, the funding rate is 0.2143 basis points and the settled rate is 0.3562 basis points, with an open interest of 629725.456 coins. For Solana, the funding rate is 0.0715 basis points and the settled rate is -0.3279 basis points, with an open interest of 2726278.38 coins. A news scan found the source https://www.circle.com/transparency, which reported the following data as of September 07, 2026: the balance is 74.28, with issue and redeem amounts of 8.8, 7.86, 36.59, 34.37, 333.52, and 331.64 for different time periods. The source https://www.usdc.com/ could not be parsed. The source https://stablecoins.llama.fi/stablecoin/2 reported a raw CMS value with a circulating amount of 74386006033.1783 and breakdowns for Ethereum, Solana, and Hyperliquid L1. The collection ended with no eligible funding and no buying flow. https://www.okx.com/api/v5/public/funding-rate ; https://www.okx.com/api/v5/public/open-interest ; https://www.circle.com/transparency ; https://stablecoins.llama.fi/stablecoin/2

A fund for crypto research was observed on 2026-09-10T13:53:16.456843+00:00. The fund got data from https://runway.adityaperswal.workers.dev/data/crypto-snapshots and https://blockstream.info/liquid/api/blocks. The fund looked at the following cryptocurrencies: BTC with a buy price of 77008.18 and a sell price of 77021.65, ETH with a buy price of 2428.8 and a sell price of 2429.61, SOL with a buy price of 99.5 and a sell price of 99.5635, USDC with a buy price of 0.9995 and a sell price of 1.00001, USDG with a buy price of 0.99999 and a sell price of 1.00001, XRP with a buy price of 1.3608 and a sell price of 1.366, UNI with a buy price of 5.904 and a sell price of 5.9152, LTC with a buy price of 52.0543 and a sell price of 52.3707, HYPE with a buy price of 80.92 and a sell price of 81.33, and LINK with a buy price of 11.6656 and a sell price of 11.69. The fund also looked at the age of each cryptocurrency, which is the time since the last update, and the depth in USD, which is the amount of the cryptocurrency that can be bought or sold. The hurdle percentage, which is the minimum percentage change required for a trade, was also considered. The fund had 8 out of 10 fresh quotes. The recipient had 598.50044115 BTC and the reserve had 3601.47203772 BTC. The recent page was 31 and the tip was 4051325. There was an error with a connection timeout. The operator notes had 2 entries and the operator hash was 1fa338f6b403063d815b51327749e5cafdd00164b7e7f0e4f2a7f8568410238b. The operator claimed that following the recent incident affecting the Liquid Network and the movement of funds, Blockstream made a statement. The fund attempted to access 7 panels and was eligible for 0 of them, and the balances were not for sales or recovery. https://runway.adityaperswal.workers.dev/data/crypto-snapshots ; https://blockstream.info/liquid/api/blocks ; https://x.com/Blockstream

A clock check at 13:52:44 UTC and record creation at 13:52:36-37 showed that the approximate retrieval times were misstated as 13:53/13:54. Both were actually fetched before 13:52:37, during the time period of 13:51-13:52 UTC. The publication and reference dates, as well as all numerical release values, remain unchanged. It is recommended to use the tool and record timestamps, rather than the inaccurate rounded prose times, which refer to written descriptions of times that are not precise. https://www.dol.gov/ui/data.pdf ; https://www.ecb.europa.eu/press/pr/date/2026/html/ecb.mp260910~314e508016.en.html

A news scan found the headline about the European Central Bank, or ECB, decision. On September 10 at 13:54 UTC, a request was made and a successful HTTP 200 response was received, but there was an internal error when trying to open the web page. The ECB decided to raise interest rates by 25 basis points, or bp, which will take effect on September 16, with a deposit rate of 2.50%, a main refinancing rate, or MRO, of 2.65%, and a marginal lending rate of 2.90%. The forecast for 2026, 2027, and 2028 is for headline inflation to be 3.0%, 2.5%, and 2.1%, and for inflation excluding food and energy to be 2.5%, 2.6%, and 2.3%. The forecast for economic growth is 0.9%, 1.4%, and 1.5% for the same years. The inflation forecast for 2027 and 2028 was revised up, and the growth forecast for 2026 and 2027 was also revised up. The Asset Purchase Programme, or APP, and the Pandemic Emergency Purchase Programme, or PEPP, will mature without being reinvested. The forecasts are scenarios created by the issuer, not actual outcomes that have been observed, and there is no commitment to follow a specific path. The decision confirms the previous decision, with a reference number of 22795, and there are no additional policy changes. https://www.ecb.europa.eu/press/pr/date/2026/html/ecb.mp260910~314e508016.en.html

The US Department of Labor released its claims report on September 10 at 8:30 ET, which was verified at 13:53 UTC. The initial number of seasonally adjusted claims for the week ending September 5 was 206,000, a decrease of 1,000 from the revised 207,000 in the previous week, which was initially reported as 206,000. The four-week average was 206,000, a decrease of 1,500 from the revised 207,500 in the previous week, which was initially reported as 207,250. The number of insured unemployed for the week ending August 29 was 1,774,000, a decrease of 1,000 from the revised 1,775,000, which was initially reported as 1,779,000, with an unemployment rate of 1.2%, which remained unchanged. The initial non-seasonally adjusted number was 176,567, an increase of 5,164. The initial 206,000 claims reported this time is the same as the previous headline, but it is for a new week, and the publication date and revisions have resolved the ambiguity caused by stale data. Note that benefit claims do not account for workers who are ineligible or not claiming benefits, and there is no estimate of the impact on the broader labor market or cryptocurrency. https://www.dol.gov/ui/data.pdf

A news scan found the headline about the August Producer Price Index (PPI) release on September 10. The US Bureau of Labor Statistics (BLS) released the report, which showed that the final demand for August increased by 0.4% from the previous month and 5.4% from the same time last year. The July and June readings were revised to 0.1% and -0.1%, respectively. Excluding food, energy, and trade, the increase was 0.3% from the previous month and 4.7% from the same time last year, compared to 0.4% in July. The report also showed that goods increased by 1.1%, energy by 4.2%, goods excluding food and energy by 0.4%, services by 0.1%, and transportation and warehousing by 2.3%, while trade decreased by 0.2%. The price of diesel increased by 24.1%, which accounts for over a third of the increase in goods. The readings for April to July were revised, but the exact changes are not clear without comparing them to the previous versions. This report replaces the earlier one, which was outdated. There is no verified consensus or expected payoff related to cryptocurrency. https://www.bls.gov/news.release/ppi.nr0.htm

On September 10 at 12:33:41 UTC, the primary tip, which is a suggested transaction, was updated to 4051246, with a transaction count of 1 and a unique identifier of 40795bffeab0f1c55d2700952300d6a9b48bae9509ff06cddf78f57524ed1171. This update was based on previous tips 4051245 and 4051244. Since the last snapshot at 12:32, two new blocks were added to the chain, 60 seconds apart. However, there was a timeout error when trying to connect to a secondary explorer, which is a tool used to track transactions. This update indicates progress on the primary chain, but it does not confirm the full health of the network, authorization of payouts, recovery of assets, or sale of Bitcoin. The recipient and reserve amounts remain unchanged at 22777, and the operator hash remains the same. https://blockstream.info/liquid/api/blocks ; https://liquid.network/api/blocks

A news scan found the following data line: Python recomputation of a quarter of the Kelly formula, which is a calculation used to determine the optimal amount to invest, based on hourly payoffs. The results are as follows: for Bitcoin, the percentage of positive outcomes is 41.6667%, the average win is 2.7120158%, the average loss is 4.1207999%, and there are 12 negative outcomes, resulting in a value of -0.1174213; for Ethereum, the percentage of positive outcomes is 44.4444%, the average win is 2.5091341%, the average loss is 5.1951198%, and there are 9 negative outcomes, resulting in a value of -0.1764560; for Solana, the percentage of positive outcomes is 50%, the average win is 3.8428040%, the average loss is 5.3172454%, and there are 8 negative outcomes, resulting in a value of -0.0479611. The allocation was clamped to long-only, meaning no short selling was allowed, and all positions were held. The 2026 data was reused from a holdout set, rather than being used for new independent tests. There was one actual operator liquidation for Bitcoin, but it was not set up as an edge, which refers to a competitive advantage. The fund has a cap of 235.86, with an account position of 212.7475 and a nominal risk of 8.5099, and cannot rescue ineligible edges. https://runway.adityaperswal.workers.dev/data/crypto-bars ; recorded backtests2183/2184/2185

A news scan found the headline about the European Central Bank's primary decision. The decision was made on September 10 and became available at 12:34 UTC via HTTP requests. The issuer's RSS feed had a publication date of September 10 at 14:15 +0200, which is equivalent to 12:15 UTC, resolving the ambiguity in the Central European Time label. The decision raises all three rates by 25 basis points, which is a unit of measurement for interest rates, effective September 16: the deposit rate to 2.50%, the main refinancing rate, also known as the MRO, to 2.65%, and the marginal lending rate to 2.90%. The staff headline reported inflation projections for 2026, 2027, and 2028 as 3.0%, 2.5%, and 2.1%, respectively, and excluding energy and food, the projections are 2.5%, 2.6%, and 2.3%. The growth projections are 0.9%, 1.4%, and 1.5% for the same years. The projections for 2027 and 2028 inflation were revised up, and the growth projections for 2026 and 2027 were also revised up. The issuer's projected numbers are not actual outcomes, and there is no precommitment path. This decision supersedes the previous unverified status of 22769 ECB, but gaps in data from the Bureau of Labor Statistics and the Department of Labor remain. This is a primary policy source and not independent causal evidence for crypto prices. https://www.ecb.europa.eu/press/pr/date/2026/html/ecb.mp260910~314e508016.en.html ; https://www.ecb.europa.eu/rss/press.html

The fund is for crypto research only and was observed on 2026-09-10T12:32:07.852909+00:00. A news scan found the source https://www.okx.com/api/v5/public/time with an offset of 0.6991 seconds and an uncertainty of 0.4599 seconds. Another source, https://www.okx.com/api/v5/public/funding-rate and https://www.okx.com/api/v5/public/open-interest, provided the following data: for Bitcoin, the funding rate was 1.0 basis point and the settled rate was 0.7581 basis point, with an open interest of 28111.536 coins. For Ethereum, the funding rate was 0.6693 basis point and the settled rate was 0.3562 basis point, with an open interest of 604460.589 coins. For Solana, the funding rate was 0.235 basis point and the settled rate was -0.3279 basis point, with an open interest of 2767714.56 coins. A website, https://www.circle.com/transparency, reported on September 07, 2026, that the issuer had 74.28 billion in circulation, with 8.8 billion issued and 7.86 billion redeemed in the last 7 days, and 36.59 billion issued and 34.37 billion redeemed in the last 30 days. The same website also reported 333.52 billion issued and 331.64 billion redeemed in the last 365 days. A news scan found the headline https://www.usdc.com/ with an unparsed raw cms value and unparsed issuer date. Another source, https://stablecoins.llama.fi/stablecoin/2, reported that on timestamp 1788998400, the circulating supply was 74386006033.1783, with 47146319984.45 on Ethereum, 7103929420.46 on Solana, and 6887999675.1 on Hyperliquid L1. The crypto collection has ended, with definitions and cutoffs differing, and the flow is not buying, resulting in 0 eligible funds. https://www.okx.com/api/v5/public/funding-rate ; https://www.okx.com/api/v5/public/open-interest ; https://www.circle.com/transparency ; https://stablecoins.llama.fi/stablecoin/2

A fund for crypto research made observations on September 10, 2026, at 12:32:48.920338 UTC. The data came from https://runway.adityaperswal.workers.dev/data/crypto-snapshots and https://blockstream.info/liquid/api/blocks. The fund looked at the following cryptocurrencies: BTC, which had a buy price of 77577.37 and a sell price of 77589.997, with an age of 40 and a depth in USD of 77.5/77.0, and a hurdle percentage of 0.518; ETH, which had a buy price of 2451.7 and a sell price of 2452.3, with an age of 41 and a depth in USD of 81.0/81.7, and a hurdle percentage of 0.526; SOL, which had a buy price of 100.922 and a sell price of 101.04, with an age of 40 and a depth in USD of 95.5/95.8, and a hurdle percentage of 0.619; USDC, which had a buy price of 1 and a sell price of 1.00001, with an age of 901 and a depth in USD of 26.7/10000100.0, and a hurdle percentage of 0.503; USDG, which had a buy price of 0.99999 and a sell price of 1.00001, with an age of 31247 and a depth in USD of 9999900.0/10000100.0, and a hurdle percentage of 0.504; XRP, which had a buy price of 1.372 and a sell price of 1.37806, with an age of 40 and a depth in USD of 5755.9/5737.0, and a hurdle percentage of 0.946; UNI, which had a buy price of 6.0112 and a sell price of 6.024, with an age of 41 and a depth in USD of 4221.5/4222.4, and a hurdle percentage of 0.716; LTC, which had a buy price of 52.208 and a sell price of 52.51, with an age of 43 and a depth in USD of 3526.5/3558.5, and a hurdle percentage of 1.083; HYPE, which had a buy price of 82.08 and a sell price of 82.47, with an age of 42 and a depth in USD of 62914.3/63395.5, and a hurdle percentage of 0.979; and LINK, which had a buy price of 11.764 and a sell price of 11.79, with an age of 43 and a depth in USD of 13514.4/13528.8, and a hurdle percentage of 0.724. The fund received 8 out of 10 fresh quotes. The recipient had 598.50043115 BTC, with a mempool net of 4.62e-05 and 291 transactions. The reserve had 3601.47203772 BTC, with a mempool net of 0.0 and 590 transactions. The recent page was 31, and it was not fully controlled, with a note that the history was not full. The tip was 4051244, with an age of 99 seconds and a timestamp of 1789043470. There was an error with a connection timeout. The operator had 2 notes and a hash of 1fa338f6b403063d815b51327749e5cafdd00164b7e7f0e4f2a7f8568410238b, and mentioned an incident affecting the Liquid Network and the movement of funds. The fund attempted 7 out of 7 panels and had 0 eligible, with balances but no sales or recovery. https://runway.adityaperswal.workers.dev/data/crypto-snapshots ; https://blockstream.info/liquid/api/blocks ; https://x.com/Blockstream

A news scan found the following information. On September 10, at 12:32 UTC, it was observed that the web BLS summary still showed the August 13 July 2026 release. The DOL pdf referenced weeks of August 29 and August 22 on September 3. The direct BLS 403 was included in the September 10 archive, but this does not prove a release delay. The ECB page listed times of 14:15 CET and 14:45 CET, but the rates from June 17 were 2.25, 2.40, and 2.65. The latest decision retrieved from the home page was from July 23. The primary decision and timestamp for today have not been verified. At 12:16 UTC, AP reported a 25 basis point hike to 2.50, but this is secondary information and has not been independently confirmed, so it will not be used as trade input. It is important not to confuse prior period values with the actual September 10 data. A calendar at https://www.bls.gov/schedule/2026/09_sched_list.htm confirms that the PPI will be released at 08:30 ET today and the CPI will be released tomorrow. https://www.bls.gov/news.release/ppi.nr0.htm ; https://www.dol.gov/ui/data.pdf ; https://www.ecb.europa.eu/press/press_conference/html/index.en.html

On September 10, a review of the system, called a readback, showed that jobs 178 and 179 had 9 quotes and 4 funding messages with numeric information from 08:15 to 12:01, all of which were marked as complete, and were between 1281 and 1464 characters long. The quotes were sent with a 15-minute gap at first, then about a 30-minute gap, while the funding messages were sent at about 60, 75, and 75 minutes apart. At 12:31, a note with 1936 characters was received, but it only contained 34 characters of actual data, and the scripts stopped running at 12:31 even though 8 out of 1 runs were still remaining. The system's stored numeric data does not confirm that the 15-minute execution was complete. The full audit is available in the C9101230 notes. query_research fund=crypto notes; list_jobs fund=crypto

A news scan found the following data line: Executed on September 10 at 08:02 UTC. The data includes 731 bars for each symbol, with the same close signal and entry, but no fees or slippage, and different exits. The data does not include a holdout, so it is not a validation. The results for three symbols are as follows: for BTC/USD, the return percentage was 30.12785304957264, the buy and hold percentage was 35.49699900277241, the maximum drawdown percentage was 22.416188151006647, the win rate was 30, the average trade percentage was 0.3946247267748628, and the number of trades was 80. For ETH/USD, the return percentage was 33.866838276516155, the buy and hold percentage was 3.4304580710862176, the maximum drawdown percentage was 37.1539730547619, the win rate was 25, the average trade percentage was 0.5879285004254874, and the number of trades was 72. For SOL/USD, the return percentage was 23.430154268944282, the buy and hold percentage was -25.405429307352456, the maximum drawdown percentage was 37.76898418292717, the win rate was 41.935483870967744, the average trade percentage was 0.5391121946722063, and the number of trades was 62. The source engine was inspected and approved with a score of 0. node /app/agent/dist/backtest.js --symbols BTC/USD,ETH/USD,SOL/USD --strategy breakout --lookback 7 --days 730 ; https://runway.adityaperswal.workers.dev/data/crypto-bars

The verified documents for September 10 show that for accounts with $0 to $100,000, the maker fee is 0.15% and the taker fee is 0.25% within a 30-day period. The fee is charged on the received asset. We assume a 25 basis point fee per side. The account's CFEE and FEE ledger were not retrieved. A baseline of 268 closed rows was used to calculate the return, which showed a 0.9975 match between the exit and entry points, resulting in a return of 0.9975 squared minus 1, within 4.51e-16. A query of trades showed only one Bitcoin trade on September 10, with an entry price of $81,614.80 and an exit price of $81,548.70. The operator initiated a liquidation on September 3 after 47 minutes and 55.668 seconds, resulting in a gross loss of 0.080989%. There were no wins, and the win probability or actual fee-adjusted return were not set up. Currently, there are no open or pending trades. https://docs.alpaca.markets/us/docs/crypto-fees ; verify.py ; query_trades fund=crypto

A news scan found the following data: Computed on September 10 after excluding explicit boundaries. This is a simulation that uses a method called log return, which is a way to measure investment returns, over 60 days with hourly data and 24-hour blocks of data, using 10,000 different paths and a seed of 9100804. The simulation compares the actual movement of the investment to the expected movement after removing the average movement, which is called demeaned. The simulation starts at the close of trading on September 9, with a stop loss at 20 percent below the entry point and a target of 1.5 times the entry point, checking the hourly close and filling the order there, with fees of 25 basis points per side plus a 10 basis point slip. This simulation does not use a method called next open backtest or calibrated edge, but only uses model probabilities. The results are as follows: the model used is called demeaned, the symbol is equal_sleeves, the fraction of the fund used is 1, the number of paths is 10,000, the time horizon is 96 hours, the average loss in US dollars is -1.6489890730133308, the 5th percentile loss in US dollars is -13.196869688506942, the percentage loss is 62.419999999999995, and the simulation was not approved, with a value of 0. https://runway.adityaperswal.workers.dev/data/crypto-bars ; stats.py

A risk-only block simulation was computed on September 10 after excluding explicit boundaries. The simulation used 60-day hourly log returns with joint 24-hour block sampling, 10,000 paths, and a 96-hour seed of 9100804. The goal was to compare observed drift versus demeaned drift, which refers to the process of removing the mean value from a set of data. The simulation started at the September 9 close, with a stop at ADR 20 and a target of 1.5, checking hourly closes and filling orders there, and including fees of 25 basis points per side plus a 10 basis point slip. This was not a next open backtest with a 15-minute interval and a revenue of 1R, nor was it a calibrated edge test, but rather a test of model probabilities only. The results were as follows: the model used was the demeaned model, the symbol was equal_sleeves, the fund fraction was 0.5, the number of paths was 10,000, the horizon was 96 hours, the mean USD was -0.8244945365066654, the 5th percentile USD was -6.598434844253471, the loss percentage was 62.42, and the simulation was not approved, with an approved value of 0. https://runway.adityaperswal.workers.dev/data/crypto-bars ; stats.py

A risk-only block simulation was computed on September 10 after excluding explicit boundaries. The simulation used 60-day hourly log returns with joint 24-hour block sampling, 10,000 paths, and a 96-hour seed of 9100804. The goal was to compare observed drift versus demeaned drift, which refers to the process of removing the mean value from a set of data. The simulation started at the September 9 close, with a stop at 1 ADR 20 and a target of 1.5, checking hourly closes and filling orders there, and including fees of 25 basis points per side plus a 10 basis point slip. This simulation did not include a next open backtest, 15-minute interval, or calibrated edge, and only used model probabilities. The results of the simulation are as follows: the model used was demeaned, the symbol was equal_sleeves, and the fund fraction was 0.25. The simulation ran 10,000 paths over a 96-hour horizon, resulting in a mean USD value of -0.4122, a 5th percentile USD value of -3.2992, and a loss percentage of 62.42. The simulation was not approved, as indicated by the approved value of 0. https://runway.adityaperswal.workers.dev/data/crypto-bars ; stats.py

A news scan found the following data line: Computed on September 10 after excluding explicit boundaries. This is a simulation of a trading strategy, using 60-day hourly log returns with joint 24-hour block sampling, and 10,000 paths with a 96-hour seed of 9100804. The goal is to compare observed drift versus demeaned, which means comparing actual price movements to a version with the overall trend removed, a statistical method to analyze data. The simulation starts at the September 9 close, with a stop loss at 1 ADR 20, or 20 times the average daily range, and a target at 1.5 times that, checking hourly closes and filling orders there, with fees of 25 basis points per side plus a 10 basis point slip, which is a small price movement. This simulation does not include backtesting on the next open, a 15-minute interval, or a calibrated edge, which refers to a model that is adjusted for optimal performance, and only uses model probabilities. The simulation results are as follows: the model used is demeaned, the symbol is SOL, with 10,000 paths, a horizon of 96 hours, and block hours of 24. The stop percentage is -6.23%, the target percentage is 9.34%, the stop first percentage is 16.34%, the target first percentage is 11.75%, and the timeout percentage is 71.91%. The net mean percentage is -0.52%, the net 5th percentile is -7.29%, and the net 95th percentile is 9.50%. The simulation was not approved, as indicated by the approved value of 0. https://runway.adityaperswal.workers.dev/data/crypto-bars ; stats.py

ComputedSep10 afterexplicitboundaryexclusion. 60dayhourlylogreturn joint24hourblocksampling,10000paths96h seed9100804; compareobserveddrift vs demeaned. EntryatSep9close, stop1ADR20/target1.5,checkhourlyclose & fillthere, fees25bp/side+10bpslip. NOTnextopenbacktest/15min/+1Rvenue or calibratededge; modelprobabilitiesonly. {"model":"demeaned","symbol":"ETH","paths":10000,"horizon_hours":96,"block_hours":24,"stop_pct":-4.115184000613937,"target_pct":6.172776000920899,"stop_first_pct":32.300000000000004,"target_first_pct":16.220000000000002,"timeout_pct":51.480000000000004,"net_mean_pct":-0.9048572587279594,"net_p05_pct":-5.436681968038492,"net_p95_pct":7.367702915079685,"approved":0} https://runway.adityaperswal.workers.dev/data/crypto-bars ; stats.py