28,321 calendar tests across 11 conditions · 1135 distinct findings after correction · as of 2026-10-23
The research behind the Patterns page, reading the same data: how the next sessions line up with the market’s surviving patterns, what tended to happen after days of each kind, every finding as a sortable table, and the arithmetic that says how much of it chance explains.
The market's own surviving buckets, matched to the labels each session ahead carries. Each is a long-run frequency against every session NOT in that bucket, not a forecast, and not combinable: a Friday that is also a turn-of-month day has two separate claims about it, not a stronger one.
Each bucket is also measured INSIDE each market state, with its baseline drawn from that same state, comparing a bucket in risk-on against every session would mostly measure the regime. F.46 derives the state from the price cross-section, so it reaches 2002 rather than 2023. Adding the conditions took the family from 167 tests to 550 and the survivors DOWN from 38 to 22: more questions asked, fewer answers kept, which is the correction working rather than failing.
Nothing survives at 60 sessions. Every bucket was tested at that horizon and none cleared the correction, printed because an omitted horizon reads as one that was never asked.
`n` counts overlapping windows: a five-session return measured from every session shares four days with its neighbour. `n_effective` divides by the horizon and is the honest sample size, across this dataset the raw count overstates it about sixfold. The p-value comes from rotating the series rather than shuffling it, so the null keeps that overlap.
A long-run conditional distribution for sessions of this description, measured over 6,135 sessions from 2002, NOT a forecast for this date. Each row is one bucket against every session outside it; two rows about the same session are two separate claims and not a stronger one. Past performance does not indicate future results.
| Utilities | (monthly expiry) | n 550 | -0.34pp | 0.011 | |
| HINDUNILVR | (month) | n 32 | -4.59pp | 0.015 | |
| SBC | (month) | n 13 | +24.82pp | 0.015 | |
| GODREJCP | (month) | n 24 | +6.50pp | 0.015 | |
| MMTC | (day of week) | n 816 | -0.41pp | 0.015 | |
| JWL | (month) | n 16 | -18.03pp | 0.015 | |
| NILKAMAL | (day of week) | n 1,510 | -0.33pp | 0.015 | |
| GHCL | (month) | n 32 | +9.63pp | 0.015 | |
| TEXRAIL | (month) | n 22 | -12.89pp | 0.015 | |
| VLSFINANCE | (month) | n 29 | +15.79pp | 0.015 | |
| GUFICBIO | (day of week) | n 1,065 | -0.42pp | 0.015 | |
| OAL | (month) | n 11 | +21.39pp | 0.015 | |
| Utilities | (month) | n 17 | +11.01pp | 0.020 | |
| Automobile and Auto Components | (monthly expiry) | n 1,050 | -0.13pp | 0.023 | |
| Healthcare | (moon phase) | n 810 | +0.14pp | 0.023 | |
| Telecommunication | (moon phase) | n 810 | +0.22pp | 0.023 | |
| ASIANPAINT | (month) | n 25 | +5.38pp | 0.023 | |
| BALAMINES | (day of week) | n 949 | -0.41pp | 0.023 | |
| NAVNETEDUL | (month) | n 31 | +7.82pp | 0.023 | |
| PARACABLES | (month) | n 19 | +19.96pp | 0.023 | |
| KECL | (month) | n 16 | -16.01pp | 0.023 | |
| GTLINFRA | (month) | n 20 | +25.14pp | 0.023 | |
| GODFRYPHLP | (day of week) | n 1,519 | +0.26pp | 0.023 | |
| ARIES | (day of week) | n 908 | -0.44pp | 0.023 | |
| TFCILTD | (month) | n 32 | -8.83pp | 0.023 | |
| CAPLIPOINT | (month) | n 13 | +15.41pp | 0.023 | |
| CESC | (day of week) | n 1,574 | -0.29pp | 0.023 | |
| GENNEX | (month) | n 25 | -14.35pp | 0.023 | |
| TITAGARH | (month) | n 18 | -16.91pp | 0.023 | |
| SAKSOFT | (month) | n 21 | -12.99pp | 0.023 | |
| TORNTPHARM | (month) | n 25 | +6.04pp | 0.023 | |
| KOKUYOCMLN | (month) | n 18 | -10.56pp | 0.023 | |
| Services | (day of month) | n 153 | +0.42pp | 0.027 | |
| Healthcare | (day of month) | n 264 | -0.26pp | 0.030 | |
| Textiles | (nth weekday) | n 352 | -0.25pp | 0.030 | |
| HIMATSEIDE | (month) | n 32 | -7.61pp | 0.030 | |
| JAIBALAJI | (month) | n 23 | +16.47pp | 0.030 | |
| ADSL | (month) | n 19 | -16.74pp | 0.030 | |
| MTNL | (month) | n 31 | +10.17pp | 0.030 | |
| SILLYMONKS | (month) | n 22 | -11.97pp | 0.030 | |
| SUNDROP | (month) | n 24 | -7.81pp | 0.030 | |
| EDELWEISS | (month) | n 19 | +11.66pp | 0.030 | |
| BIRLANU | (month) | n 32 | +10.36pp | 0.030 | |
| JBCHEPHARM | (month) | n 24 | +6.44pp | 0.030 | |
| POLYPLEX | (month) | n 32 | +9.73pp | 0.030 | |
| XCHANGING | (month) | n 22 | -11.96pp | 0.030 | |
| NEULANDLAB | (month) | n 26 | +14.50pp | 0.030 | |
| TORNTPOWER | (day of week) | n 969 | -0.32pp | 0.030 | |
| ONIDA | (month) | n 24 | +12.51pp | 0.030 | |
| Fast Moving Consumer Goods | (day of month) | n 261 | -0.24pp | 0.033 | |
| JAICORPLTD | (day of week) | n 1,275 | -0.39pp | 0.037 | |
| GRANULES | (month) | n 22 | +8.10pp | 0.037 | |
| 20MICRONS | (month) | n 18 | -11.98pp | 0.037 | |
| KERNEX | (month) | n 21 | -15.20pp | 0.037 | |
| EMKAY | (month) | n 20 | -13.83pp | 0.037 | |
| VENKEYS | (day of week) | n 1,128 | -0.31pp | 0.037 | |
| LALPATHLAB | (month) | n 11 | -8.00pp | 0.037 | |
| SPLPETRO | (month) | n 25 | -9.24pp | 0.037 | |
| HONAUT | (month) | n 31 | +7.98pp | 0.037 | |
| HIRECT | (month) | n 20 | -12.25pp | 0.037 | |
| MONTECARLO | (month) | n 12 | -11.13pp | 0.037 | |
| Forest Materials | (day of month) | n 261 | -0.37pp | 0.037 | |
| Metals & Mining | (day of month) | n 267 | -0.38pp | 0.037 | |
| Services | (month) | n 31 | +4.64pp | 0.039 | |
| Realty | (nth weekday) | n 366 | -0.32pp | 0.041 | |
| EIDPARRY | (month) | n 24 | -7.48pp | 0.042 | |
| NOCIL | (day of week) | n 1,549 | -0.30pp | 0.042 | |
| STYRENIX | (month) | n 32 | -6.66pp | 0.042 | |
| FDC | (month) | n 29 | +6.36pp | 0.042 | |
| KINETICENG | (month) | n 25 | -11.07pp | 0.042 | |
| SPICEJET | (month) | n 25 | +19.50pp | 0.042 | |
| JTEKTINDIA | (month) | n 30 | -7.55pp | 0.042 | |
| FIEMIND | (month) | n 20 | -8.43pp | 0.042 | |
| BOSCH-HCIL | (month) | n 30 | -8.43pp | 0.042 | |
| OFSS | (day of week) | n 1,199 | -0.24pp | 0.042 | |
| Oil Gas & Consumable Fuels | (day of month) | n 263 | +0.36pp | 0.044 | |
| Telecommunication | (day of month) | n 267 | -0.33pp | 0.047 | |
| CINEVISTA | (month) | n 26 | -12.19pp | 0.048 | |
| PNB | (month) | n 24 | -8.43pp | 0.048 | |
| BANCOINDIA | (month) | n 20 | -9.69pp | 0.048 | |
| WELENT | (month) | n 22 | +11.79pp | 0.048 | |
| ESTER | (month) | n 15 | -14.23pp | 0.048 | |
| VAIBHAVGBL | (month) | n 22 | +10.88pp | 0.048 | |
| HERITGFOOD | (month) | n 30 | +9.73pp | 0.048 | |
| Consumer Services | (month) | n 31 | +3.67pp | 0.049 | |
| Forest Materials | (month) | n 32 | +5.39pp | 0.049 | |
| Healthcare | (nth weekday) | n 352 | -0.19pp | 0.049 | |
| Oil Gas & Consumable Fuels | (day of month) | n 153 | +0.47pp | 0.049 |
Bars are to one scale across every scope, so a row is comparable with the row above it. The noise share is published once for the whole family (1,416 of 28,321) and apportioned to each scope by its share of the tests, it is an apportionment, not a per-scope measurement. Indices producing no survivors on 6,600 tests is a fact about a history that starts in 2023, not a fact about indices.
28,321 calendar tests. 4,946 clear an uncorrected p<0.05, against roughly 1,416 that pure noise would give; 1,552 survive a family-wide correction, and 1135 of those are distinct claims, 88 of them ones the market does not also have. Indices: 0 of 6,600, that history starts in 2023, so month-shaped questions cannot resolve at all. A stock or sector 'surviving' a bucket the MARKET also survives has not found an edge of its own, it was carried by one. Every table here shows the market's row for the same bucket beside it, and `inherited` counts how many survivors are of that kind. Every test is BUCKET VERSUS REST, never bucket versus zero. Month and quarter effects are measured on compounded periods, 24 Januaries, not 500 January days. The p-value is a permutation test over 2,000 label shuffles, and Benjamini-Hochberg runs across the whole family. There is no point-in-time universe before 2023-03-31, so a 24-year effect is measured over securities that survived into the panel. That inflates every mean and cannot be corrected with this data. Nothing here is a backtest feature: each row pools the whole history, including any period a reader would want to trade.
For learning and research only, not a buy, sell or hold recommendation, and not investment advice. What this platform is · Plans · Privacy · Terms · Refunds · Delivery · About · Contact