Half the year is gone, which makes this a reasonable moment to look back at whatever you concluded in January. Most people do not, and the ones who do usually conduct the review in a way that guarantees a flattering result.
The difficulty is not honesty. It is that memory reorganises itself around outcomes, and it does so automatically and invisibly.
The problem with reviewing from memory
Once you know how something turned out, it becomes very difficult to reconstruct how uncertain you were beforehand. The outcome feels more predictable in retrospect than it was in prospect, and the reasoning that pointed toward it becomes more prominent in recollection than the reasoning that did not.
The consequence is that an unrecorded forecast cannot be graded. What gets graded is a reconstruction, and the reconstruction is generated by a process that already knows the answer.
This is why the single most valuable thing in any review is a contemporaneous written record. Not a summary written afterwards — the actual thing you wrote at the time, with the date on it.
Separating the decision from the outcome
The second discipline is harder and matters more. A good decision can produce a bad outcome and a bad decision can produce a good one, because the world is probabilistic. Judging decisions by their results — grading the process by the outcome — systematically rewards luck and punishes sound reasoning.
The questions that actually separate them:
- Was the reasoning sound given what was knowable then? Not what is known now. Information that arrived in April is not evidence about a January decision.
- Was the uncertainty acknowledged? A forecast stated as a range with a stated confidence is a different object from one stated as a fact, even if both point the same way.
- Was the position sized for being wrong? The most consequential question. Someone right four times out of five who sizes as though they are always right will eventually lose more on the fifth than they made on the other four.
- Would the same process produce the same decision again? If yes, and the outcome was poor, the process may still be correct. If no, identify what specifically would change it.
The three failures a mid-year review should look for
Confusing the call with the timing. The most common error in market forecasting is being directionally right and catastrophically early. A view that eventually proves correct after a position has been closed at a loss was, in practical terms, wrong. Timing is not a separate skill layered on top of analysis; it is part of the forecast.
Unfalsifiable positions. A forecast that cannot be shown to be wrong cannot be shown to be right either. “Markets will be volatile” and “risks are elevated” survive every outcome. If your January view had no conditions under which you would have conceded error, it was not a forecast.
Selective attention. Recalling the two calls that worked and not the six that did not is the default, not an aberration. The correction is a complete list written in advance, not a recollection of highlights.
What to change, and what not to
The temptation after a poor six months is to overhaul the approach. This is usually a mistake, and the reason is sample size.
Six months is a very small sample. A sound process will produce losing periods regularly, and abandoning it after one of them means switching approaches on evidence that cannot distinguish skill from variance. The pattern of repeatedly adopting whatever worked recently is a reliable way to buy high and sell low with additional steps.
The changes worth making are the ones that would have been right regardless of outcome:
- Write things down with dates. The precondition for every future review.
- State forecasts as ranges with confidence levels. “Probably up” is not gradeable. “60 percent chance of ending the year higher” is.
- Specify in advance what would change your mind. Written at the time of the decision, not afterwards.
- Size positions against the bad case. The question is not what happens if you are right, but whether you can survive being wrong for longer than you expect.
The uncomfortable finding
Most people who conduct this exercise honestly discover that their forecasting record is unimpressive, and that the returns they did earn came from being invested rather than from being right about anything in particular.
That is not a demoralising conclusion. It is the finding that supports the most reliably successful approach available to a non-professional investor: low costs, low turnover, broad diversification, and a plan that does not require correctly anticipating the next six months — the case we set out in index funds versus active management.
The value of the review is not in improving your forecasting. It is in establishing how much weight your forecasts can bear.
