Most investment memos I have reviewed cannot be wrong. They contain a fair value, a set of qualitative reasons the business is good, and a target price. Six quarters later, when the position is down and someone asks whether the thesis broke, there is no way to answer, because the memo never stated a condition that could fail. The discussion becomes a debate about sentiment, and the position gets held or cut on the basis of how the room feels.
A reverse DCF fixes this structurally rather than through better writing discipline. Instead of asserting a value, you state what the current price requires, state what you expect instead, and name the gap. All three are numbers. The thesis becomes a documented disagreement with the market, and a documented disagreement can be tracked, tested, and attributed when it closes.
The thesis as a stated disagreement
The mechanics are the same reverse DCF anyone runs. Fix the discount rate from house policy so it cannot be tuned, take enterprise value as given, solve for the free cash flow growth over the explicit period that reconciles the two. What changes at an institutional level is what you do with the output.
The memo opens with three lines rather than a conclusion. Market implied path: the growth and margin trajectory the current price requires. House path: what your forecast delivers instead. Gap: the difference, expressed both in growth points per year and in value terms.
That structure forces something useful before a single share is bought. It requires you to name what the market believes, which most analysts have never explicitly done for the names they own. In my experience roughly one memo in four does not survive this step, because the analyst discovers the market implied path is broadly what they were forecasting anyway. There was no disagreement, therefore no thesis, therefore no position. Catching that before deployment is worth more than most of the rest of the process.
What makes the gap falsifiable
A gap is only testable if you decompose it into observables with dates attached. Three fields do most of the work.
The first is the driver split. State how much of your gap comes from revenue growth, how much from margin, and how much from capital intensity. A gap that is entirely a margin call is a different risk from one that is entirely a volume call, and it is monitored differently. It also stops the common failure where an analyst is right about the top line, wrong about margins, and records the thesis as broadly correct.
The second is the reporting milestone. For each driver, name the specific line in a specific future report that will move your confidence, and state the threshold. Not "we will monitor margins" but "gross margin below a stated level in the next two reported periods contradicts the path". This is the field that gets skipped, and it is the field that makes the difference between a thesis and a preference.
The third is the falsifier that is not about the company. Expectations gaps close for reasons other than fundamentals: the discount environment moves, or the market simply revalues the sector. Write down in advance what evidence would tell you the gap closed for reasons you did not predict, because that outcome is a lucky win and should not be logged as skill.
The Company Valuation Engine screen above computes a mispricing figure per company alongside its composite score and verdict, which is a useful triage input and a poor thesis input, for a reason worth being explicit about. A single mispricing number compresses the entire disagreement into one figure and discards the decomposition. Two names both showing a 26 percent gap can be completely different trades, one resting on a margin assumption testable in two quarters and the other on a terminal assumption testable in a decade. Use the column to populate the work queue. Do the decomposition yourself.
Exit rules that follow from the implied path
The strongest argument for framing positions this way is that the exit rule writes itself, and it is mechanical rather than discretionary.
There are exactly three ways a gap thesis ends. The gap closes because the price moved toward your path, which is the win, and the exit is triggered by the implied path converging on your forecast rather than by a target price. The gap closes because your path moved toward the market, which is the thesis breaking, and the exit is triggered by your own forecast revision crossing a stated threshold. Or the gap persists past the horizon you gave it, which is the ambiguous case and the one that needs a rule most.
Rerunning the reverse DCF on a fixed cadence, monthly or at each report, gives you the first two rules for free. The implied growth path is recomputed at the new price, the gap is remeasured, and when it reaches a stated fraction of the original you trim or exit on schedule. No judgement required, no discussion of whether there is still upside, because the arithmetic answers it.
The third case needs a time stop, and it should be set from the milestone structure rather than from the calendar. If your gap rested on margin expansion that should be visible within four reporting periods, and four periods have passed with the margin flat, the thesis has failed even though the price has done nothing. Positions that die of old age without a stop are how a book fills with names nobody wants to defend and nobody will sell.
Attribution when the position closes
The reason to do all of this becomes clear only at the post mortem, where the gap structure lets you attribute the outcome instead of narrating it.
Four questions, each answerable from the file. Was the market implied path correctly characterised at entry, or did you mis-specify what the market believed? Was your own path right, driver by driver? Did the gap close, and through which mechanism? And was the size appropriate to the confidence the decomposition supported?
The combination that matters most is being right on the path and wrong on the outcome, or wrong on the path and right on the outcome. Both are common and both are invisible in conventional post mortems, which tend to code profitable positions as good decisions. A team that consistently forecasts drivers accurately while gaps fail to close has a market timing or crowding problem, not a research problem, and those need entirely different fixes. You cannot see that distinction without the decomposition, and once you have it across thirty closed positions it is usually obvious.
Capacity, crowding, and the gaps that never close
One structural caution that belongs in any process built on expectations gaps. The gap is a disagreement with the market, and the market is not a fixed counterparty. If the disagreement is one that many similar processes will reach from the same public inputs, the gap can close before you finish building, or it can persist for years because the marginal holder is not running your model.
Both failure modes are worth measuring rather than worrying about. On the first, track the time between your gap identification and the first meaningful price move toward your path, across your closed positions. If that number is consistently short, you are late to a crowded signal and the gap you are measuring has already been arbitraged in the names where it works. On the second, look at your open positions with the largest gaps and the longest holding periods. A gap that has been 25 percent for three years is not an opportunity being ignored. It is usually evidence that your discount rate policy differs from the market's view of the risk, and the honest response is to review the policy rather than the position.
Neither check requires new data. Both come out of the same file you already built to make the thesis falsifiable, which is the argument for the structure in the first place: the fields you write down to be honest at entry turn out to be the fields you need to learn anything at exit.