Analysis: the log of every test the nightly job ran, and the method
Findings log, 79 tests
LearnHow to read it
79 tests and 36 article candidates, written by the nightly job on 2026-09-22: date, statistic, sample, p-value and one descriptive sentence per test. A test at p under 0.01 on at least 8 independent points is flagged for the Learn section.
79 tests on the last run; 36 reach the article-candidate rule (p under 0.01 on at least 8 independent points). A click on a column's header sorts by it.
| leverage:MARA:365price | 2026-09-22 | 365 | 1.28 | 0.44 | under 0.0001 | Over the last 365 sessions, MARA's daily log return moved 1.28 for each 1.00 of bitcoin's (95% interval 1.13 to 1.43; r 0.67, R squared 0.44, p under 0.0001, n 365). | article candidate |
|---|---|---|---|---|---|---|---|
| leverage:CLSK:365price | 2026-09-22 | 365 | 1.33 | 0.41 | under 0.0001 | Over the last 365 sessions, CLSK's daily log return moved 1.33 for each 1.00 of bitcoin's (95% interval 1.16 to 1.49; r 0.64, R squared 0.41, p under 0.0001, n 365). | article candidate |
| leverage:BTBT:365price | 2026-09-22 | 365 | 1.24 | 0.34 | under 0.0001 | Over the last 365 sessions, BTBT's daily log return moved 1.24 for each 1.00 of bitcoin's (95% interval 1.07 to 1.42; r 0.58, R squared 0.34, p under 0.0001, n 365). | article candidate |
| leverage:RIOT:365price | 2026-09-22 | 365 | 1.09 | 0.30 | under 0.0001 | Over the last 365 sessions, RIOT's daily log return moved 1.09 for each 1.00 of bitcoin's (95% interval 0.92 to 1.27; r 0.55, R squared 0.30, p under 0.0001, n 365). | article candidate |
| leverage:HIVE:365price | 2026-09-22 | 365 | 1.17 | 0.29 | under 0.0001 | Over the last 365 sessions, HIVE's daily log return moved 1.17 for each 1.00 of bitcoin's (95% interval 0.98 to 1.36; r 0.54, R squared 0.29, p under 0.0001, n 365). | article candidate |
| leverage:DMGGF:365price | 2026-09-22 | 365 | 1.14 | 0.26 | under 0.0001 | Over the last 365 sessions, DMGGF's daily log return moved 1.14 for each 1.00 of bitcoin's (95% interval 0.94 to 1.34; r 0.51, R squared 0.26, p under 0.0001, n 365). | article candidate |
| concordance:MARA:365price | 2026-09-22 | 365 | 106.09 | not computed | under 0.0001 | Over the last 365 sessions, MARA closed up on 75% of the days bitcoin closed up and on 21% of the days it did not (chi-square 106.09, p under 0.0001, n 365). | article candidate |
| leverage:BTDR:365price | 2026-09-22 | 365 | 1.22 | 0.25 | under 0.0001 | Over the last 365 sessions, BTDR's daily log return moved 1.22 for each 1.00 of bitcoin's (95% interval 1.00 to 1.44; r 0.50, R squared 0.25, p under 0.0001, n 365). | article candidate |
| leverage:KEEL:365price | 2026-09-22 | 365 | 1.23 | 0.25 | under 0.0001 | Over the last 365 sessions, KEEL's daily log return moved 1.23 for each 1.00 of bitcoin's (95% interval 1.01 to 1.46; r 0.50, R squared 0.25, p under 0.0001, n 365). | article candidate |
| concordance:BTDR:365price | 2026-09-22 | 365 | 88.27 | not computed | under 0.0001 | Over the last 365 sessions, BTDR closed up on 77% of the days bitcoin closed up and on 28% of the days it did not (chi-square 88.27, p under 0.0001, n 365). | article candidate |
| concordance:CLSK:365price | 2026-09-22 | 365 | 87.96 | not computed | under 0.0001 | Over the last 365 sessions, CLSK closed up on 76% of the days bitcoin closed up and on 27% of the days it did not (chi-square 87.96, p under 0.0001, n 365). | article candidate |
| leverage:HUT:365price | 2026-09-22 | 365 | 1.06 | 0.20 | under 0.0001 | Over the last 365 sessions, HUT's daily log return moved 1.06 for each 1.00 of bitcoin's (95% interval 0.84 to 1.27; r 0.45, R squared 0.20, p under 0.0001, n 365). | article candidate |
| concordance:BTBT:365price | 2026-09-22 | 365 | 77.99 | not computed | under 0.0001 | Over the last 365 sessions, BTBT closed up on 71% of the days bitcoin closed up and on 24% of the days it did not (chi-square 77.99, p under 0.0001, n 365). | article candidate |
| concordance:DMGGF:365price | 2026-09-22 | 365 | 72.53 | not computed | under 0.0001 | Over the last 365 sessions, DMGGF closed up on 69% of the days bitcoin closed up and on 25% of the days it did not (chi-square 72.53, p under 0.0001, n 365). | article candidate |
| concordance:HIVE:365price | 2026-09-22 | 365 | 70.80 | not computed | under 0.0001 | Over the last 365 sessions, HIVE closed up on 71% of the days bitcoin closed up and on 27% of the days it did not (chi-square 70.80, p under 0.0001, n 365). | article candidate |
| concordance:KEEL:365price | 2026-09-22 | 365 | 69.21 | not computed | under 0.0001 | Over the last 365 sessions, KEEL closed up on 72% of the days bitcoin closed up and on 28% of the days it did not (chi-square 69.21, p under 0.0001, n 365). | article candidate |
| leverage:VIP:365price | 2026-09-22 | 365 | 1.12 | 0.17 | under 0.0001 | Over the last 365 sessions, VIP's daily log return moved 1.12 for each 1.00 of bitcoin's (95% interval 0.87 to 1.37; r 0.42, R squared 0.17, p under 0.0001, n 365). | article candidate |
| leverage:CIFR:365price | 2026-09-22 | 365 | 1.04 | 0.16 | under 0.0001 | Over the last 365 sessions, CIFR's daily log return moved 1.04 for each 1.00 of bitcoin's (95% interval 0.80 to 1.29; r 0.40, R squared 0.16, p under 0.0001, n 365). | article candidate |
| leverage:IREN:365price | 2026-09-22 | 365 | 0.92 | 0.15 | under 0.0001 | Over the last 365 sessions, IREN's daily log return moved 0.92 for each 1.00 of bitcoin's (95% interval 0.70 to 1.15; r 0.39, R squared 0.15, p under 0.0001, n 365). | article candidate |
| concordance:VIP:365price | 2026-09-22 | 365 | 55.76 | not computed | under 0.0001 | Over the last 365 sessions, VIP closed up on 67% of the days bitcoin closed up and on 28% of the days it did not (chi-square 55.76, p under 0.0001, n 365). | article candidate |
| leverage:WULF:365price | 2026-09-22 | 365 | 0.89 | 0.14 | under 0.0001 | Over the last 365 sessions, WULF's daily log return moved 0.89 for each 1.00 of bitcoin's (95% interval 0.66 to 1.11; r 0.38, R squared 0.14, p under 0.0001, n 365). | article candidate |
| concordance:IREN:365price | 2026-09-22 | 365 | 50.27 | not computed | under 0.0001 | Over the last 365 sessions, IREN closed up on 71% of the days bitcoin closed up and on 34% of the days it did not (chi-square 50.27, p under 0.0001, n 365). | article candidate |
| concordance:HUT:365price | 2026-09-22 | 365 | 49.64 | not computed | under 0.0001 | Over the last 365 sessions, HUT closed up on 73% of the days bitcoin closed up and on 37% of the days it did not (chi-square 49.64, p under 0.0001, n 365). | article candidate |
| leverage:DGXX:365price | 2026-09-22 | 365 | 1.00 | 0.13 | under 0.0001 | Over the last 365 sessions, DGXX's daily log return moved 1.00 for each 1.00 of bitcoin's (95% interval 0.73 to 1.26; r 0.36, R squared 0.13, p under 0.0001, n 365). | article candidate |
| concordance:RIOT:365price | 2026-09-22 | 365 | 48.86 | not computed | under 0.0001 | Over the last 365 sessions, RIOT closed up on 71% of the days bitcoin closed up and on 34% of the days it did not (chi-square 48.86, p under 0.0001, n 365). | article candidate |
| leverage:CORZ:365price | 2026-09-22 | 365 | 0.60 | 0.12 | under 0.0001 | Over the last 365 sessions, CORZ's daily log return moved 0.60 for each 1.00 of bitcoin's (95% interval 0.43 to 0.77; r 0.35, R squared 0.12, p under 0.0001, n 365). | article candidate |
| concordance:CIFR:365price | 2026-09-22 | 365 | 45.57 | not computed | under 0.0001 | Over the last 365 sessions, CIFR closed up on 68% of the days bitcoin closed up and on 32% of the days it did not (chi-square 45.57, p under 0.0001, n 365). | article candidate |
| concordance:WULF:365price | 2026-09-22 | 365 | 39.08 | not computed | under 0.0001 | Over the last 365 sessions, WULF closed up on 68% of the days bitcoin closed up and on 36% of the days it did not (chi-square 39.08, p under 0.0001, n 365). | article candidate |
| leverage:BGDE:365price | 2026-09-22 | 365 | 1.18 | 0.09 | under 0.0001 | Over the last 365 sessions, BGDE's daily log return moved 1.18 for each 1.00 of bitcoin's (95% interval 0.80 to 1.56; r 0.31, R squared 0.09, p under 0.0001, n 365). | article candidate |
| concordance:ANY:365price | 2026-09-22 | 365 | 35.68 | not computed | under 0.0001 | Over the last 365 sessions, ANY closed up on 60% of the days bitcoin closed up and on 29% of the days it did not (chi-square 35.68, p under 0.0001, n 365). | article candidate |
| concordance:CORZ:365price | 2026-09-22 | 365 | 32.05 | not computed | under 0.0001 | Over the last 365 sessions, CORZ closed up on 69% of the days bitcoin closed up and on 39% of the days it did not (chi-square 32.05, p under 0.0001, n 365). | article candidate |
| leverage:ANY:365price | 2026-09-22 | 365 | 0.87 | 0.08 | under 0.0001 | Over the last 365 sessions, ANY's daily log return moved 0.87 for each 1.00 of bitcoin's (95% interval 0.57 to 1.16; r 0.29, R squared 0.08, p under 0.0001, n 365). | article candidate |
| concordance:DGXX:365price | 2026-09-22 | 365 | 25.65 | not computed | under 0.0001 | Over the last 365 sessions, DGXX closed up on 62% of the days bitcoin closed up and on 36% of the days it did not (chi-square 25.65, p under 0.0001, n 365). | article candidate |
| concordance:BGDE:365price | 2026-09-22 | 365 | 18.58 | not computed | under 0.0001 | Over the last 365 sessions, BGDE closed up on 53% of the days bitcoin closed up and on 31% of the days it did not (chi-square 18.58, p under 0.0001, n 365). | article candidate |
| leverage:ARBK:365price | 2026-09-22 | 365 | 0.97 | 0.05 | under 0.0001 | Over the last 365 sessions, ARBK's daily log return moved 0.97 for each 1.00 of bitcoin's (95% interval 0.51 to 1.42; r 0.22, R squared 0.05, p under 0.0001, n 365). | article candidate |
| concordance:ARBK:365price | 2026-09-22 | 365 | 12.74 | not computed | 0.0004 | Over the last 365 sessions, ARBK closed up on 50% of the days bitcoin closed up and on 31% of the days it did not (chi-square 12.74, p 0.0004, n 365). | article candidate |
| indicator:sma50_over_sma200:21indicator | 2026-09-22 | 8,834 | -0.05 | not computed | under 0.0001 | Pooled across 18 miners, the 21-session forward log return averaged -0.06 on sessions with 50-session SMA above the 200-session SMA and -0.00 without it (difference -0.05, Welch t -11.48, p under 0.0001, n 8834 against 11716; overlapping windows, so p overstates). | no |
| indicator:macd_hist_negative:21indicator | 2026-09-22 | 10,945 | -0.04 | not computed | under 0.0001 | Pooled across 18 miners, the 21-session forward log return averaged -0.05 on sessions with MACD histogram negative and -0.00 without it (difference -0.04, Welch t -9.88, p under 0.0001, n 10945 against 12593; overlapping windows, so p overstates). | no |
| indicator:macd_hist_positive:21indicator | 2026-09-22 | 12,593 | 0.04 | not computed | under 0.0001 | Pooled across 18 miners, the 21-session forward log return averaged -0.00 on sessions with MACD histogram positive and -0.05 without it (difference 0.04, Welch t 9.88, p under 0.0001, n 12593 against 10945; overlapping windows, so p overstates). | no |
| d_revenue_vs_d_cost_of_revenuetime-series | 2026-09-22 | 176 | 0.59 | 0.34 | under 0.0001 | Across the 176 quarter-pairs pooled across companies, quarter-to-quarter log change in revenue and quarter-to-quarter log change in cost of revenue move together (r 0.59, R squared 0.34, p under 0.0001, n 176). | no |
| indicator:close_over_cloud:21indicator | 2026-09-22 | 7,680 | -0.03 | not computed | under 0.0001 | Pooled across 18 miners, the 21-session forward log return averaged -0.04 on sessions with close above the Ichimoku cloud and -0.01 without it (difference -0.03, Welch t -6.13, p under 0.0001, n 7680 against 15066; overlapping windows, so p overstates). | no |
| indicator:sma50_over_sma200:5indicator | 2026-09-22 | 9,045 | -0.01 | not computed | under 0.0001 | Pooled across 18 miners, the 5-session forward log return averaged -0.01 on sessions with 50-session SMA above the 200-session SMA and -0.00 without it (difference -0.01, Welch t -6.01, p under 0.0001, n 9045 against 11793; overlapping windows, so p overstates). | no |
| indicator:close_under_bb_lower:21indicator | 2026-09-22 | 1,060 | 0.05 | not computed | under 0.0001 | Pooled across 18 miners, the 21-session forward log return averaged 0.03 on sessions with close below the lower Bollinger band and -0.02 without it (difference 0.05, Welch t 4.68, p under 0.0001, n 1060 against 22730; overlapping windows, so p overstates). | no |
| indicator:rsi_over_70:21indicator | 2026-09-22 | 1,449 | -0.04 | not computed | under 0.0001 | Pooled across 18 miners, the 21-session forward log return averaged -0.06 on sessions with RSI(14) over 70 and -0.02 without it (difference -0.04, Welch t -4.44, p under 0.0001, n 1449 against 22431; overlapping windows, so p overstates). | no |
| indicator:rsi_under_30:5indicator | 2026-09-22 | 1,058 | 0.03 | not computed | under 0.0001 | Pooled across 18 miners, the 5-session forward log return averaged 0.02 on sessions with RSI(14) under 30 and -0.00 without it (difference 0.03, Welch t 4.42, p under 0.0001, n 1058 against 23110; overlapping windows, so p overstates). | no |
| indicator:rsi_under_30:21indicator | 2026-09-22 | 1,058 | 0.05 | not computed | 0.0007 | Pooled across 18 miners, the 21-session forward log return averaged 0.02 on sessions with RSI(14) under 30 and -0.02 without it (difference 0.05, Welch t 3.41, p 0.0007, n 1058 against 22822; overlapping windows, so p overstates). | no |
| d_assets_vs_d_sharestime-series | 2026-09-22 | 254 | 0.15 | 0.02 | 0.014 | Across the 254 quarter-pairs pooled across companies, quarter-to-quarter log change in total assets and quarter-to-quarter log change in shares outstanding lean together, at a strength that would arise by chance about once in 100 (r 0.15, R squared 0.02, p 0.014, n 254). | no |
| indicator:close_under_bb_lower:5indicator | 2026-09-22 | 1,064 | 0.01 | not computed | 0.018 | Pooled across 18 miners, the 5-session forward log return averaged 0.01 on sessions with close below the lower Bollinger band and -0.00 without it (difference 0.01, Welch t 2.37, p 0.018, n 1064 against 23014; overlapping windows, so p overstates). | no |
| indicator:close_over_bb_upper:21indicator | 2026-09-22 | 1,776 | 0.02 | not computed | 0.023 | Pooled across 18 miners, the 21-session forward log return averaged -0.00 on sessions with close above the upper Bollinger band and -0.02 without it (difference 0.02, Welch t 2.28, p 0.023, n 1776 against 22014; overlapping windows, so p overstates). | no |
| hashrate_vs_revenue:log10cross-section | 2026-09-22 | 10 | 0.69 | 0.47 | 0.028 | Across the 10 miners with both on file, log realized hashrate and log annualized revenue lean together, at a strength that would arise by chance about 3 times in 100 (r 0.69, R squared 0.47, p 0.028, n 10). | no |
| indicator:close_over_sma50:21indicator | 2026-09-22 | 10,020 | -0.01 | not computed | 0.036 | Pooled across 18 miners, the 21-session forward log return averaged -0.03 on sessions with close above the 50-session SMA and -0.02 without it (difference -0.01, Welch t -2.10, p 0.036, n 10020 against 13230; overlapping windows, so p overstates). | no |
| indicator:close_under_sma50:21indicator | 2026-09-22 | 13,230 | 0.01 | not computed | 0.036 | Pooled across 18 miners, the 21-session forward log return averaged -0.02 on sessions with close below the 50-session SMA and -0.03 without it (difference 0.01, Welch t 2.10, p 0.036, n 13230 against 10020; overlapping windows, so p overstates). | no |
| indicator:close_over_cloud:5indicator | 2026-09-22 | 7,688 | -0.00 | not computed | 0.041 | Pooled across 18 miners, the 5-session forward log return averaged -0.01 on sessions with close above the Ichimoku cloud and -0.00 without it (difference -0.00, Welch t -2.05, p 0.041, n 7688 against 15346; overlapping windows, so p overstates). | no |
| market_cap_vs_ev_per_eh:log10cross-section | 2026-09-22 | 6 | 0.80 | 0.63 | 0.058 | Across the 6 miners with both on file, log market capitalization and log EV per EH/s show no measurable relation at this sample size (r 0.80, R squared 0.63, p 0.058, n 6). With 6 points one company can move the result. | no |
| d_revenue_vs_d_sharestime-series | 2026-09-22 | 229 | 0.11 | 0.01 | 0.096 | Across the 229 quarter-pairs pooled across companies, quarter-to-quarter log change in revenue and quarter-to-quarter log change in shares outstanding show no measurable relation at this sample size (r 0.11, R squared 0.01, p 0.096, n 229). | no |
| indicator:close_over_bb_upper:5indicator | 2026-09-22 | 1,783 | 0.01 | not computed | 0.112 | Pooled across 18 miners, the 5-session forward log return averaged 0.00 on sessions with close above the upper Bollinger band and -0.00 without it (difference 0.01, Welch t 1.59, p 0.112, n 1783 against 22295; overlapping windows, so p overstates). | no |
| hashrate_vs_revenue:levelcross-section | 2026-09-22 | 10 | 0.53 | 0.28 | 0.112 | Across the 10 miners with both on file, realized hashrate and annualized revenue show no measurable relation at this sample size (r 0.53, R squared 0.28, p 0.112, n 10). | no |
| indicator:macd_hist_negative:5indicator | 2026-09-22 | 10,998 | -0.00 | not computed | 0.145 | Pooled across 18 miners, the 5-session forward log return averaged -0.01 on sessions with MACD histogram negative and -0.00 without it (difference -0.00, Welch t -1.46, p 0.145, n 10998 against 12828; overlapping windows, so p overstates). | no |
| indicator:macd_hist_positive:5indicator | 2026-09-22 | 12,828 | 0.00 | not computed | 0.145 | Pooled across 18 miners, the 5-session forward log return averaged -0.00 on sessions with MACD histogram positive and -0.01 without it (difference 0.00, Welch t 1.46, p 0.145, n 12828 against 10998; overlapping windows, so p overstates). | no |
| btc_held_vs_xbrl_units:levelcross-section | 2026-09-22 | 5 | 0.73 | 0.53 | 0.161 | Across the 5 miners with both on file, bitcoin held and the XBRL bitcoin count show no measurable relation at this sample size (r 0.73, R squared 0.53, p 0.161, n 5). With 5 points one company can move the result. | no |
| market_cap_vs_hashrate:log10cross-section | 2026-09-22 | 9 | -0.48 | 0.23 | 0.189 | Across the 9 miners with both on file, log market capitalization and log realized hashrate show no measurable relation at this sample size (r -0.48, R squared 0.23, p 0.189, n 9). | no |
| market_cap_vs_ev_per_eh:levelcross-section | 2026-09-22 | 6 | 0.62 | 0.38 | 0.190 | Across the 6 miners with both on file, market capitalization and EV per EH/s show no measurable relation at this sample size (r 0.62, R squared 0.38, p 0.190, n 6). With 6 points one company can move the result. | no |
| efficiency_vs_direct_cost:levelcross-section | 2026-09-22 | 6 | 0.61 | 0.37 | 0.199 | Across the 6 miners with both on file, fleet efficiency in J/TH and direct cost per bitcoin show no measurable relation at this sample size (r 0.61, R squared 0.37, p 0.199, n 6). With 6 points one company can move the result. | no |
| market_cap_vs_hashrate:levelcross-section | 2026-09-22 | 9 | -0.47 | 0.22 | 0.203 | Across the 9 miners with both on file, market capitalization and realized hashrate show no measurable relation at this sample size (r -0.47, R squared 0.22, p 0.203, n 9). | no |
| indicator:close_under_cloud:21indicator | 2026-09-22 | 12,068 | 0.01 | not computed | 0.231 | Pooled across 18 miners, the 21-session forward log return averaged -0.02 on sessions with close below the Ichimoku cloud and -0.02 without it (difference 0.01, Welch t 1.20, p 0.231, n 12068 against 10678; overlapping windows, so p overstates). | no |
| hashrate_vs_btc_held:levelcross-section | 2026-09-22 | 8 | 0.47 | 0.22 | 0.243 | Across the 8 miners with both on file, realized hashrate and bitcoin held show no measurable relation at this sample size (r 0.47, R squared 0.22, p 0.243, n 8). | no |
| indicator:close_under_cloud:5indicator | 2026-09-22 | 12,294 | -0.00 | not computed | 0.252 | Pooled across 18 miners, the 5-session forward log return averaged -0.01 on sessions with close below the Ichimoku cloud and -0.00 without it (difference -0.00, Welch t -1.15, p 0.252, n 12294 against 10740; overlapping windows, so p overstates). | no |
| indicator:close_over_sma50:5indicator | 2026-09-22 | 10,098 | 0.00 | not computed | 0.301 | Pooled across 18 miners, the 5-session forward log return averaged -0.00 on sessions with close above the 50-session SMA and -0.01 without it (difference 0.00, Welch t 1.03, p 0.301, n 10098 against 13440; overlapping windows, so p overstates). | no |
| indicator:close_under_sma50:5indicator | 2026-09-22 | 13,440 | -0.00 | not computed | 0.301 | Pooled across 18 miners, the 5-session forward log return averaged -0.01 on sessions with close below the 50-session SMA and -0.00 without it (difference -0.00, Welch t -1.03, p 0.301, n 13440 against 10098; overlapping windows, so p overstates). | no |
| d_btc_value_vs_d_revenuetime-series | 2026-09-22 | 81 | 0.08 | 0.01 | 0.455 | Across the 81 quarter-pairs pooled across companies, quarter-to-quarter log change in digital assets at fair value and quarter-to-quarter log change in revenue show no measurable relation at this sample size (r 0.08, R squared 0.01, p 0.455, n 81). | no |
| hashrate_vs_btc_held:log10cross-section | 2026-09-22 | 7 | 0.32 | 0.10 | 0.486 | Across the 7 miners with both on file, log realized hashrate and log bitcoin held show no measurable relation at this sample size (r 0.32, R squared 0.10, p 0.486, n 7). With 7 points one company can move the result. | no |
| btc_held_vs_xbrl_units:log10cross-section | 2026-09-22 | 4 | 0.50 | 0.25 | 0.503 | Across the 4 miners with both on file, log bitcoin held and log the XBRL bitcoin count show no measurable relation at this sample size (r 0.50, R squared 0.25, p 0.503, n 4). With 4 points one company can move the result. | no |
| market_cap_vs_btc_held:levelcross-section | 2026-09-22 | 8 | -0.26 | 0.07 | 0.538 | Across the 8 miners with both on file, market capitalization and bitcoin held show no measurable relation at this sample size (r -0.26, R squared 0.07, p 0.538, n 8). | no |
| indicator:rsi_over_70:5indicator | 2026-09-22 | 1,449 | -0.00 | not computed | 0.587 | Pooled across 18 miners, the 5-session forward log return averaged -0.01 on sessions with RSI(14) over 70 and -0.00 without it (difference -0.00, Welch t -0.54, p 0.587, n 1449 against 22719; overlapping windows, so p overstates). | no |
| market_cap_vs_btc_held:log10cross-section | 2026-09-22 | 7 | 0.20 | 0.04 | 0.671 | Across the 7 miners with both on file, log market capitalization and log bitcoin held show no measurable relation at this sample size (r 0.20, R squared 0.04, p 0.671, n 7). With 7 points one company can move the result. | no |
| revenue_per_mw_vs_efficiency:levelcross-section | 2026-09-22 | 8 | 0.13 | 0.02 | 0.760 | Across the 8 miners with both on file, annualized revenue per MW and fleet efficiency in J/TH show no measurable relation at this sample size (r 0.13, R squared 0.02, p 0.760, n 8). | no |
| direct_cost_vs_margin:levelcross-section | 2026-09-22 | 4 | 0.02 | 0.00 | 0.977 | Across the 4 miners with both on file, direct cost per bitcoin and gross margin show no measurable relation at this sample size (r 0.02, R squared 0.00, p 0.977, n 4). With 4 points one company can move the result. | no |
| all_in_cost_vs_margin:levelcross-section | 2026-09-22 | 2 | not computed | not computed | not computed | all-in cost per bitcoin and gross margin: the 2 miners with both on file are too few for a test. | no |
| matrix:minersprice | 2026-09-22 | 153 | 0.49 | not computed | not computed | Across 153 miner pairs over the last 365 common sessions, the daily-return correlations run from 0.13 to 0.84 (mean 0.49, n 153 pairs). | no |
Method
MethodologyWhat was computed, from which files, under which rules.
- Inputs: 38 files under data/, each named with its hash in the document; the research packs of 2026-09-06 (Tier 1) and 2026-09-06 (Tier 2), and each company's XBRL facts read under the period rule on the methodology page.
- Sample statistics throughout: the variance divides by n minus 1; the regression has n minus 2 degrees of freedom; p-values are two-sided from the Student t distribution; Spearman's p uses the t approximation on the rank correlation.
- A test needs 3 points; under 8 it is shown with the caveat that one company can move it; a size pair is also tested on log10 of both figures, and a non-positive value is dropped and named.
- Prices: data/prices_history.json, written by scripts/prices.py, each close tagged with its source; bitcoin for every equity comparison is the 16:00 ET sample on NYSE session dates (spec 5.5). Every indicator is computed on closes, and the site's own indicator code is checked against the job's fixture to four decimals.
- The job is
site/scripts/analysis.pywith returns_study.py, standard library only, run nightly by the analysis workflow; version 2e81c67f5cd6 at fd011aa. It forecasts nothing and writes no sentence that says what a relation means.