ensemble-of-five - Predictions for:

Forecaster A (historical base rates): 14%. Russia has repeatedly operated near NATO territory, but deliberate kinetic attacks on NATO members remain rare because the escalation costs are exceptionally high. The short remaining window also limits the opportunity for a major operation.

Forecaster B (current evidence): 24%. The armed explosive-carrying drone entering Lithuania from the Belarusian direction, Poland's warning about possible disguised Russian strikes, and repeated Russian activity near NATO borders are meaningful upward indicators. If the Lithuanian incident is later attributed to Russia or Belarus and judged intentional, the event could resolve yes without a new attack.

Forecaster C (incentives and game theory): 18%. Russia may have incentives to test NATO cohesion through a deniable, limited strike, especially against a logistics or border target. However, even a small attack risks triggering a much larger NATO response and could undermine Russia's strategic position, making deliberate escalation unattractive.

Forecaster D (quantitative/statistical reasoning): 17%. The event requires several conditions simultaneously: kinetic action, NATO territory or assets, Russian or proxy responsibility, deliberate intent, and an Article 5-testing purpose. Nearby incidents raise the chance of a qualifying event, but most plausible incidents will instead be accidental spillover, ambiguous airspace violations, or hybrid operations. With roughly two months remaining, the cumulative probability is elevated but still below one in five.

Forecaster E (devil's advocate): 15%. The research may overinterpret public warnings and ambiguous incidents. Poland and other frontline states have strong incentives to publicize threats, while Russia has generally avoided an unmistakable attack on NATO despite many opportunities. The Lithuanian drone may ultimately be attributed to a non-Russian actor, an accident, or an operation without demonstrable Article 5-testing intent.

Aggregating these estimates gives a value around 18%. The main reason not to use the document's 22% central estimate is the strict resolution standard: a future incident must be shown to be deliberate, Russian-directed or Belarusian-proxy, and intended at least partly to test NATO. The Lithuanian incident is a significant near miss and could itself later qualify, but its attribution and intent remain unresolved. A small deniable drone or missile strike is considerably more likely than a tank incursion or sustained attack, while a major Russian assault remains very unlikely.

My aggregate estimate is 16%, somewhat below the research document's 22% central estimate. The risk is clearly elevated: an explosive-carrying drone entered Lithuanian airspace from the Belarusian direction, Russian attacks have occurred near NATO borders, and Poland has publicly warned of possible deliberately ambiguous Russian strikes. These developments make a limited drone or missile incident the most plausible pathway, particularly against Poland or the Baltic states.

However, the event definition is demanding. It requires not merely a kinetic incident or airspace violation, but evidence that Russia or a Russian-directed Belarusian proxy acted deliberately and that the action was intended in part to test NATO's Article 5 response. The Lithuanian incident remains unattributed and could ultimately be accidental, Ukrainian, criminal, or otherwise unrelated to a Russian decision. The July Polish missile incident and other spillovers likewise do not qualify absent evidence of deliberate targeting.

Forecaster A, historical base rates, would place the probability around 10-12%: Russia has strong incentives to avoid an unmistakable attack on NATO, while prior near misses have generally stopped short of deliberate kinetic attacks. Forecaster B, emphasizing current evidence, would estimate 20-22% because of the Lithuanian drone, Polish intelligence warning, and increasing border proximity. Forecaster C, focusing on incentives, would assign about 14-16%: a deniable probe could serve Russian objectives, but the escalation risks remain severe and NATO defenses are improving. Forecaster D, using the short 71-day horizon and conditional event rates, would estimate roughly 15-17%; there may be opportunities for incidents, but the probability of clear attribution and qualifying intent is much lower. Forecaster E, as devil's advocate, would give 10-12%, stressing that public warnings may reflect deterrence messaging and that the evidence may be suffering from recency bias.

Combining these views yields approximately 16%. A major or sustained Russian attack is far less likely, probably around 2-4%; most of the 16% risk is concentrated in a small, deniable drone or missile strike whose deliberate nature is established only after investigation.

Forecaster A (historical base rates): Deliberate Russian kinetic attacks on NATO members have been rare, while near misses, airspace violations, and hybrid operations are considerably more common. The short remaining window through November 30 further limits the chance of a planned attack. Estimate: 10%.

Forecaster B (current evidence): The environment is unusually tense: Russia continues intensive drone and missile operations, reconnaissance activity near NATO borders persists, Belarus provides a possible launch or transit route, and ambiguous spillover incidents have occurred. However, there is no clear offensive buildup, no recent qualifying incident, and diplomacy remains active. Estimate: 13%.

Forecaster C (incentives and game theory): Moscow has incentives to probe NATO cohesion, particularly with a small deniable drone or missile incident, but the costs of a clearly attributable attack are very high. Russia can achieve much of its coercive objective through hybrid activity and airspace pressure without crossing the kinetic threshold. Estimate: 12%.

Forecaster D (quantitative reasoning): Starting from a low annual base rate for deliberate Russia-NATO attacks, the elevated operational tempo and roughly 75-day remaining period justify a meaningful increase, but the strict requirement for deliberate intent and attribution substantially reduces the probability relative to the likelihood of accidental spillover. Estimate: 11%.

Forecaster E (devil’s advocate): The forecast may understate escalation risk because repeated ambiguous incidents can normalize Russian probing, and a small attack could be designed specifically to exploit attribution uncertainty. A failed diplomatic process or renewed major Russian air campaign could sharply increase risk. Nevertheless, absent such a trigger, 14% is probably somewhat high. Estimate: 9%.

Aggregating these views produces a central estimate of approximately 11–12%. I use 12% as the final probability. The most plausible qualifying scenario is a limited, deliberate drone or missile strike against eastern NATO territory or a military asset, potentially involving Belarus and followed by Russian denials. A major invasion remains far less likely, while accidental spillover is more probable than the event as strictly defined but does not qualify on its own.

Forecaster A (historical base rates): A deliberate Russian kinetic attack on NATO territory has a very low base rate because the escalation risks are extreme and Russia has generally remained below the Article 5 threshold despite repeated opportunities. The remaining time is short, and there is no evidence of preparations for a major attack. However, limited deniable drone or missile incidents are more plausible than a major invasion, supporting a low-teens estimate rather than a near-zero probability.

Forecaster B (current evidence): The evidence shows elevated operational danger but no confirmed qualifying incident. Russian missile and drone attacks on Ukraine, repeated reconnaissance near NATO borders, ambiguous drone spillover, Belarusian military integration, and Russian-attributed hybrid activity all increase risk. Offsetting factors include active diplomacy, NATO defensive preparations, the absence of a major Russian buildup, and the fact that recent incidents remain ambiguous or accidental. Current evidence supports roughly 12–15%.

Forecaster C (incentives and game theory): Moscow may have incentives to test NATO cohesion with a small, deniable strike, especially against a military or logistics target and possibly via Belarus. Yet the likely consequences—stronger NATO unity, expanded military support for Ukraine, and possible direct conflict with the Alliance—make an intentional attack unattractive unless Russia believes attribution or allied response would remain uncertain. Russia can obtain many coercive benefits through hybrid operations and airspace probing without crossing the kinetic threshold. This keeps the probability in the low teens.

Forecaster D (quantitative reasoning): With approximately 2.5 months remaining, the relevant hazard is concentrated in rare limited incidents rather than a sustained invasion. A rough annualized risk for a deliberate qualifying incident in this environment, discounted for the short window and the strict attribution/intent criteria, yields a probability around 10–15%. The chance of some accidental spillover is higher, but accidental or unresolved incidents do not count. A point estimate of about 13% is reasonable.

Forecaster E (devil’s advocate): The 13–14% estimate may understate tail risk because Russia has already accepted increasing risks around NATO airspace, and ambiguity could make a small strike politically attractive. A failed or deteriorating diplomatic process, a renewed Russian air campaign, or a Belarus-linked operation could produce rapid escalation. Nevertheless, this argument raises the estimate only modestly because such an attack would still be a major strategic gamble and the forecast excludes many likely spillover and hybrid events.

Aggregating the five views, the probability is approximately 13%. This is slightly below the research document’s 14% central estimate because the resolution criteria require not merely a kinetic incident, but credible evidence of deliberate Russian or Russian-directed action and an intent to test Article 5. The most likely qualifying pathway is a small, ambiguous drone or missile attack before December 1, 2026; a tank incursion or major sustained attack remains highly unlikely.

Forecaster A (historical base rates): Deliberate Russian kinetic attacks on NATO territory are rare because the escalation costs are exceptionally high, and Russia has generally preferred coercion, probing, and hybrid activity below the Article 5 threshold. The recent Poland missile incident and other drone incursions did not meet the required intent standard. This supports a low probability, roughly 8–12%, though the ongoing war creates elevated exposure.

Forecaster B (current evidence): The risk is higher than normal because Russian missile and drone operations remain intense, Baltic airspace incidents and reconnaissance flights are recurring, Belarus provides a plausible launch vector, and Russian-attributed hybrid operations show willingness to operate inside NATO. Against this, there is no clear offensive buildup, no recent qualifying incident, and active U.S.-Russia diplomacy. This points to roughly 12–15%.

Forecaster C (incentives and game theory): Moscow could gain information or political leverage from a small, ambiguous strike, especially against a military or logistics asset. However, deliberately testing Article 5 risks NATO cohesion, stronger military involvement, and loss of diplomatic leverage. Russia can obtain many benefits through sabotage, intimidation, and airspace probing without crossing the kinetic threshold. Incentives therefore favor restraint, with a limited deniable strike as the main residual risk.

Forecaster D (quantitative reasoning): About 82 days remain, so even a low daily hazard accumulates to a meaningful probability. The most plausible event is a small drone or missile incident, but the forecast requires not merely impact or airspace violation—it requires credible evidence of deliberate Russian or Russian-directed intent. Applying that substantial resolution filter to the elevated regional activity yields approximately 12–14%.

Forecaster E (devil’s advocate): The 14% central estimate may understate tail risk. A failed diplomatic effort, renewed Russian air campaign, Belarusian involvement, or a casualty-producing incident could rapidly create pressure for a deliberate probe. Ambiguity could also let Moscow test NATO while attempting to avoid uncontrollable escalation. Nonetheless, the absence of current mobilization or explicit crisis indicators makes a near-term attack more likely to remain below the threshold than to cross it.

Aggregating these estimates, I assign a 13% probability. This is close to the recent forecast history and slightly below the research document’s 14% central estimate because the event definition requires deliberate intent and excludes accidental spillover, debris, and unresolved attribution. The most likely qualifying scenario remains a small, deniable drone or missile attack, potentially involving Belarus, rather than a tank incursion or sustained invasion.

Forecaster A (historical base rates): 10%. Russia has repeatedly used coercion, airspace probing, and hybrid operations against NATO countries while avoiding an overt kinetic attack. The absence of a deliberate Article 5 test over a long period is meaningful, though the war in Ukraine creates more opportunities for escalation.

Forecaster B (current evidence): 14%. The most relevant evidence is the continuing high-volume Russian missile and drone campaign, repeated activity near Baltic airspace, Russian reconnaissance flights, Belarusian military integration, and recent Russian-attributed hybrid operations in NATO states. However, the July Polish missile incident remains unattributed as deliberate, and no new qualifying incident or major offensive buildup has appeared.

Forecaster C (incentives and game theory): 11%. A deliberate strike would risk NATO military retaliation, greater Western unity, and loss of diplomatic leverage. Russia can achieve many of its objectives through sabotage, intimidation, cyber operations, and ambiguous airspace violations without crossing the kinetic threshold. A small deniable attack is possible, but the strategic incentives generally favor staying below that threshold.

Forecaster D (quantitative/statistical reasoning): 13%. With roughly 84 days remaining, the risk is concentrated in a limited number of plausible pathways: a deliberate drone or missile strike, a Belarus-linked incident, or escalation during a renewed Russian air campaign. The daily probability is low, but the elevated operational tempo and growing number of near misses make the cumulative probability material. The stringent requirement to establish both deliberate Russian responsibility and an intent to test NATO reduces the chance relative to the probability of any spillover incident.

Forecaster E (devil’s advocate): 17%. Russia may view a small, ambiguous attack as a way to test NATO’s political cohesion, especially if it believes Washington or European governments would respond cautiously. The density of military activity, ambiguous drone incidents, and Russian willingness to conduct operations inside NATO countries could provide cover for such an action. The main counterargument is that a kinetic strike could produce consequences Moscow cannot control.

Aggregating these estimates gives a central probability of approximately 13%. The probability of any drone or missile incident in NATO territory is higher, but the event requires credible evidence of a deliberate attack by Russia or a Russian-directed Belarusian proxy, rather than accidental spillover, debris, or an unattributed violation. A major ground invasion remains far less likely, probably only around 1–3%. The strongest upward triggers would be a confirmed deliberate strike, weapons launched from Belarus, a collapse of current diplomacy followed by intensified Russian operations, or unusual Russian-Belarusian force movements near the Suwalki corridor. The strongest downward triggers would be a durable ceasefire, explicit U.S. response commitments, or sustained Russian de-escalation.

Forecaster A (historical base rates): Deliberate Russian attacks on NATO territory have been rare because the escalation costs are exceptionally high, while Russia has generally preferred deniable hybrid activity and operations below the Article 5 threshold. A limited incident is more plausible than a major invasion, but the stringent requirement for deliberate intent makes the base rate relatively low: 10%.

Forecaster B (current evidence): The risk is elevated by sustained Russian drone and missile operations against Ukraine, repeated activity near Baltic airspace, Russian reconnaissance flights, Belarusian military integration, and recent Russian-attributed hybrid operations in NATO countries. However, the July Polish missile incident remains unattributed as deliberate, there is no major offensive buildup, and diplomatic engagement is ongoing. Estimate: 14%.

Forecaster C (incentives and game theory): Russia has some incentive to test NATO cohesion with a small, ambiguous strike, especially if it perceives Western political divisions. But a clearly attributable attack could trigger military retaliation and unify NATO, making the strategic payoff uncertain and the downside large. Moscow therefore has stronger incentives to remain below the kinetic threshold than to openly test Article 5. Estimate: 11%.

Forecaster D (quantitative/statistical reasoning): Roughly 84 days remain, giving meaningful opportunity for a drone or missile incident but limited time for a major planned operation. The most likely qualifying pathway is a small strike or deliberately directed drone attack. Increasing operational density raises the incident rate, but the requirement to establish deliberate Russian or proxy intent substantially lowers the probability relative to the chance of any spillover event. Estimate: 13%.

Forecaster E (devil's advocate): The apparent calm and diplomacy could be misleading. Russia may exploit ambiguity, Belarusian territory, or a renewed air campaign to conduct a deniable probe, and repeated near misses could normalize escalation. Hybrid activity demonstrates willingness to accept risk inside NATO. These considerations justify a somewhat higher estimate, though not one approaching even odds because an attack would still carry severe and unpredictable consequences. Estimate: 17%.

Aggregating the five views produces a central estimate of approximately 13%. The probability is higher than the chance of a major invasion, but lower than the probability of an accidental spillover or hybrid operation, both of which are excluded. The principal downward factors are deterrence, NATO readiness, Russia's ongoing commitment in Ukraine, and the lack of evidence of offensive preparations. The principal upward factors are the short but nontrivial remaining time window, frequent drone and missile activity near NATO borders, Belarus's role, reconnaissance and counter-drone operations, and the possibility of a small deniable strike intended to probe allied resolve.

Forecaster A (historical base rates): Deliberate Russian kinetic attacks on NATO territory are rare because the escalation costs are exceptionally high, while Russia has repeatedly preferred sabotage, cyberattacks, airspace violations, and other below-threshold actions. This supports a low single-digit baseline, modestly increased by the ongoing war and repeated near misses.

Forecaster B (current evidence): The recent missile landing in Poland, repeated drone incursions, intensified Russian strikes near NATO borders, and Russian-linked activity in Germany indicate a more dangerous environment than when the prior estimate was 8%. However, none clearly constitutes a deliberate qualifying attack, and NATO has strengthened eastern-flank defenses. The evidence supports an upward revision, but not a dramatic one.

Forecaster C (incentives and game theory): Russia has incentives to probe NATO cohesion, especially through a small, ambiguous drone or missile incident. But a clearly attributable attack could trigger military retaliation and undermine Russia’s broader strategic position. Moscow can often obtain similar coercive benefits through hybrid operations, making a deliberate kinetic test less attractive. Belarus could provide ambiguity, but would not eliminate escalation risks.

Forecaster D (quantitative/statistical reasoning): With roughly three months remaining, the opportunity window is limited but nontrivial. The relevant hazard is elevated by the density of Russian attacks around Ukraine and the number of airspace incidents, yet the event definition excludes accidents, debris, and ambiguous violations. A reasonable estimate is in the low-to-mid teens, with major sustained conflict far below that.

Forecaster E (devil’s advocate): The probability could be higher if Russia interprets Western political signals as weak, if a Belarus-linked operation is used to create ambiguity, or if an accidental spillover is deliberately reframed as a probe. Conversely, the stringent requirement for clear deliberate intent makes many plausible incidents non-qualifying. The strongest counterargument to a higher estimate is that Russia has already found effective ways to pressure NATO without crossing this threshold.

Aggregating these views, I estimate a 14% probability before December 1, 2026. This is above the previous 8% forecast because of the deteriorating environment and repeated kinetic near misses, but below the research document’s upper range because the remaining time is short and the resolution criteria require deliberate, attributable kinetic action rather than accidental spillover or hybrid activity.

My aggregate estimate is 8%. Forecaster A (historical base rates) assigns roughly 4–6%: Russia has repeatedly used hybrid pressure and airspace violations against NATO members but has avoided an acknowledged kinetic attack, despite many opportunities. Forecaster B (current evidence) assigns 9–12%, reflecting increased drone incursions, missile incidents, concern about NATO cohesion, and reported assessments that Moscow could consider a limited test. Forecaster C (incentives and game theory) assigns 6–9%: Russia has incentives to probe Western resolve, especially if it perceives U.S. disengagement or allied division, but a deliberate strike risks a war with the entire alliance and therefore has strongly negative expected value. Forecaster D (quantitative reasoning) assigns about 8–10%, broadly consistent with the cited estimates of around 9% for limited Baltic action and approximately 17% for a broader Russia–NATO clash; the narrower event definition warrants a lower figure than the broader clash estimate. Forecaster E (devil’s advocate) assigns 3–5%, emphasizing that the cited intelligence and market signals may conflate gray-zone incidents, accidents, and a general military clash with a deliberate attack intended to test Article 5. The final estimate is above the long-run baseline because of the Ukraine war, Russian coercive behavior, Belarusian integration, and possible perceived Western weakness, but remains below 10% because NATO deterrence is strong, Russia is heavily constrained by the war in Ukraine, and even a limited overt attack could trigger uncontrollable escalation. Hybrid incidents and accidental spillover are substantially more likely than the narrowly defined event and are excluded from this estimate.