A post-result endpoint audit found that the auction execution was not authorized. The system used the official daily close instead of the final 30-minute bar close, and kept the entries and splits. This calculation was done on September 7. The results are shown in a table with the following columns: one-way break-even point, return, drawdown, win rate, sample size, average win percentage, average loss percentage, mean trade percentage, quarter Kelly, annualized percentage, and cash reconcile absolute. The rows of the table are: [0, 5.5512706498, 0.84854306756, 64.556962025, 79, 0.17400443483, 0.13267322349, 0.06866737172, 0.09383174055, 8.338080731, 7.2759576142e-12], [2, 2.2679993798, 0.88819571941, 54.430379747, 79, 0.16218681062, 0.13085689001, 0.028647909064, 0.044158814386, 3.3802938032, 1.0913936421e-11], [10, -9.8750717558, 9.9641116523, 16.455696203, 79, 0.22620985574, 0.20168266976, -0.13126997569, 0, -14.283535019, 3.4560798667e-11].
https://runway.adityaperswal.workers.dev/data/stock-bars
EEM QIM90708 official-close diagnostic recent rod = 3
Quant replication2026-09-07 08:15
A post-result endpoint audit found that the auction execution was not authorized. The system used the official daily close instead of the final 30-minute bar close, and kept the entries and splits. This was calculated on September 7. The diagnostic results are shown in a table with the following columns: one-way break-even point, return, drawdown, win rate, sample size, average win percentage, average loss percentage, mean trade percentage, quarter Kelly criterion, annualized percentage, and cash reconciliation absolute value. The rows of the table show the following values: for a value of 0, the results are 4.5832846713, 0.65593272921, 56.382978723, 94, 0.14674446473, 0.093764074552, 0.047826744909, 0.071283445097, 6.8685617353, and 1.8189894035e-11; for a value of 2, the results are 0.72396306969, 1.0841156661, 46.808510638, 94, 0.13249230655, 0.10189987011, 0.0078156168363, 0.014747303145, 1.0750402347, and 9.0949470177e-12; for a value of 10, the results are -13.340753548, 13.426369487, 10.638297872, 94, 0.18148567276, 0.1917778481, -0.1520689629, 0, -19.123984577, and 6.3664629124e-11.
https://runway.adityaperswal.workers.dev/data/stock-bars
A post-result endpoint audit found that the auction execution was not authorized. As a result, the official daily close was used instead of the final 30-minute bar close, and entries and splits were kept. This calculation was done on September 7. The columns in the data are: one-way break-even point, return, drawdown, win rate, sample size, average win percentage, average loss percentage, mean trade percentage, quarter Kelly criterion, annualized percentage, and cash reconciliation absolute value. The rows of data are: [0, 5.3095425351, 0.84854306756, 58.333333333, 96, 0.17108392005, 0.12176928992, 0.054135469641, 0.071692519692, 7.970496882, 1.0913936421e-11], [2, 1.3423143507, 1.374925192, 48.958333333, 96, 0.15965413333, 0.12547040168, 0.014121818583, 0.022113142781, 1.9962104002, 3.6379788071e-12], [10, -13.087312619, 13.173178947, 14.583333333, 96, 0.21233650799, 0.20691346391, -0.14577284301, 0, -18.773120204, 6.0026650317e-11].
https://runway.adityaperswal.workers.dev/data/stock-bars
A post-result endpoint audit found that the auction execution was not authorized. This uses the official daily close instead of the final 30-minute bar close, and keeps entries and splits. The calculation was done on September 7. The results are shown in a table with columns including one-way break-even point, return, drawdown, win rate, sample, average win percentage, average loss percentage, mean trade percentage, quarter Kelly, annualized percentage, and cash reconcile absolute. The rows of the table are: row 0 with values 0, -6.9321490714, 7.7048030275, 43.570057582, 521, 0.098154015859, 0.12204380063, -0.013688280904, 0, -1.4436172185, 3.092281986e-11, row 2 with values 2, -24.439866736, 24.816467422, 31.669865643, 521, 0.085375606283, 0.11812233131, -0.053674807887, 0, -5.514476752, 8.3673512563e-11, and row 10 with values 10, -67.170459989, 67.170459989, 4.4145873321, 521, 0.11211574213, 0.22849776221, -0.2134610816, 0, -20.184482736, 7.3214323493e-11.
https://runway.adityaperswal.workers.dev/data/stock-bars
EEM QIM90708 official-close diagnostic holdout rod = 3
Quant replication2026-09-07 08:15
A post-result endpoint audit found that the system was not authorized to execute an auction. Instead, it used the official daily close, rather than the final 30-minute bar close, and kept the entries and splits from the September 7 calculation. The results are shown in a table with columns including one-way break-even point, return, drawdown, win rate, sample size, average win percentage, average loss percentage, mean trade percentage, quarter Kelly, annualized percentage, and cash reconcile absolute value. The rows of the table show the following data: for a value of 0, the results were a -9.6757495485 return, 11.229184771 drawdown, 43.264659271 win rate, 631 sample size, 0.095154443533 average win percentage, 0.12151466038 average loss percentage, -0.016026451739 mean trade percentage, 0 quarter Kelly, -2.0387365548 annualized percentage, and 2.546585165e-11 cash reconcile absolute value; for a value of 2, the results were a -29.824027679 return, 30.12099325 drawdown, 32.963549921 win rate, 631 sample size, 0.076722802506 average win percentage, 0.12128118785 average loss percentage, -0.056012043641 mean trade percentage, 0 quarter Kelly, -6.9177103726 annualized percentage, and 2.9103830457e-11 cash reconcile absolute value; for a value of 10, the results were a -74.430355825 return, 74.430355825 drawdown, 4.2789223455 win rate, 631 sample size, 0.097352466948 average win percentage, 0.22979287594 average loss percentage, -0.21579458076 mean trade percentage, 0 quarter Kelly, -24.121669391 annualized percentage, and 7.8671291703e-11 cash reconcile absolute value.
https://runway.adityaperswal.workers.dev/data/stock-bars
A post-result endpoint audit found that the system was not authorized to execute an auction. Instead, it used the official daily close instead of the final 30-minute bar close, which is a type of trading data, and kept the entries and splits. This calculation was done on September 7. The results are shown in a table with columns including one-way break-even point, or the point at which a trade becomes profitable, return, drawdown, win rate, sample, average win percentage, average loss percentage, mean trade percentage, quarter Kelly, which is a measure of trading performance, annualized percentage, and cash reconcile absolute. The rows of the table show the following results: for a value of 0, the one-way break-even point is 0, the return is -9.0769607112, the drawdown is 10.170905457, the win rate is 43.349753695, the sample is 609, the average win percentage is 0.099042226094, the average loss percentage is 0.12446792335, the mean trade percentage is -0.015518355319, the quarter Kelly is 0, the annualized percentage is -1.9076344197, and the cash reconcile absolute is 1.2732925825e-11. For a value of 2, the results are one-way break-even point of 2, return of -28.734422575, drawdown of 29.193078047, win rate of 32.019704433, sample of 609, average win percentage of 0.085014887251, average loss percentage of 0.12169065367, mean trade percentage of -0.055504150418, quarter Kelly of 0, annualized percentage of -6.626969338, and cash reconcile absolute of 6.4574123826e-11. For a value of 10, the results are one-way break-even point of 10, return of -73.103038493, drawdown of 73.103038493, win rate of 4.7619047619, sample of 609, average win percentage of 0.10113139796, average loss percentage of 0.2311084444, mean trade percentage of -0.21528749952, quarter Kelly of 0, annualized percentage of -23.340420061, and cash reconcile absolute of 8.1399775809e-11.
https://runway.adityaperswal.workers.dev/data/stock-bars
A post-result endpoint audit found that the system was not authorized to execute an auction. Instead, it used the official daily close, rather than the final 30-minute bar close, and kept the entries and splits. This calculation was done on September 7. The results are shown in a table with columns including one-way break-even point, return, drawdown, win rate, sample size, average win percentage, average loss percentage, mean trade percentage, quarter Kelly, annualized percentage, and cash reconciliation absolute value. The rows of the table show the following values: for a value of 0, the results are 0, 2.7355273399, 5.700003291, 44.122965642, 553, 0.16034824822, 0.14464115335, 0.0050995353978, 0, 0.54687297531, and 4.3655745685e-11; for a value of 2, the results are 2, -17.651685444, 19.197133418, 37.974683544, 553, 0.14135891108, 0.14280476008, -0.034894505208, 0, -3.8487172098, and 2.9103830457e-11; for a value of 10, the results are 10, -66.006888602, 66.006888602, 11.934900542, 553, 0.15387454609, 0.24195234972, -0.19471080338, 0, -19.591653461, and 1.1368683772e-10.
https://runway.adityaperswal.workers.dev/data/stock-bars
EEM QIM90708 official-close diagnostic train rod = 3
Quant replication2026-09-07 08:15
A post-result endpoint audit found that the system was not authorized to execute an auction. Instead, it used the official daily close, rather than the final 30-minute bar close, and kept entries and splits. This calculation was done on September 7. The results are shown in a table with columns including one-way break-even point, return, drawdown, win rate, sample size, average win percentage, average loss percentage, mean trade percentage, quarter Kelly, annualized percentage, and cash reconcile absolute. The rows of the table show the following data: for a value of 0, the results were 7.1792266855 for return, 6.027416852 for drawdown, 46.569343066 for win rate, 685 for sample size, 0.15154105431 for average win percentage, 0.13895954569 for average loss percentage, 0.010322060223 for mean trade percentage, 0 for quarter Kelly, 1.410962487 for annualized percentage, and 3.6379788071e-11 for cash reconcile absolute. For a value of 2, the results were -18.508196224 for return, 21.398668064 for drawdown, 39.124087591 for win rate, 685 for sample size, 0.13457167813 for average win percentage, 0.13523248678 for average loss percentage, -0.029674068975 for mean trade percentage, 0 for quarter Kelly, -4.0516615625 for annualized percentage, and 9.0949470177e-12 for cash reconcile absolute. For a value of 10, the results were -72.765015689 for return, 72.765015689 for drawdown, 10.948905109 for win rate, 685 for sample size, 0.16321520337 for average win percentage, 0.23286517832 for average loss percentage, -0.18949871317 for mean trade percentage, 0 for quarter Kelly, -23.113929823 for annualized percentage, and 9.4132701633e-11 for cash reconcile absolute.
https://runway.adityaperswal.workers.dev/data/stock-bars
A post-result endpoint audit found that the system was not authorized to execute an auction. Instead, it used the official daily close instead of the final 30-minute bar close, and kept the entries and splits. This calculation was done on September 7. The results are shown in a table with columns including one-way break-even point, return, drawdown, win rate, sample size, average win percentage, average loss percentage, mean trade percentage, quarter Kelly, annualized percentage, and cash reconciliation absolute value. The rows of the table show the following values: for a value of 0, the results are 0, 4.6961805442, 5.04161817, 45.030120482, 664, 0.15759813158, 0.14322156077, 0.0071208725222, 0, 0.93173146166, 6.3664629124e-11; for a value of 2, the results are 2, -19.724647082, 20.776930148, 38.855421687, 664, 0.1377756643, 0.14131635901, -0.032873976457, 0, -4.3428365943, 8.185452316e-12; and for a value of 10, the results are 10, -72.254809037, 72.254809037, 11.445783133, 664, 0.15482087223, 0.23761032915, -0.19269350489, 0, -22.825008945, 1.0322764865e-10.
https://runway.adityaperswal.workers.dev/data/stock-bars
SPY QIM90708 annual main results and worst close discrepancies = 5
Quant replication2026-09-07 08:15
A calculation from September 7 shows the annual main results and worst close discrepancies for the SPY QIM90708 fund. The results are presented in a table with the following columns: year, one-way basis point, return, drawdown, win rate, sample, average win percentage, average loss percentage, mean trade percentage, quarter Kelly, annualized percentage, and cash reconcile absolute. The rows of the table are: 2021 with 10 one-way basis point, -31.940489655 return, 31.940489655 drawdown, 10.738255034 win rate, 149 sample, 0.073344150353 average win percentage, 0.29751632982 average loss percentage, -0.25769238564 mean trade percentage, 0 quarter Kelly, -32.044746228 annualized percentage, and 5.7298166212e-11 cash reconcile absolute, 2022 with 10 one-way basis point, -24.169157645 return, 24.169157645 drawdown, 23.140495868 win rate, 121 sample, 0.21040717996 average win percentage, 0.35972926486 average loss percentage, -0.2277968644 mean trade percentage, 0 quarter Kelly, -24.336810014 annualized percentage, and 4.638422979e-11 cash reconcile absolute, 2023 with 10 one-way basis point, -22.780084846 return, 22.823719014 drawdown, 15.492957746 win rate, 142 sample, 0.12172653892 average win percentage, 0.23731389155 average loss percentage, -0.18168790936 mean trade percentage, 0 quarter Kelly, -23.101387696 annualized percentage, and 7.5488060247e-11 cash reconcile absolute, 2024 with 10 one-way basis point, -26.717054116 return, 26.717054116 drawdown, 8.5526315789 win rate, 152 sample, 0.13170601642 average win percentage, 0.23548857411 average loss percentage, -0.2040837736 mean trade percentage, 0 quarter Kelly, -26.990991883 annualized percentage, and 5.8207660913e-11 cash reconcile absolute, 2025 with 10 one-way basis point, -23.909575744 return, 23.909575744 drawdown, 12.857142857 win rate, 140 sample, 0.15778171738 average win percentage, 0.24677494936 average loss percentage, -0.19476052078 mean trade percentage, 0 quarter Kelly, -24.329295206 annualized percentage, and 6.2755134422e-11 cash reconcile absolute. The worst close discrepancies are: on 2020-03-12 00:00:00 with -279.4016506189823 basis point discrepancy, entry at 228.78, exit at 226.14, and official close at 232.64, on 2020-03-18 00:00:00 with 225.7704124383197 basis point discrepancy, entry at 211.31, exit at 219.67, and official close at 214.82, and on 2025-04-09 00:00:00 with -95.56730378202704 basis point discrepancy, entry at 532.93, exit at 535.81, and official close at 540.98.
https://runway.adityaperswal.workers.dev/data/stock-bars
A post-result endpoint audit found that the auction execution was not authorized. The system uses the official daily close instead of the final 30-minute bar close, and it keeps the entries and splits from the September 7 calculation. The columns in the data are: oneway_bp, return, drawdown, win_rate, sample, average_win_pct, average_loss_pct, mean_trade_pct, quarter_kelly, annualized_pct, and cash_reconcile_abs. The rows of data are: [0, 1.8495538789, 0.72161981203, 50.649350649, 77, 0.12728886792, 0.082175103559, 0.023917037837, 0.046973938548, 2.7538827066, 9.0949470177e-12], [2, -1.23959529, 1.4711305434, 35.064935065, 77, 0.13621142964, 0.098324345533, -0.016084528265, 0, -1.832012453, 9.0949470177e-12], and [10, -12.687228228, 12.687228228, 9.0909090909, 77, 0.15343956552, 0.20886794473, -0.17593089834, 0, -18.218237323, 4.0017766878e-11].
https://runway.adityaperswal.workers.dev/data/stock-bars
SPY QIM90708 official-close diagnostic recent rod = 3
Quant replication2026-09-07 08:15
A post-result endpoint audit found that the auction execution was not authorized. Instead, the official daily close was used, rather than the final 30-minute bar close, and entries and splits were kept. This calculation was done on September 7. The results are shown in a table with columns including one-way break-even point, return, drawdown, win rate, sample size, average win percentage, average loss percentage, mean trade percentage, quarter Kelly, annualized percentage, and cash reconcile absolute. The rows of the table show the following results: for a value of 0, the return was -0.086815645063, the drawdown was 1.3747975486, the win rate was 45.348837209, the sample size was 86, the average win percentage was 0.096106625643, the average loss percentage was 0.08324067356, the mean trade percentage was -0.00094084399631, the quarter Kelly was 0, the annualized percentage was -0.12866447749, and the cash reconcile absolute was 1.8189894035e-12. For a value of 2, the return was -3.4653846459, the drawdown was 3.4653846459, the win rate was 30.23255814, the sample size was 86, the average win percentage was 0.095946959407, the average loss percentage was 0.10024688788, the mean trade percentage was -0.040932468934, the quarter Kelly was 0, the annualized percentage was -5.0937301473, and the cash reconcile absolute was 7.2759576142e-12. For a value of 10, the return was -15.875180854, the drawdown was 15.875180854, the win rate was 3.488372093, the sample size was 86, the average win percentage was 0.13538210219, the average loss percentage was 0.21288807371, the mean trade percentage was -0.20073911409, the quarter Kelly was 0, the annualized percentage was -22.605324546, and the cash reconcile absolute was 3.6379788071e-11.
https://runway.adityaperswal.workers.dev/data/stock-bars
A post-result endpoint audit was performed, but auction execution was not authorized. The official daily close was used instead of the final 30-minute bar close, and entries and splits were kept. The calculation was done on September 7. The results are shown in a table with columns including one-way break-even point, return, drawdown, win rate, sample size, average win percentage, average loss percentage, mean trade percentage, quarter Kelly, annualized percentage, and cash reconcile absolute. The rows of the table show the following results: for a value of 0, the return was 1.3957801616, the drawdown was 0.95690598699, the win rate was 48.387096774, the sample size was 93, the average win percentage was 0.1248425628, the average loss percentage was 0.087944457061, the mean trade percentage was 0.015017004161, the quarter Kelly was 0.030071883787, the annualized percentage was 2.0759871331, and the cash reconcile absolute was 1.4551915228e-11. For a value of 2, the return was -2.3068470299, the drawdown was 2.3068470299, the win rate was 33.333333333, the sample size was 93, the average win percentage was 0.13330067676, the average loss percentage was 0.10412184234, the mean trade percentage was -0.02498100264, the quarter Kelly was 0, the annualized percentage was -3.4004601038, and the cash reconcile absolute was 1.8189894035e-12. For a value of 10, the return was -15.813762394, the drawdown was 15.813762394, the win rate was 7.5268817204, the sample size was 93, the average win percentage was 0.15343956552, the average loss percentage was 0.21234534749, the mean trade percentage was -0.18481314974, the quarter Kelly was 0, the annualized percentage was -22.521549643, and the cash reconcile absolute was 4.9112713896e-11.
https://runway.adityaperswal.workers.dev/data/stock-bars
A news scan found the following data line: Post-result endpoint audit, not authorized auction execution. It uses the official daily close instead of the final 30-minute bar close, which is a price at the end of a 30-minute trading period; it keeps entries and splits. This calculation was done on September 7. The columns in the data are: one-way break-even point, return, drawdown, win rate, sample, average win percentage, average loss percentage, mean trade percentage, quarter Kelly, annualized percentage, and cash reconcile absolute. The rows of data are: [0, -9.4911329344, 12.676319384, 46.0456942, 569, 0.17202530592, 0.18162998586, -0.017190906117, 0, -1.9982418467, 5.4569682106e-12], [2, -27.914866966, 27.948764043, 35.32513181, 569, 0.17725500498, 0.18522124564, -0.05717603233, 0, -6.4106134999, 8.731149137e-11], [10, -70.995594574, 70.995594574, 13.708260105, 569, 0.18356181019, 0.28058286836, -0.21695670856, 0, -22.160950179, 9.0494722826e-11].
https://runway.adityaperswal.workers.dev/data/stock-bars
SPY QIM90708 official-close diagnostic holdout rod = 3
Quant replication2026-09-07 08:15
A post-result endpoint audit found that the system was not authorized to execute an auction. Instead, it used the official daily close instead of the final 30-minute bar close, which is a type of price data, and kept the entries and splits. This calculation was done on September 7. The system analyzed several columns of data, including one-way break-even point, return, drawdown, win rate, sample size, average win percentage, average loss percentage, mean trade percentage, quarter Kelly, annualized percentage, and cash reconcile absolute. The rows of data showed the following results: for a value of 0, the return was -3.678375616, the drawdown was 6.0771111273, the win rate was 48.81305638, the sample size was 674, the average win percentage was 0.15872271377, the average loss percentage was 0.16451005586, the mean trade percentage was -0.0052657805686, the quarter Kelly was 0, the annualized percentage was -0.7557075492, and the cash reconcile absolute was 3.092281986e-11. For a value of 2, the return was -26.440629286, the drawdown was 26.475219603, the win rate was 38.575667656, the sample size was 674, the average win percentage was 0.15517546371, the average loss percentage was 0.1711303046, the mean trade percentage was -0.045255675877, the quarter Kelly was 0, the annualized percentage was -6.0263172445, and the cash reconcile absolute was 5.1841198001e-11. For a value of 10, the return was -74.979568939, the drawdown was 74.979568939, the win rate was 14.391691395, the sample size was 674, the average win percentage was 0.14562350146, the average loss percentage was 0.26400836325, the mean trade percentage was -0.20505540943, the quarter Kelly was 0, the annualized percentage was -24.454419025, and the cash reconcile absolute was 9.6406438388e-11.
https://runway.adityaperswal.workers.dev/data/stock-bars
A post-result endpoint audit found that the system was not authorized to execute an auction. Instead, it used the official daily close, rather than the final 30-minute bar close, and kept the entries and splits from the September 7 calculation. The columns analyzed were one-way break-even point, return, drawdown, win rate, sample, average win percentage, average loss percentage, mean trade percentage, quarter Kelly, annualized percentage, and cash reconcile absolute. The results were as follows: the system had a return of -6.4084791425, a drawdown of 12.231428112, and a win rate of 48.153409091 for a sample of 704, with an average win percentage of 0.16359515097 and an average loss percentage of 0.17234256412, resulting in a mean trade percentage of -0.0091082732068, a quarter Kelly of 0, and an annualized percentage of -1.3316214647, with a cash reconcile absolute of 9.0949470177e-12. For a different set of parameters, the return was -29.378135691, the drawdown was 29.411344686, and the win rate was 37.642045455, with an average win percentage of 0.16288961129 and an average loss percentage of 0.17706099271, resulting in a mean trade percentage of -0.049096631826, a quarter Kelly of 0, and an annualized percentage of -6.7983000699, with a cash reconcile absolute of 7.185008144e-11. For another set of parameters, the return was -77.104516297, the drawdown was 77.104516297, and the win rate was 14.204545455, with an average win percentage of 0.15728417769 and an average loss percentage of 0.26951512583, resulting in a mean trade percentage of -0.20889022476, a quarter Kelly of 0, and an annualized percentage of -25.799438265, with a cash reconcile absolute of 7.9580786405e-11.
https://runway.adityaperswal.workers.dev/data/stock-bars
A post-result endpoint audit found that the system was not authorized to execute an auction. Instead, it used the official daily close, keeping entries and splits, for a calculation on September 7. The results are shown in a table with columns including one-way break-even point, return, drawdown, win rate, sample, average win percentage, average loss percentage, mean trade percentage, quarter Kelly, annualized percentage, and cash reconcile absolute. The rows of the table show the following data: for a value of 0, the results were a return of -6.5081536976, a drawdown of 8.492818411, a win rate of 47.202797203, a sample of 572, an average win percentage of 0.17183768569, an average loss percentage of 0.18272385301, a mean trade percentage of -0.011208074098, a quarter Kelly of 0, an annualized percentage of -1.3507448627, and a cash reconcile absolute of 3.6379788071e-12. For a value of 2, the results were a return of -25.628394209, a drawdown of 25.924706887, a win rate of 36.713286713, a sample of 572, an average win percentage of 0.17468564449, an average loss percentage of 0.18223166994, a mean trade percentage of -0.051195592964, a quarter Kelly of 0, an annualized percentage of -5.8081525416, and a cash reconcile absolute of 7.7307049651e-11. For a value of 10, the results were a return of -70.218895005, a drawdown of 70.218895005, a win rate of 9.965034965, a sample of 572, an average win percentage of 0.31171491901, an average loss percentage of 0.26883814619, a mean trade percentage of -0.21098583025, a quarter Kelly of 0, an annualized percentage of -21.712698053, and a cash reconcile absolute of 9.6406438388e-11.
https://runway.adityaperswal.workers.dev/data/stock-bars
SPY QIM90708 official-close diagnostic train rod = 3
Quant replication2026-09-07 08:15
A post-result endpoint audit found that the auction execution was not authorized. The system uses the official daily close instead of the final 30-minute bar close, and it keeps entries and splits. This calculation was done on September 7. The results are shown in a table with the following columns: one-way break-even point, return, drawdown, win rate, sample size, average win percentage, average loss percentage, mean trade percentage, quarter Kelly, annualized percentage, and cash reconcile absolute. The rows of the table show the following values: for a value of 0, the results are 3.614048323 for return, 8.0108632829 for drawdown, 48.714285714 for win rate, 700 for sample size, 0.17340308307 for average win percentage, 0.16099559947 for average loss percentage, 0.0055842295831 for mean trade percentage, 0.0027455252421 for quarter Kelly, 0.72003694843 for annualized percentage, and 3.6379788071e-11 for cash reconcile absolute. For a value of 2, the results are -21.690186964 for return, 24.218820056 for drawdown, 38.571428571 for win rate, 700 for sample size, 0.17355324502 for average win percentage, 0.16499158037 for average loss percentage, -0.034410004862 for mean trade percentage, 0 for quarter Kelly, -4.8208452351 for annualized percentage, and 9.5496943686e-11 for cash reconcile absolute. For a value of 10, the results are -74.449093243 for return, 74.461302606 for drawdown, 11 for win rate, 700 for sample size, 0.29467493862 for average win percentage, 0.25465316951 for average loss percentage, -0.19422707762 for mean trade percentage, 0 for quarter Kelly, -24.099313634 for annualized percentage, and 9.231371223e-11 for cash reconcile absolute.
https://runway.adityaperswal.workers.dev/data/stock-bars
A post-result endpoint audit found that the system was not authorized to execute an auction. Instead, it used the official daily close, rather than the final 30-minute bar close, and kept the entries and splits from the September 7 calculation. The columns analyzed included one-way break-even point, return, drawdown, win rate, sample size, average win percentage, average loss percentage, mean trade percentage, quarter Kelly, annualized percentage, and cash reconciliation absolute value. The results were as follows: for a value of 0, the results were -7.2470431972 for return, 9.5499394267 for drawdown, 48.245614035 for win rate, 684 for sample size, 0.16202696375 for average win percentage, 0.17838920895 for average loss percentage, -0.010502095036 for mean trade percentage, 0 for quarter Kelly, -1.5087995429 for annualized percentage, and 2.0008883439e-11 for cash reconciliation absolute value. For a value of 2, the results were -29.44873707 for return, 29.46750534 for drawdown, 37.134502924 for win rate, 684 for sample size, 0.16404366993 for average win percentage, 0.17721437486 for average loss percentage, -0.050489896238 for mean trade percentage, 0 for quarter Kelly, -6.8066141922 for annualized percentage, and 9.0039975476e-11 for cash reconciliation absolute value. For a value of 10, the results were -76.383645336 for return, 76.383645336 for drawdown, 9.649122807 for win rate, 684 for sample size, 0.29069760949 for average win percentage, 0.26378385963 for average loss percentage, -0.21028126173 for mean trade percentage, 0 for quarter Kelly, -25.297399792 for annualized percentage, and 9.0949470177e-11 for cash reconciliation absolute value.
https://runway.adityaperswal.workers.dev/data/stock-bars
TLT QIM90708 regressions recent = 170
Quant replication2026-09-07 08:15
A news scan found the following data: On September 7, a calculation was made using a statistical method called Ordinary Least Squares, or OLS, which is a way to find the best fit line for a set of data. This calculation looked at the return on investment for the TLT fund over the last half hour and the rest of the day. The results are based on a method called HAC5, which is a type of statistical test, and show a descriptive association, meaning they describe a relationship, but do not imply causation. The returns are given as decimal values. The results are as follows: for the TLT fund, with 170 data points, the intercept and first return had a beta of -0.000035467817224784146 and 0.002881191988335393, respectively, with HAC5 t-values of -0.43988466029270523 and 0.17845078982475207, and an R2 value of 0.0002058620007591161. Another calculation for the TLT fund, also with 170 data points, looked at the intercept, first return, and 12th return, and found beta values of -0.00003503362099835063, 0.002779649248119993, and -0.022034903656255284, with HAC5 t-values of -0.4382212061073656, 0.17218994894517625, and -0.3039588866198136, and an R2 value of 0.0008390464066386771. A third calculation for the TLT fund, with 170 data points, looked at the intercept and rest of day return, and found beta values of -0.00003448571699550978 and 0.008816427816437475, with HAC5 t-values of -0.42999529499548744 and 0.6633577356506899, and an R2 value of 0.003025284165777631.
https://runway.adityaperswal.workers.dev/data/stock-bars
TLT QIM90708 regressions holdout = 1245
Quant replication2026-09-07 08:15
A news scan found the following data: On September 7, a calculation was made using a method called Ordinary Least Squares, or OLS, which is a way to find the best fit line for a set of data. This calculation looked at the return on investment for the TLT symbol, which is a type of bond, over the last half hour and the rest of the day. The results showed a descriptive association, which means it describes a relationship but does not imply cause and effect, between the returns. The returns are shown as decimal numbers. The calculation was done three times with different sets of data, and the results are as follows: for the first set, the intercept, or starting point, was 0.000248959387413663 and the beta, or measure of volatility, for the first return was 0.02359578057256328, with a HAC5 t-value, which is a measure of statistical significance, of 4.628045478418234 and 3.306330830166599, respectively, and an R2, or measure of how well the model fits the data, of 0.012466878925200042, and a correlation between the first return and the last return of 0.1116551786761366. For the second set, the intercept was 0.0002476156188481961, the beta for the first return was 0.02364430208445091, and the beta for the 12th return was 0.056441261545553505, with HAC5 t-values of 4.586398758528841, 3.298463752957348, and 1.1046236516478027, respectively, and an R2 of 0.014622877250393573, and the same correlation as before. For the third set, the intercept was 0.00025177431556341044, the beta for the return on the rest of the day was 0.02636233138984397, with HAC5 t-values of 4.682144620096881 and 4.479883138698519, respectively, and an R2 of 0.024079154323103924, and the same correlation as before.
https://runway.adityaperswal.workers.dev/data/stock-bars
TLT QIM90708 regressions train = 1244
Quant replication2026-09-07 08:15
A news scan found the following data: on September 7, a calculation was performed using ordinary least squares, or OLS, which is a method to find the best fit line, to analyze the last half hour return on the first return and/or the 12th return for the rest of the day. The results used HAC5, which stands for a statistical method to calculate standard errors, to show descriptive association, not causation. The returns are given as decimal numbers. The results for the TLT symbol, which represents a type of investment, are as follows: for the train split, which is a part of the data, with 1244 data points, the columns analyzed were intercept and r1, which refers to the first return, and the beta values, which measure the change, were 0.00004360245005053721 and -0.013059649928567042, with HAC5 t-values of 0.8423265376344887 and -0.5120607995475798, and an R2 value, which measures the goodness of fit, of 0.002468234035397643. The correlation between r1 and the last return was -0.04968132481524003. Another analysis with columns intercept, r1, and r12, which refers to the 12th return, had beta values of 0.00004125964744352307, -0.014882731457980625, and -0.05213426640036488, with HAC5 t-values of 0.7847925821060477, -0.6330631969103219, and -0.4372889290961637, and an R2 value of 0.004993636529019296. A third analysis with columns intercept and rod, which refers to the rest of the day, had beta values of 0.000032547142905647357 and 0.02740994507041745, with HAC5 t-values of 0.5514953420360819 and 1.1908025231207757, and an R2 value of 0.016747267682585942.
https://runway.adityaperswal.workers.dev/data/stock-bars
EFA QIM90708 regressions recent = 170
Quant replication2026-09-07 08:15
On September 7, a calculation was done using a method called Ordinary Least Squares, or OLS, which is a way to find the best fit line for some data. This calculation looked at the return of the EFA stock over the last half hour and the first return and/or the 12th return for the rest of the day. The HAC5 method was used, which is a way to correct for problems with the data, and it showed a descriptive association, not a cause-and-effect relationship. The returns were given as decimal numbers. The results were as follows: for the first calculation, the intercept was 0.00015170849876177867 and the beta for the first return was 0.03658996323586186, with a HAC5 t-value of 1.1529085227867857 and 2.884724129409552, an R2 value of 0.04861774586731127, and a correlation between the first return and the last return of 0.22049432162146704. For the second calculation, the intercept was 0.00014556836680757729, the beta for the first return was 0.03840888674023462, and the beta for the 12th return was 0.13304567227240102, with HAC5 t-values of 1.1250293504118587, 3.0899130124802117, and 1.6004161865311206, an R2 value of 0.07568781471580954, and the same correlation. For the third calculation, the intercept was 0.0001564316243796275 and the beta for the rest of the day was 0.025416704846470205, with HAC5 t-values of 1.2269028120568617 and 1.8060868682511708, an R2 value of 0.0284113043429689, and the same correlation.
https://runway.adityaperswal.workers.dev/data/stock-bars
EFA QIM90708 regressions holdout = 1245
Quant replication2026-09-07 08:15
A news scan found the following data: On September 7, a calculation was made using Ordinary Least Squares, or OLS, which is a method to find the best fit line. The calculation looked at the return on the first return and/or the 12th return for the rest of the day, for the last half hour. This used HAC5, which is a statistical method to calculate standard errors, to find a descriptive association, not a cause-and-effect relationship. The returns are given as decimal numbers. The results for the symbol EFA, which is a stock market ticker, in the holdout set, which is a part of the data not used for training, are as follows: for the first model, the intercept was 0.00006689610017491031 and the beta for the first return was -0.001546302068832937, with a HAC5 t-value of 1.503591315437261 and -0.174465164833615, and an R2 value of 0.000057569478913999106. For the second model, the intercept was 0.00006647647062517855, the beta for the first return was -0.0011561300853593882, and the beta for the 12th return was 0.08983520315772228, with HAC5 t-values of 1.4936961304155791, -0.13156560478161847, and 1.6709349852506306, and an R2 value of 0.006664019267157539. For the third model, the intercept was 0.00006422681531507994 and the beta for the rest of the day was 0.006965383680357814, with HAC5 t-values of 1.4300355424297628 and 1.0343336252494053, and an R2 value of 0.001827527257958761. The correlation between the first return and the last return was -0.007587455364873097 for all models.
https://runway.adityaperswal.workers.dev/data/stock-bars
EFA QIM90708 regressions train = 1247
Quant replication2026-09-07 08:15
A news scan found the following data: On September 7, a calculation was performed using Ordinary Least Squares, or OLS, which is a method to find the best fit line. The calculation looked at the return on the first return and/or the 12th return for the rest of the day, with the last half hour return. This used HAC5, which is a statistical method to calculate standard errors, to find descriptive association, not causation. The returns were in decimal form. The results were as follows: for the symbol EFA, which is a stock ticker, in the train split, with 1247 data points, the columns included the intercept, which is a starting point, and r1, which is the first return, with beta values of 0.00006564872991631818 and 0.024224035523706743, and HAC5 t values of 1.1669704765137179 and 1.4991421711960247, with an R2 value of 0.012101133557348542, and a correlation of 0.11000515241273141. Another calculation for EFA in the train split with 1247 data points included the columns intercept, r1, and r12, which is the 12th return, with beta values of 0.00006390020005458628, 0.02379356088473252, and 0.054938540214414966, and HAC5 t values of 1.1388856980901652, 1.515158307983134, and 0.5561713662900177, with an R2 value of 0.013414668404556296, and a correlation of 0.11000515241273141. A third calculation for EFA in the train split with 1247 data points included the columns intercept and rod, which is the rest of the day return, with beta values of 0.00006189093248617362 and 0.022569496216873777, and HAC5 t values of 1.1003262009401087 and 1.8439993952355738, with an R2 value of 0.014810817198015358, and a correlation of 0.11000515241273141.
https://runway.adityaperswal.workers.dev/data/stock-bars