Using Dialectical Behavior Therapy to enhance Superforecasting

What each one is actually for

Dialectical Behavior Therapy

DBT, short for Dialectical Behavior Therapy, was developed by Marsha Linehan originally for people with intense, fast-moving emotional swings, and its most exported piece is a set of concrete skills for surviving a crisis moment without making it worse. One of those skills, check the facts, asks you to pause in the middle of a strong reaction and separate the feeling from the situation: is my fear right now proportionate to the actual likelihood and actual severity of what I'm afraid of, or is my body responding to something closer to a worst-case story than to the real odds. Another, sometimes taught as the STOP skill, an acronym for Stop, Take a step back, Observe, Proceed mindfully, is a literal pause sequence you run before you act on an urge, buying enough time for check the facts to actually happen instead of the reaction just running its course. Together they're a practical answer to the question 'what do I actually do in the sixty seconds a crisis is happening,' not a general philosophy about emotions.

Superforecasting

Calibration training comes out of research, most visibly Philip Tetlock's work on 'superforecasters,' people who consistently predict geopolitical and other real-world events more accurately than professional analysts. The core discipline is refusing to answer an uncertain question in vague words, 'probably,' 'unlikely,' 'could go either way,' and instead forcing yourself to state an actual number: I think there's a 70 percent chance this project ships on time. That number does something a vague hedge can't: it can be scored later against what actually happened, and if you do this repeatedly, tracking many predictions over months, you start to see your own specific biases, whether you're systematically overconfident, or you hedge everything toward 50 percent to avoid being wrong, and you can correct for it. The practice is picture-able: you keep a running list of predictions with numbers attached, and you check back on them.

What you get out of pairing them

Check the facts, done honestly, is already asking you to estimate an odds question, is this actually likely to happen, but most people answer it with a feeling rather than a number, and a feeling of 'this is definitely going to go badly' and a feeling of 'this is definitely going to be fine' both masquerade as clarity while carrying no information about your actual track record at judging situations like this one. What calibration training adds is a way of finding out, later, whether your gut is any good at this specific kind of call, which turns 'I have a bad feeling about this' from a conversation-ending certainty into a testable claim you can get better at. For someone doing calibration work on its own, the payoff runs the other way: forecasters who take a hard public miss badly, spiraling into either defensive rationalizing or panic that they've lost their edge, tend to make worse calls right afterward, exactly the pattern DBT's crisis skills exist to interrupt before it compounds.

What Dialectical Behavior Therapy rests on

• Acceptance reduces struggle randomised trials
  Allowing an internal experience to be present without resisting or acting on it reduces the psychological struggle that resistance would otherwise intensify.
• Affect labeling reduces arousal randomised trials
  Putting an emotion into words reduces the physiological and psychological intensity it would otherwise carry while unexamined.
• Cognitive reappraisal changes emotional response randomised trials
  Checking an emotional reaction against the actual facts of a situation can down-regulate a response calibrated to a threat that isn't really present.
• Pausing before reacting prevents escalation clinical practice
  Inserting a brief pause between an urge and an action prevents an impulsive reaction from escalating a crisis.

What Superforecasting rests on

• Decomposition improves estimate accuracy observational studies
  Breaking a complex uncertain question into smaller more tractable sub-questions produces more accurate estimates than judging the whole question at once.
• Explicit probability estimates improve calibration randomised trials
  Forcing a numeric probability rather than a vague verbal hedge makes a prediction falsifiable and trainable.
• Outcome feedback corrects judgment bias randomised trials
  Regularly scoring predictions against actual outcomes and reviewing the misses corrects systematic biases in judgment over time.

Where they actually meet

Both practices distrust the first read a situation gets. DBT's check the facts skill distrusts the emotional read, the felt certainty that a threat is as large as it feels, and asks you to hold it against the actual facts of the situation before you act on it. Calibration training distrusts the intellectual read, the confident hunch about how a real-world question will turn out, and asks you to state it as a number precise enough to be wrong, then check it. In each case the practice is refusing to let a feeling of certainty, whether that certainty is emotional or intellectual, stand in for an actual answer, and instead making the person produce something specific enough to be tested, a fact-check in one case, a scored prediction in the other.

The principle underneath both

A strong sense of certainty, whether it shows up as dread or as confidence, is not evidence about the world; it's information about your current internal state, and the only way to find out if it's tracking reality is to make it specific enough to check against what actually happens. Treating the feeling of certainty itself as proof is where both a crisis reaction and a bad prediction usually go wrong.

What that makes the real goal

What both practices are training is the same underlying skill: catching the moment a strong sense of certainty shows up, and instead of acting on the certainty directly, converting it into something specific enough to be checked, a fact-based comparison in the middle of a crisis, a numeric estimate in the middle of an uncertain forecast, so that later you can find out whether the certainty was warranted. That skill, converting a felt sense of certainty into a checkable claim, isn't owned by crisis therapy or by forecasting research specifically; it's a more general habit either domain can be used to build.

Other practices that serve the same goal

Cognitive Behavioral Therapy's thought records aim at a close cousin of this objective from the therapy side: writing down an automatic thought, rating your belief in it as a percentage, then listing the actual evidence for and against it, which forces the same move from felt certainty to a checkable number, and carries a large base of controlled trial evidence for reducing anxious and depressive thinking. On the judgment-and-decision-making side, structured pre-mortems, a technique where a team imagines a plan has already failed and works backward to explain why, force a similarly explicit accounting of what could go wrong before commitment, with observational support for reducing overconfidence in group planning. Neither tradition was built with the other in mind; both independently landed on making an uncertain internal read explicit and checkable as the lever.

A tradition that already does this

The Stoics, whose crisis-tolerance ideas run underneath much of DBT's philosophical lineage, already had a practice for exactly this kind of forecasting discipline. Premeditatio malorum, Latin for 'premeditation of adversity,' meant deliberately imagining specific bad outcomes in advance, not to spiral into dread but to take the emotional sting out of an outcome by having already looked at it clearly and estimated how likely and how survivable it actually was. It's a much older, less quantified cousin of sitting down and forcing yourself to name a real probability instead of a vague fear.

Using Dialectical Behavior Therapy (DBT) to enhance Superforecasting

A hard, public miss, a prediction that was confidently wrong, is the moment calibration training is most likely to produce a bad reaction rather than a useful lesson: either a defensive story about why the miss doesn't really count, or a spike of anxiety that curdles into either overcorrecting toward extreme caution on the next call or abandoning the practice altogether. Borrow DBT's STOP sequence, stop, take a literal step back from the screen or the spreadsheet, observe what you're actually feeling in your body right now, then proceed mindfully, before you touch the post-mortem on a bad miss. Concretely: when you notice the flush of shame or the urge to explain the miss away, name it out loud, 'I'm feeling defensive about being wrong here,' and let a minute pass before you write the review of what happened. The point isn't to skip the review, it's to keep the emotional reaction to being wrong from writing the review for you, which is exactly how forecasters develop blind spots they can't see.

Using Superforecasting to enhance Dialectical Behavior Therapy (DBT)

Check the facts, as DBT teaches it, usually ends at a yes-or-no judgment: is my reaction proportionate, or not. That binary is easier to argue yourself out of in the heat of a crisis than an actual number is. Borrow calibration training's discipline directly: mid-crisis, instead of just asking 'is this proportionate,' force yourself to name an actual percentage, 'I'd put it at maybe 15 percent that this conversation ends the relationship,' even roughly. A vague sense of doom resists being examined; a number invites the next question, which is what would have to be true for that number to be right, and answering that question is usually what breaks the spell of a catastrophic feeling that has nothing behind it. The concrete move is refusing to let 'this feels catastrophic' stand as an answer, and making yourself produce a specific figure instead, however imprecise, because the act of producing it does most of the work.

Waiting on a text back after a fight, at 11pm

You sent a text twenty minutes ago after an argument and there's no reply, and your chest is tight, and the thought arriving on a loop is that this is over, they've had enough, this is the fight that ends it. Before you send a second message, or three, you run STOP: stop typing, take a physical step back from the phone, set it on the counter. Observe: your jaw is clenched, your breathing is shallow, and the thought is arriving as certainty, not as a guess. Now do the number, out loud if you're alone: realistically, given every other fight you two have had and how those actually ended, what's the real percentage this one is different, this one is the end. Most of the time, said honestly, it's closer to five or ten percent than the ninety percent your body is currently insisting on. You're not pretending the fight didn't matter, and you're not texting again yet either. You're waiting, with a number attached to the waiting instead of just the raw feeling running the show.

When this is worth reaching for

The test results aren't back yet, and you've spent the last two days certain — genuinely certain — of the worst outcome, in a way that has nothing to do with what the actual odds are. The feeling and the real probability have drifted apart, and reacting on the feeling alone, before the numbers exist to check it against, is how a bad two days turns into a bad decision on top of it.

Where this does not apply

In an acute crisis where safety is genuinely on the line, self-harm risk, a medical emergency, immediate danger, stopping to produce a calibrated probability estimate is the wrong move entirely; DBT's crisis skills exist to get a person through those minutes safely, full stop, and adding a forecasting exercise on top only delays action a situation like that needs immediately. And for low-stakes daily annoyances, running a full check-the-facts-plus-probability-estimate process on something like a delayed email reply is more machinery than the moment calls for; a two-second gut check is enough for most ordinary friction, and reserving the fuller process for genuinely high-stakes uncertainty keeps it from becoming its own source of overthinking.

What the evidence does and does not support

DBT's crisis skills, including check the facts and the STOP sequence, are supported by a solid clinical evidence base, originally built and tested in populations with severe emotional dysregulation, with the broader claims about pausing before reacting and labeling affect additionally supported by controlled research outside DBT specifically. Calibration training's core claim, that forcing explicit numeric probability estimates and scoring them against outcomes improves judgment over time, has randomized and quasi-experimental support from forecasting-tournament research, most visibly the work behind the term superforecasting. Neither literature has tested what happens when a person applies DBT's crisis pause to a forecasting miss or applies calibration's numeric-estimate habit inside an emotional crisis; the connection drawn here is a reasoned extension built from two well-supported but separately studied bodies of work, not a documented result of combining them.