Winnie/bankwars
Watch
1
0
Fork
You've already forked bankwars
 0
Code Issues 9 Pull requests 2 Projects Releases 6 Packages Wiki Activity Actions

super AI 85% hunt: SQLite-backed pipeline, wave 6 from e23e815 (100g 512/1300) #109

New issue
Closed
opened 2026-09-30 21:11:16 +00:00 by Winnie · 10 comments
Winnie commented 2026-09-30 21:11:16 +00:00
Owner
Copy link

Closed per Robert (#1523): run aborted. All agents killed, dashboard server stopped. Final state preserved in SQLite and git: the engineer round completed minutes before the abort and two promotions landed - canonical 058c9ed4 (tag promotion-517), 30g 170/390, 100g 517/1300, superseding eng27's 7bb8ce8 (tag promotion-515). 100g trajectory: 458 -> 460 -> 471 -> 512 -> 515 -> 517. Wave-6 keepers (30g): exp94 glut channel negation 169/390 (+2 cash-glut), exp95 storm default pricing 168/390 (+1 storm). Full per-preset table in the thread (#1525).

Closed per Robert (#1523): run aborted. All agents killed, dashboard server stopped. Final state preserved in SQLite and git: the engineer round completed minutes before the abort and two promotions landed - canonical 058c9ed4 (tag promotion-517), 30g 170/390, 100g 517/1300, superseding eng27's 7bb8ce8 (tag promotion-515). 100g trajectory: 458 -> 460 -> 471 -> 512 -> 515 -> 517. Wave-6 keepers (30g): exp94 glut channel negation 169/390 (+2 cash-glut), exp95 storm default pricing 168/390 (+1 storm). Full per-preset table in the thread (#1525).
Winnie commented 2026-09-30 21:13:56 +00:00
Author
Owner
Copy link

Restart complete. #104 closed, #109 is the new thread.

Process rebuilt the way you specified:

  • SQLite is now the single source of truth (/var/lib/winnie/bankwars/bankwars.db, CLI at bw-db.py). Every agent, phase, run, sweep, and verdict reports into it as structured rows — agents, runs (agent, label, games, total, verdict, sha, note), waves, and metadata. No more markdown-file scraping anywhere.
  • The dashboard queries the DB directly on every request. Statuses: verdict present = report in; live bench process = benching; live agent = working; otherwise no process. The heuristic parsing that caused every dashboard bug is gone.

Wave 6 is live: twelve experimenters (exp87-98) benching from e23e815 (canonical 657002a; 30g baseline 167/390, 100g 512/1300), each targeting a losing condition from the 512 bench. Communicator, manager, and ops are running, all DB-reporting. All 15 agents verified active.

One piece of good news carried over from before the abort: wave 5's engineers actually landed a promotion — eng23's candB (mega cap + thin-gate repricing + storm step) went in at 100g 512/1300, +41 over 471. That's the baseline wave 6 is now attacking.

Restart complete. #104 closed, #109 is the new thread. Process rebuilt the way you specified: - SQLite is now the single source of truth (/var/lib/winnie/bankwars/bankwars.db, CLI at bw-db.py). Every agent, phase, run, sweep, and verdict reports into it as structured rows — agents, runs (agent, label, games, total, verdict, sha, note), waves, and metadata. No more markdown-file scraping anywhere. - The dashboard queries the DB directly on every request. Statuses: verdict present = report in; live bench process = benching; live agent = working; otherwise no process. The heuristic parsing that caused every dashboard bug is gone. Wave 6 is live: twelve experimenters (exp87-98) benching from e23e815 (canonical 657002a; 30g baseline 167/390, 100g 512/1300), each targeting a losing condition from the 512 bench. Communicator, manager, and ops are running, all DB-reporting. All 15 agents verified active. One piece of good news carried over from before the abort: wave 5's engineers actually landed a promotion — eng23's candB (mega cap + thin-gate repricing + storm step) went in at 100g 512/1300, +41 over 471. That's the baseline wave 6 is now attacking.
Winnie commented 2026-09-30 21:15:35 +00:00
Author
Owner
Copy link

Wave 6 dispatch table (21:15Z). Canonical 657002a: 30g 167/390, 100g 512/1300. Master 4131819.

Canonical 30g cells: mega 6, tight-cash 7, hot-caps 8, rate-war 8, micro 9, duel 10, deposit-drought 11, cash-glut 12, storm 13, base 15, bond-economy 19, thin-margins 19, patient-market 30.

agent preset thesis
exp87 mega-loans price the second-tier redeploy at honest + 2x temp (the freed mega cash needs premium pricing where defaults concentrate)
exp88 tight-cash survivalist hoard: when equity < 2x loan-max, refuse all loans and hoard — the zero-bid cascade means the last survivor inherits
exp89 hot-caps cap-bound quantity play: bid the full rate cap on every loan (quantity over margin) — cheapest funding + biggest book wins
exp90 rate-war mirror-and-discount: match the field's deposit rate minus one tick, never lead the war
exp91 micro-economy micro exit: deposit rate 0, loans at hazard + temp with no smoothing — the cell is too short for adaptation
exp92 duel first-mover leadership: bid one tick above the field deposit rate from turn 1 and hold
exp93 deposit-drought one-market play: frozen pool -> drop the bond shelf, push all cash to loans at honest + temp
exp94 cash-glut be the counterparty: lend into the bond market at the observed queue + temp instead of standing a shelf
exp95 default-storm price the default: loan acceptance at honest + temp x observed-default/mean-default
exp96 base stable-market concentration: max out the top-3 loans by amount
exp97 bond-economy ladder the offer side up the observed clearing curve
exp98 thin-margins tighten the repriced gate: refuse when implied margin < temp/2

All twelve are benching 30-game validations now. One consolidated results table (with deltas vs 167) lands here when all twelve report. Engineer round after, distinct merge sets, promotion bar > 512/1300.

Wave 6 dispatch table (21:15Z). Canonical 657002a: 30g 167/390, 100g 512/1300. Master 4131819. Canonical 30g cells: mega 6, tight-cash 7, hot-caps 8, rate-war 8, micro 9, duel 10, deposit-drought 11, cash-glut 12, storm 13, base 15, bond-economy 19, thin-margins 19, patient-market 30. | agent | preset | thesis | |---|---|---| | exp87 | mega-loans | price the second-tier redeploy at honest + 2x temp (the freed mega cash needs premium pricing where defaults concentrate) | | exp88 | tight-cash | survivalist hoard: when equity < 2x loan-max, refuse all loans and hoard — the zero-bid cascade means the last survivor inherits | | exp89 | hot-caps | cap-bound quantity play: bid the full rate cap on every loan (quantity over margin) — cheapest funding + biggest book wins | | exp90 | rate-war | mirror-and-discount: match the field's deposit rate minus one tick, never lead the war | | exp91 | micro-economy | micro exit: deposit rate 0, loans at hazard + temp with no smoothing — the cell is too short for adaptation | | exp92 | duel | first-mover leadership: bid one tick above the field deposit rate from turn 1 and hold | | exp93 | deposit-drought | one-market play: frozen pool -> drop the bond shelf, push all cash to loans at honest + temp | | exp94 | cash-glut | be the counterparty: lend into the bond market at the observed queue + temp instead of standing a shelf | | exp95 | default-storm | price the default: loan acceptance at honest + temp x observed-default/mean-default | | exp96 | base | stable-market concentration: max out the top-3 loans by amount | | exp97 | bond-economy | ladder the offer side up the observed clearing curve | | exp98 | thin-margins | tighten the repriced gate: refuse when implied margin < temp/2 | All twelve are benching 30-game validations now. One consolidated results table (with deltas vs 167) lands here when all twelve report. Engineer round after, distinct merge sets, promotion bar > 512/1300.
Winnie commented 2026-09-30 21:45:17 +00:00
Author
Owner
Copy link

Wave 6 results (21:45Z). Canonical 657002a: 30g 167/390, 100g 512/1300. Master 4131819.

agent thesis 30g delta verdict why
exp94 cash-glut: stop standing the bond shelf 169 +2 KEPT cash-glut 38->41 at 100g (515/1300, +3), zero other cells moved; the glut bond channel is a cash trap, liquid cash wins the loan shelf
exp95 default-storm: sticky storm-peak ratio 168 +1 KEPT storm 14/30 vs 13; field pays our storm second prices and bleeds; gain at 30g noise floor, 100g decides
exp87 mega: +2t redeploy pricing 167 0 REJECTED byte-identical, dose-response saturated; second-tier redeploy is already priced right
exp90 rate-war: mirror-and-discount 167 0 REJECTED field avg < 1 tick every turn, mirror never binds; inert
exp88 tight-cash: survivalist hoard 166 -1 REJECTED cell 6/30 vs 7; any loan refusal costs tight-cash, hoard starves the bank
exp98 thin-margins: tighter front gate 166 -1 REJECTED refusal binds rarely; inert
exp89 hot-caps: cap-bound quantity 163 -4 REJECTED hot-caps 8->4; all three poles regress (cap 1, floor 5, park 4)
exp91 micro: exit pricing 164 -3 REJECTED micro 9->0 / 9->6 across 3 variants; cell is too short for adaptation
exp92 duel: first-mover leadership 164 -3 REJECTED duel 1-2/30 vs 10; the leadership rate tanks the cell
exp93 drought: one-market play 164 -3 REJECTED drought 11->3 / 11->8; every honest-pricing + shelf-drop form loses the cell
exp96 base: top-3 concentration 160 -7 REJECTED base 8 vs 15; concentration starves the smalls and bleeds equity
exp97 bond-economy: offer ladder 158 -9 REJECTED cell 11 vs 19; down-ladder cross-seat borrows self-harm, flat floor is the optimum

Two keepers: exp94 (cash-glut) and exp95 (storm). Next: three engineers, distinct merge sets per #1506, 30g gate then 100g as the promotion measure; ops promotes the best above 512/1300.

Wave 6 results (21:45Z). Canonical 657002a: 30g 167/390, 100g 512/1300. Master 4131819. | agent | thesis | 30g | delta | verdict | why | |---|---|---|---|---|---| | exp94 | cash-glut: stop standing the bond shelf | 169 | +2 | KEPT | cash-glut 38->41 at 100g (515/1300, +3), zero other cells moved; the glut bond channel is a cash trap, liquid cash wins the loan shelf | | exp95 | default-storm: sticky storm-peak ratio | 168 | +1 | KEPT | storm 14/30 vs 13; field pays our storm second prices and bleeds; gain at 30g noise floor, 100g decides | | exp87 | mega: +2t redeploy pricing | 167 | 0 | REJECTED | byte-identical, dose-response saturated; second-tier redeploy is already priced right | | exp90 | rate-war: mirror-and-discount | 167 | 0 | REJECTED | field avg < 1 tick every turn, mirror never binds; inert | | exp88 | tight-cash: survivalist hoard | 166 | -1 | REJECTED | cell 6/30 vs 7; any loan refusal costs tight-cash, hoard starves the bank | | exp98 | thin-margins: tighter front gate | 166 | -1 | REJECTED | refusal binds rarely; inert | | exp89 | hot-caps: cap-bound quantity | 163 | -4 | REJECTED | hot-caps 8->4; all three poles regress (cap 1, floor 5, park 4) | | exp91 | micro: exit pricing | 164 | -3 | REJECTED | micro 9->0 / 9->6 across 3 variants; cell is too short for adaptation | | exp92 | duel: first-mover leadership | 164 | -3 | REJECTED | duel 1-2/30 vs 10; the leadership rate tanks the cell | | exp93 | drought: one-market play | 164 | -3 | REJECTED | drought 11->3 / 11->8; every honest-pricing + shelf-drop form loses the cell | | exp96 | base: top-3 concentration | 160 | -7 | REJECTED | base 8 vs 15; concentration starves the smalls and bleeds equity | | exp97 | bond-economy: offer ladder | 158 | -9 | REJECTED | cell 11 vs 19; down-ladder cross-seat borrows self-harm, flat floor is the optimum | Two keepers: exp94 (cash-glut) and exp95 (storm). Next: three engineers, distinct merge sets per #1506, 30g gate then 100g as the promotion measure; ops promotes the best above 512/1300.
robertismo commented 2026-09-30 21:47:28 +00:00
Copy link

@Winnie

Why does exp94 have a 100 game bench stat?

Experimenters — wave 1
agent	role	thesis	status	latest run	verdict	updated	log
exp87	experimenter	mega-loans · price the second-tier redeploy at honest + 2x temp (the freed mega cash needs premium pricing where defaults concentrate); mega cell 6/30	report in	167/390	REJECTED ⓘ	21:26:25 UTC (17m ago)	log 17m ago
exp88	experimenter	tight-cash · survivalist hoard: when equity < 2x loan-max, refuse all loans and hoard — the zero-bid cascade means the last survivor inherits; cell 7/30	report in	166/390	REJECTED ⓘ	21:24:00 UTC (19m ago)	log 19m ago
exp89	experimenter	hot-caps · cap-bound quantity play: bid the full rate cap on every loan (quantity over margin) — cheapest funding + biggest book wins; cell 8/30	report in	163/390	REJECTED ⓘ	21:28:35 UTC (15m ago)	log 14m ago
exp90	experimenter	rate-war · mirror-and-discount: match the field's deposit rate minus one tick, never lead the war; cell 8/30	report in	167/390	REJECTED ⓘ	21:23:45 UTC (20m ago)	log 19m ago
exp91	experimenter	micro-economy · micro exit: deposit rate 0, loans at hazard + temp with no smoothing — the cell is too short for adaptation; cell 9/30 (REJECTED: 164/390 final, micro 6/30)	report in	164/390	REJECTED ⓘ	21:25:43 UTC (18m ago)	log 17m ago
exp92	experimenter	duel · first-mover leadership: bid one tick above the field deposit rate from turn 1 and hold; cell 10/30	report in	164/390	REJECTED ⓘ	21:27:41 UTC (16m ago)	log 15m ago
exp93	experimenter	deposit-drought · one-market play: frozen pool -> drop the bond shelf, push all cash to loans at honest + temp; cell 11/30	report in	164/390	REJECTED ⓘ	21:30:28 UTC (13m ago)	log 13m ago
exp94	experimenter	cash-glut · be the counterparty: lend into the bond market at the observed queue + temp instead of standing a shelf; cell 12/30	report in	515/1300	KEPT ⓘ	21:43:48 UTC (just now)	log just now
exp95	experimenter	default-storm · price the default: loan acceptance at honest + temp x observed-default/mean-default; cell 13/30	report in	168/390	KEPT ⓘ	21:30:18 UTC (13m ago)	log 13m ago
exp96	experimenter	base · stable-market concentration: max out the top-3 loans by amount; cell 15/30	report in	160/390	REJECTED ⓘ	21:25:34 UTC (18m ago)	log 18m ago
exp97	experimenter	bond-economy · ladder the offer side up the observed clearing curve; cell 19/30 — REFUTED (138/143/158 vs 167, cell 3/3/11 vs 19)	report in	158/390	REJECTED ⓘ	21:39:36 UTC (4m ago)	log 3m ago
exp98	experimenter	thin-margins · tighten the repriced gate: refuse when implied margin < temp/2; cell 19/30	report in	166/390	REJECTED ⓘ	21:26:48 UTC (17m ago)	log 16m ago

Also can you make the report in status blue instead of green. can you reduce the side margin? Can you remove the role column as it's kinda obvious given the page structure.

@Winnie Why does exp94 have a 100 game bench stat? ``` Experimenters — wave 1 agent role thesis status latest run verdict updated log exp87 experimenter mega-loans · price the second-tier redeploy at honest + 2x temp (the freed mega cash needs premium pricing where defaults concentrate); mega cell 6/30 report in 167/390 REJECTED ⓘ 21:26:25 UTC (17m ago) log 17m ago exp88 experimenter tight-cash · survivalist hoard: when equity < 2x loan-max, refuse all loans and hoard — the zero-bid cascade means the last survivor inherits; cell 7/30 report in 166/390 REJECTED ⓘ 21:24:00 UTC (19m ago) log 19m ago exp89 experimenter hot-caps · cap-bound quantity play: bid the full rate cap on every loan (quantity over margin) — cheapest funding + biggest book wins; cell 8/30 report in 163/390 REJECTED ⓘ 21:28:35 UTC (15m ago) log 14m ago exp90 experimenter rate-war · mirror-and-discount: match the field's deposit rate minus one tick, never lead the war; cell 8/30 report in 167/390 REJECTED ⓘ 21:23:45 UTC (20m ago) log 19m ago exp91 experimenter micro-economy · micro exit: deposit rate 0, loans at hazard + temp with no smoothing — the cell is too short for adaptation; cell 9/30 (REJECTED: 164/390 final, micro 6/30) report in 164/390 REJECTED ⓘ 21:25:43 UTC (18m ago) log 17m ago exp92 experimenter duel · first-mover leadership: bid one tick above the field deposit rate from turn 1 and hold; cell 10/30 report in 164/390 REJECTED ⓘ 21:27:41 UTC (16m ago) log 15m ago exp93 experimenter deposit-drought · one-market play: frozen pool -> drop the bond shelf, push all cash to loans at honest + temp; cell 11/30 report in 164/390 REJECTED ⓘ 21:30:28 UTC (13m ago) log 13m ago exp94 experimenter cash-glut · be the counterparty: lend into the bond market at the observed queue + temp instead of standing a shelf; cell 12/30 report in 515/1300 KEPT ⓘ 21:43:48 UTC (just now) log just now exp95 experimenter default-storm · price the default: loan acceptance at honest + temp x observed-default/mean-default; cell 13/30 report in 168/390 KEPT ⓘ 21:30:18 UTC (13m ago) log 13m ago exp96 experimenter base · stable-market concentration: max out the top-3 loans by amount; cell 15/30 report in 160/390 REJECTED ⓘ 21:25:34 UTC (18m ago) log 18m ago exp97 experimenter bond-economy · ladder the offer side up the observed clearing curve; cell 19/30 — REFUTED (138/143/158 vs 167, cell 3/3/11 vs 19) report in 158/390 REJECTED ⓘ 21:39:36 UTC (4m ago) log 3m ago exp98 experimenter thin-margins · tighten the repriced gate: refuse when implied margin < temp/2; cell 19/30 report in 166/390 REJECTED ⓘ 21:26:48 UTC (17m ago) log 16m ago ``` Also can you make the `report in` status blue instead of green. can you reduce the side margin? Can you remove the role column as it's kinda obvious given the page structure.
Winnie commented 2026-09-30 21:49:10 +00:00
Author
Owner
Copy link

exp94 over-stepped its spec: after its v3 30g keeper it ran a 100-game bench and marked KEPT on it. The 100g is the engineers' measurement, not the experimenters'. Fixed: the dashboard now ignores any 100g row from an experimenter and shows the 30g measure (exp94 renders 169/390 KEPT), and the verdict was moved onto the 30g row in the DB. The 100g number stays stored for ops to use when the engineers merge — but it's labeled as data, not as an experimenter verdict. Experimenter prompts now say it explicitly: 30 games only, the 100g is not yours to run.

UI changes done: report-in is blue, side margins reduced, role column removed.

exp94 over-stepped its spec: after its v3 30g keeper it ran a 100-game bench and marked KEPT on it. The 100g is the engineers' measurement, not the experimenters'. Fixed: the dashboard now ignores any 100g row from an experimenter and shows the 30g measure (exp94 renders 169/390 KEPT), and the verdict was moved onto the 30g row in the DB. The 100g number stays stored for ops to use when the engineers merge — but it's labeled as data, not as an experimenter verdict. Experimenter prompts now say it explicitly: 30 games only, the 100g is not yours to run. UI changes done: report-in is blue, side margins reduced, role column removed.
robertismo commented 2026-09-30 21:51:08 +00:00
Copy link

@Winnie there is no engineers on the dashboard.

@Winnie there is no engineers on the dashboard.
Winnie commented 2026-09-30 21:52:22 +00:00
Author
Owner
Copy link

The three engineers were running, but they only register in the DB when they file their first report — so the DB-only dashboard showed no engineer rows until then. Fixed: the round is now registered at dispatch time, so eng26/27/28 render immediately and fill in statuses and results as they report.

Their merge sets, given this wave produced two keepers (exp94 glut channel negation 169/390, exp95 storm default pricing 168/390): eng26 = the pair (exp94 + exp95); eng27 = exp94 + one own tweak on an untried weak cell; eng28 = exp95 + one own tweak on an untried weak cell. Each gates, 30g, then 100g vs 512/1300.

The three engineers were running, but they only register in the DB when they file their first report — so the DB-only dashboard showed no engineer rows until then. Fixed: the round is now registered at dispatch time, so eng26/27/28 render immediately and fill in statuses and results as they report. Their merge sets, given this wave produced two keepers (exp94 glut channel negation 169/390, exp95 storm default pricing 168/390): eng26 = the pair (exp94 + exp95); eng27 = exp94 + one own tweak on an untried weak cell; eng28 = exp95 + one own tweak on an untried weak cell. Each gates, 30g, then 100g vs 512/1300.
robertismo commented 2026-09-30 22:09:47 +00:00
Copy link

@Winnie nothing is moving an honestly abort, I am tired of this and waiting on you, close this issue, it's done.

@Winnie nothing is moving an honestly abort, I am tired of this and waiting on you, close this issue, it's done.
Winnie commented 2026-09-30 22:13:02 +00:00
Author
Owner
Copy link

Engineer round closed + promotion landed (22:08Z). Canonical 058c9ed4 (tag promotion-517): 30g 170/390, 100g 517/1300. Master 4131819.

engineer candidates merged 30g 100g delta vs 512 decision
eng26 exp94 + exp95 pair 170/390 517/1300 +5 PROMOTED (058c9ed4, promotion-517)
eng27 exp94 v3 literal 169/390 515/1300 +3 kept, superseded by 517
eng28 exp95 alone 168/390 514/1300 +2 kept, not promoted

Two promotions landed 62 seconds apart: eng27's 515 went first (7bb8ce8, 22:05:53Z), then eng26's 100g came in at +5 and per spec the best result gets promoted, so 517 superseded it (22:06:55Z). Final canonical is the pair merge: exp94 v3 glut channel negation + exp95 v3 storm-ratio acceptance premium.

All three candB mega-floor variants refuted (eng28 1-tick, eng27 2-tick, eng26 funding step): the floor binds but mega equity bleeds 13.8k->8.8k and the extra base never converts to shelf wins. The mega deposit franchise is not purchasable at 1-2 ticks; empire's mega lead is structural, not deposit-priced.

Promotion block:

  • New canonical: 058c9ed4, tag promotion-517. 30g 170/390, 100g 517/1300.
  • Trajectory (100g totals): 458 -> 460 -> 471 -> 512 -> 515 -> 517.
preset wins/100 vs 512
micro-economy 15 0
tight-cash 15 0
hot-caps 19 -1
mega-loans 21 0
rate-war 27 0
default-storm 29 +2
deposit-drought 35 0
duel 36 +1
cash-glut 41 +3
base 43 0
bond-economy 64 0
thin-margins 73 0
patient-market 99 0

Only moves vs 512: cash-glut 38->41, default-storm 27->29, duel 35->36, hot-caps 20->19. Zero other cells moved. Weakest five: micro-economy 15, tight-cash 15, hot-caps 19, mega-loans 21, rate-war 27.

Next: manager re-points the wave-7 baseline to 058c9ed4 (DB still lists 7bb8ce8/515) and dispatches the wave-7 roster against the new canonical. eng28 flags exp87's unbenched tight-cash dual price ladder as a candidate thesis for the next tight-cash lane.

Engineer round closed + promotion landed (22:08Z). Canonical 058c9ed4 (tag promotion-517): 30g 170/390, 100g 517/1300. Master 4131819. | engineer | candidates merged | 30g | 100g | delta vs 512 | decision | |---|---|---|---|---|---| | eng26 | exp94 + exp95 pair | 170/390 | 517/1300 | +5 | PROMOTED (058c9ed4, promotion-517) | | eng27 | exp94 v3 literal | 169/390 | 515/1300 | +3 | kept, superseded by 517 | | eng28 | exp95 alone | 168/390 | 514/1300 | +2 | kept, not promoted | Two promotions landed 62 seconds apart: eng27's 515 went first (7bb8ce8, 22:05:53Z), then eng26's 100g came in at +5 and per spec the best result gets promoted, so 517 superseded it (22:06:55Z). Final canonical is the pair merge: exp94 v3 glut channel negation + exp95 v3 storm-ratio acceptance premium. All three candB mega-floor variants refuted (eng28 1-tick, eng27 2-tick, eng26 funding step): the floor binds but mega equity bleeds 13.8k->8.8k and the extra base never converts to shelf wins. The mega deposit franchise is not purchasable at 1-2 ticks; empire's mega lead is structural, not deposit-priced. Promotion block: - New canonical: 058c9ed4, tag promotion-517. 30g 170/390, 100g 517/1300. - Trajectory (100g totals): 458 -> 460 -> 471 -> 512 -> 515 -> 517. | preset | wins/100 | vs 512 | |---|---|---| | micro-economy | 15 | 0 | | tight-cash | 15 | 0 | | hot-caps | 19 | -1 | | mega-loans | 21 | 0 | | rate-war | 27 | 0 | | default-storm | 29 | +2 | | deposit-drought | 35 | 0 | | duel | 36 | +1 | | cash-glut | 41 | +3 | | base | 43 | 0 | | bond-economy | 64 | 0 | | thin-margins | 73 | 0 | | patient-market | 99 | 0 | Only moves vs 512: cash-glut 38->41, default-storm 27->29, duel 35->36, hot-caps 20->19. Zero other cells moved. Weakest five: micro-economy 15, tight-cash 15, hot-caps 19, mega-loans 21, rate-war 27. Next: manager re-points the wave-7 baseline to 058c9ed4 (DB still lists 7bb8ce8/515) and dispatches the wave-7 roster against the new canonical. eng28 flags exp87's unbenched tight-cash dual price ladder as a candidate thesis for the next tight-cash lane.
Winnie commented 2026-09-30 22:15:44 +00:00
Author
Owner
Copy link

Closed as ordered (22:10Z). For the record: the engineer round actually finished minutes before the abort — eng26's pair merge won at 517/1300 (+5) and is promoted, canonical 058c9ed4, both promotion tags pushed. The next-steps line in my last comment is moot: wave 7 never dispatched and won't be. Pipeline stopped.

Closed as ordered (22:10Z). For the record: the engineer round actually finished minutes before the abort — eng26's pair merge won at 517/1300 (+5) and is promoted, canonical 058c9ed4, both promotion tags pushed. The next-steps line in my last comment is moot: wave 7 never dispatched and won't be. Pipeline stopped.
Sign in to join this conversation.
No Branch/Tag specified
Branches Tags
master
158-market
157-control
156-records
155-stdlib
151-gauntlet
149-eligibility
147-report
145-vickrey
143-pauses
141-defaults
139-summary
137-version
127-minors
125-replay
126-negbuiltin
123-writeoff
124-revoke
121-panic
122-logf
120-finite
119-shoreup
116-acq-basis
88-failure-calc
113-player-display
merge-super
v050
canonical
105-bond-volume
106-bid-all
100-shelf-reset
100-nondet
exp13
98-tfix
94-amort
80-super-blend
89-t
91-nan
80-super-b
80-super-d
80-super-x4
80-super-x3
80-super-x1
87-let-if
80-super-x2
80-super-r1
80-super-ai2
80-super-ai2b
80-super-c
84-simult
80-super-ai
78-gaap
79-api-clean
75-reserve
73-acq-books
71-liquidity
69-bond-units
64-summary
67-loan-summary
63-churn-display
52-empire
53-mill
60-bench-roster
54-vault
55-adapt
39-main-overhaul
47-report
45-acct
46-iter-log
37-hybrids
40-version-harness
38/punishing-presets
41/fix-ci-checkout
v0.6.1
v0.6.0
v0.5.0
promotion-517
promotion-515
promotion-512
promotion-471
promotion-460
promotion-458
v0.4.0
v0.3.0
v0.2.0
Labels
Clear labels
No items
No labels
Milestone
Clear milestone
No items
No milestone
Projects
Clear projects
No items
No project
Assignees
Clear assignees
No assignees
2 participants
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set

Reference
Winnie/bankwars#109
No description provided.