What should you do when the probabilities are not knowable?

In everyday investing, risk and uncertainty often overlap. FINRA defines investment risk broadly in terms of uncertainty that can harm financial welfare. In this guide, a narrower working distinction helps organize a decision: measurable risk refers to outcomes for which a stated probability model is available; uncertainty refers to important outcomes or probabilities that you cannot estimate credibly. FINRA: Risk.

This distinction does not mean that a model makes an investment safe. The model itself depends on assumptions, data, and a view of how the future relates to the past. A precise number can still be unreliable.

The practical question is how to make a proportionate decision when you can calculate consequences more confidently than probabilities. That situation calls for transparent scenarios and constraints, rather than inventing a percentage that creates an illusion of knowledge.

Imagine a 25,000-dollar portfolio considering a 2,000-dollar exposure to a fictional business awaiting an uncertain commercial decision. The reader cannot credibly estimate the probability of approval or the market's response. They can still calculate what several assumed price outcomes would mean.

A 30% rise in the holding adds 600 dollars, or 2.4% of the starting portfolio. A 40% decline loses 800 dollars, or 3.2%. A complete loss of this ordinary unleveraged holding costs 2,000 dollars, or 8%. The other portfolio assets are assumed unchanged solely to isolate the arithmetic.

None of these numbers says the outcomes are equally likely. Averaging them as though they each had a one-third probability would introduce an unsupported assumption. The useful output is a range of consequences under stated scenarios, together with an honest statement that their probabilities and completeness remain unknown.

Create a short evidence ledger with three categories: observed facts, modeling assumptions, and unresolved questions. A published contractual deadline might be a fact. A belief that the decision will arrive on schedule is an assumption. The eventual commercial outcome remains unresolved.

For every numerical input, ask where it came from and what would invalidate it. A historical frequency from a small or poorly matched sample should not be promoted into a precise probability without explaining why the cases are comparable. If the relevant process has changed, the old frequency may answer a different question.

Use the ledger to identify research that could actually alter the decision. Reading a primary agreement might clarify a loss channel. Reading another confident opinion that supplies no new evidence might not. The goal is not to collect more words; it is to reduce the uncertainties that matter to the consequence.

Ask which adverse consequences would be unacceptable regardless of their uncertain probability. If an 800-dollar loss would prevent a required payment, the example's exposure conflicts with that constraint under a plausible scenario you have chosen to examine. You do not need to pretend the probability is 17% to recognize the conflict.

Possible responses to uncertainty include reducing the amount exposed, preserving liquid resources for known obligations, postponing a decision until relevant evidence arrives, or declining an exposure whose downside is not understood. These are decision-process options, not prescriptions for a particular investment.

Waiting also has consequences: information may arrive after prices change, and an opportunity may disappear. Write that tradeoff honestly. Flexibility has value because it preserves choices, but it is not free and does not guarantee a better eventual result. Compare the consequences of acting and waiting on the same basis.

• State the decision and the financial purpose it is meant to serve.
• Separate observed facts from assumptions and unresolved questions.
• Calculate the money consequences of several distinct adverse and favorable scenarios.
• Do not assign equal probabilities merely because the scenario list has a convenient length.
• Identify the constraints that must remain workable even if the preferred scenario fails.
• Name the specific new evidence that would justify revisiting the decision.

For the fictional business, the final note might say that the 2,000-dollar exposure could lose enough to disrupt a defined obligation, while the available evidence does not support a credible probability estimate. That is a decision-relevant statement even though it contains no prediction of the announcement.

A hypothetical opportunity offers a 600-dollar gain in one modeled outcome and an 800-dollar loss in the other. Assume, only within this deliberately restricted two-outcome exercise, that there are no other outcomes, costs, or timing differences. If the favorable probability is p, the modeled expected money result is 600 times p minus 800 times one minus p.

Setting that expression to zero gives 1,400 times p equals 800, so the break-even probability is approximately 57.14%. This is a sensitivity threshold implied by the invented payoffs. It is not evidence that the favorable probability actually exceeds 57.14%, and it is not a recommendation to proceed whenever someone supplies a larger percentage.

The threshold can make an unsupported claim easier to challenge. A statement that the chance is probably better than half does not establish positive expected money under these assumptions. Even an asserted 60% probability requires evidence about comparability, omitted outcomes, and the payoff amounts. The exact algebra highlights the evidence needed rather than filling the gap itself.

Now widen the scenario set to include a possible 2,000-dollar loss. The original two-outcome formula no longer describes the whole proposed model. You cannot simply retain the 57.14% threshold without specifying how the additional outcome changes the probabilities and payoffs. This demonstrates how a precise answer can depend on a narrow framing that excluded the consequence most important to the decision.

Finally, expected money is not the same as acceptable consequences. A modeled positive average would not ensure the owner can absorb the adverse outcome or meet a required payment afterward. Keep the probability threshold, uncertainty about the model, and household constraints in separate fields. The exercise is useful when it clarifies those distinctions, not when it produces a confident number unsupported by the available information.

Suppose a hypothetical business faces an unresolved contract decision. Three original stories are considered: prompt approval followed by a 25% price gain, a long delay accompanied by a 20% decline, and contract failure accompanied by a 70% decline. These are selected consequences for analysis, not exhaustive descriptions of what can happen or estimated probabilities of the contract outcomes.

For a 2,000-dollar ordinary unleveraged holding, the modeled money effects are positive 500, negative 400, and negative 1,400 dollars. For a 500-dollar holding, they are positive 125, negative 100, and negative 350 dollars. With no holding, the direct contribution from this opportunity is zero in each scenario, ignoring what happens to any alternative use of the money.

Assume the fictional decision maker has defined a requirement that at least 1,000 dollars of a 2,000-dollar earmarked pool remain available under each selected scenario. The larger holding leaves only 600 dollars after the assumed 70% decline and fails that particular test. The smaller holding, with 1,500 dollars held separately and unchanged, leaves 1,650 dollars and passes the selected test.

Passing does not establish safety under every event. The assumed separate cash must actually remain available, and the selected scenarios may omit other demands on it. The conclusion is limited: under these specified outcomes and resources, one scale conflicts with the stated constraint while the other does not. No probability estimate is required to identify that difference.

This method is useful when the consequence is easier to estimate than its likelihood. It compares choices against the same constraints instead of using an unsupported average to rank them. Keep the source of each constraint visible too. An arbitrary teaching threshold should never be mistaken for a general rule about how much uncertainty a real household ought to accept.

An evidence search is useful when its possible answers connect to different decisions. In a hypothetical contract situation, suppose the unresolved question is whether a missed milestone allows termination without payment. Reading the actual agreement could clarify a material loss mechanism. Reading another opinion about management's confidence might leave that contractual uncertainty exactly where it started.

Write two branches before researching. If the agreement contains the assumed termination right, what would change in the exposure analysis? If it does not, what uncertainty remains? This prevents a research session from becoming a collection of interesting facts that cannot affect the decision. It also makes clear when a question requires the primary contract rather than a summary of it.

Not all uncertainty is removable before acting. A signed agreement may establish obligations while leaving execution, commercial demand, and the future market price unresolved. Distinguish evidence about the rules from evidence about the eventual outcome. Confirming the former can improve the scenario model without supplying a defensible probability for the latter or guaranteeing that the investment responds favorably.

This is not a reason to avoid background learning. It is a way to allocate attention when a decision has a deadline. Stop describing uncertainty as reduced merely because more documents have been read. State what changed: a contract term confirmed, an assumption rejected, a cash requirement clarified, or a probability still unknown. Those concrete updates make the eventual decision easier to review without overstating what the research accomplished.

Suppose a hypothetical share trades at 20 dollars before an unresolved announcement. Acting now with 1,000 dollars buys 50 shares before costs. Waiting preserves the 1,000 dollars temporarily, but assume the price could be 25 dollars after favorable news or 12 dollars after unfavorable news. Those invented prices describe alternative information paths, not predicted reactions.

Under the favorable path, the early purchase becomes worth 1,250 dollars. A later 1,000-dollar purchase at 25 dollars would buy 40 shares. The waiting decision has missed the earlier price increase if it eventually buys. Under the unfavorable path, the early holding becomes worth 600 dollars, while waiting preserves the option to decline the exposure or reassess it with new information.

Do not compare the early buyer's realized gain in one path with the waiting buyer's avoided loss in another and conclude that either choice wins in all circumstances. Compare both choices within each path. Waiting does not deliver a favorable purchase automatically; the investor must still evaluate the new price and the remaining uncertainty when the information arrives.

A smaller initial commitment creates another decision structure but does not remove the tradeoff. A hypothetical 200-dollar initial purchase buys ten shares, leaving 800 dollars uncommitted. It participates in the selected price movement at smaller scale while retaining resources. Whether that structure serves a real objective depends on costs, instrument terms, and the reason to acquire more later, not on the appealing phrase keep options open.

The reusable worksheet records the amount committed now, what is preserved, what information is awaited, and the decision to revisit after it arrives. Include a condition for doing nothing further. Otherwise, a staged decision can turn into an implicit commitment to complete the purchase regardless of what the evidence says, undermining the flexibility that was supposed to justify waiting.

Imagine a hypothetical analyst writes that a commercial approval is uncertain and cannot be assigned a credible probability. The approval then occurs and the share rises. That result does not retroactively establish that approval was knowable or that a confident prediction would have been justified. The outcome resolves one event; it does not validate every method that happened to point in its direction.

The reverse also matters. A carefully limited exposure can lose money under a scenario that was explicitly recognized. The loss may reveal new information, but its occurrence alone does not prove the original decision ignored uncertainty. Review whether the evidence was represented honestly, the consequences were calculated correctly, and the commitment fit the stated constraints at the time.

A useful decision record includes the date, information available, unresolved questions, alternative choices, and the reason for the selected scale or decision to wait. After the event, add what actually happened and which assumptions changed. Preserve the original note rather than rewriting it to make the eventual result appear obvious. That gives the review something concrete to compare against memory.

For a hypothetical contract approval, the review might find that the formal decision arrived as expected but the price response differed because another development occurred simultaneously. That observation should improve the scenario framework by separating event outcome from investment payoff. It does not justify inventing a precise probability for the next contract or treating one observed reaction as a dependable pattern.

Finish by naming the next useful change in the process. Perhaps the original worksheet omitted a possible delay, mixed facts with assumptions, or failed to connect the downside to a cash deadline. A specific correction is more actionable than declaring the decision good because it made money or bad because it lost. Under uncertainty, review the quality of the reasoning and the consequences together, while keeping their meanings distinct.

A scenario list is not a complete catalog of future events. Several adverse developments can combine, and a loss mechanism may have been omitted. Do not label the worst scenario on your worksheet the maximum possible loss unless the instrument's actual structure supports that limit.

Another trap is treating uncertainty as a reason to ignore every measurement. Exact probabilities may be unavailable while position values, contractual obligations, and cash deadlines are still knowable. Use that concrete information. Investor.gov's discussion of allocation ties decisions to time horizon and risk tolerance, which is a useful reminder to start with the investor's circumstances. Investor.gov: Asset Allocation and Diversification.

At the next review, compare the evidence ledger with what actually changed. Update assumptions when warranted, preserve unresolved questions, and avoid rewriting the original reasoning to make an uncertain outcome look inevitable.