Configurations

Instead of asking which single pressure matters most, this page asks whether some combination of the five reliably comes before things getting worse — say high energy and high geopolitical together, while economic stays low. Each such recipe is a configuration, and with five pressures there are exactly 32 of them.

All 32 are listed, including combinations no day ever actually matched. A recipe passes only if escalation followed it reliably enough — the conventional bar is 0.80 out of 1.

Why all 32, and how the scores are computed

Corners are enumerated rather than grouped over observed patterns: a corner nobody lived through still has to be scoreable, and still has to be marked as never lived through rather than as having scored zero. Zero would claim we looked and found nothing; unobserved says the combination never arose.

Consistency and coverage are summed over every case, weighted by that case's membership in the corner — not over the subset whose crisped pattern matches exactly. Restricting to crisp matches discards the partial memberships that fuzzy sets exist to carry, and it inflates consistency: this project measured 1.000 against a true 0.948 on the same corner before the summation was fixed.

Recipe set

v3 · k=7 · median@0.007235 v1 · k=7 · median@0.202645

A set is the whole key (anchor_version, outcome_name, outcome_rule, horizon_k), not four independent choices: the median split moves as data accumulates, so each anchor version carries its own threshold and the two are only meaningful together.

No combination of pressures reliably precedes escalation

None of the 32 combinations passes. The best of them was followed by escalation 72% of the time, and the bar is 80%. So there is no pattern of these five pressures that reliably comes before things getting worse 7 days later.

That is a complete answer rather than an empty table. Every combination was checked and scored; the scores are shown in full below, next to the bar they had to clear.

Corners enumerated

the full 2⁵ space

Never happened

combinations no real day ever matched — scored, but not eligible to be reported as findings

Reliable enough (0.8+)

counting reliability alone — before the other two bars are applied

Truth table · 32 corners

must be right 0.8+ of the timemust explain 0.1+ of escalationmust have happened at least once

The vertical mark in every bar is the 0.8 pass line. It is drawn on rows that fall far short of it on purpose: a table where no bar reaches the line should look like a measured result, not like a rendering error.

#	Configuration	Consistency how often escalation actually followed — must reach 0.8	Coverage share of all escalation this recipe accounts for	Cases real days matching this recipe	Result
1	climatedigitaleconomicenergy~geopolitical	0.7167	0.3122	3	below bar
2	climatedigitaleconomicenergygeopolitical	0.6444	0.3053	4	below bar
3	~climatedigitaleconomicenergy~geopolitical	0.6323	0.2081	0	never happened
4	~climatedigital~economicenergy~geopolitical	0.5232	0.1252	0	never happened
5	climate~digitaleconomicenergy~geopolitical	0.4981	0.1364	2	below bar
6	climatedigital~economicenergy~geopolitical	0.4932	0.1052	0	never happened
7	climatedigitaleconomic~energy~geopolitical	0.4905	0.1226	0	never happened
8	~climate~digitaleconomicenergy~geopolitical	0.4879	0.1197	0	never happened
9	~climate~digital~economicenergy~geopolitical	0.4854	0.1193	1	below bar
10	climate~digital~economicenergy~geopolitical	0.4829	0.1003	0	never happened
11	~climatedigitaleconomic~energy~geopolitical	0.4706	0.1226	1	below bar
12	~climatedigitaleconomicenergygeopolitical	0.4675	0.1601	1	below bar
13	~climate~digital~economic~energy~geopolitical	0.4571	0.2588	6	below bar
14	climatedigital~economic~energy~geopolitical	0.4465	0.1087	1	below bar
15	climate~digital~economic~energy~geopolitical	0.4257	0.0993	0	never happened
16	climate~digitaleconomic~energy~geopolitical	0.4149	0.0864	0	never happened
17	~climatedigital~economic~energy~geopolitical	0.3914	0.1243	1	below bar
18	climate~digitaleconomicenergygeopolitical	0.3898	0.1099	0	never happened
19	~climate~digitaleconomic~energy~geopolitical	0.3759	0.0864	0	never happened
20	climatedigitaleconomic~energygeopolitical	0.3243	0.0943	0	never happened
21	~climatedigitaleconomic~energygeopolitical	0.3041	0.0943	0	never happened
22	~climate~digitaleconomicenergygeopolitical	0.2790	0.0733	1	below bar
23	~climatedigital~economicenergygeopolitical	0.2622	0.0667	0	never happened
24	climatedigital~economicenergygeopolitical	0.2563	0.0598	0	never happened
25	~climate~digital~economicenergygeopolitical	0.2287	0.0592	0	never happened
26	climate~digitaleconomic~energygeopolitical	0.2286	0.0581	0	never happened
27	climate~digital~economicenergygeopolitical	0.2241	0.0524	0	never happened
28	climatedigital~economic~energygeopolitical	0.2099	0.0581	0	never happened
29	~climate~digitaleconomic~energygeopolitical	0.1985	0.0581	0	never happened
30	~climatedigital~economic~energygeopolitical	0.1907	0.0649	1	below bar
31	climate~digital~economic~energygeopolitical	0.1856	0.0508	0	never happened
32	~climate~digital~economic~energygeopolitical	0.1837	0.0899	2	below bar

"Never happened" is not the same as "scored zero." A row marked never happened is a combination no day in the sample ever matched — known in the method as a logical remainder. It was scored like every other row, but from partial fits alone, so it says nothing about what would follow if such a day did occur. A row marked below bar is the opposite: real days matched it, and things still did not reliably get worse afterwards. Showing the two the same way is the single error this screen exists to prevent, because a combination that never occurred, reported as a finding, is a claim about a world that never happened.

20 of 32 combinations were never actually observed. They are excluded from the findings on purpose: without that rule, patterns that never occurred would be reported as results — one of them reaches 0.6323 on partial-fit arithmetic alone.