Run a valuation
The five valuation methods, the Auto default, and how a method preference works.
When you review a new analysis before running it, the Valuation Method control defaults to Auto: dbot studies the company first and then picks the method that fits it. If you have a strong view, you can set a preference instead. dbot treats it as a strong default, follows it in most cases, and tells you when it can't.
A bank has no capex and no working capital worth modeling, and its debt is raw material rather than financing, so a cash-flow model built for an operating company has nothing to grip. What a bank does have is a regulatory constraint: it must hold Tier 1 capital against its risk-adjusted assets. Damodaran's answer, the model behind his published Citi valuation, treats that capital as the reinvestment. Free cash flow to equity is net income minus the change in Tier 1 capital, and it is discounted at the cost of equity, because in a bank there is nothing to value but the equity.
What that measures is dividend capacity, which is often nothing like the declared dividend. Buybacks carry much of the return to shareholders and never show up in a dividend series. A growing bank spends its earnings funding the capital behind new assets, so its capacity sits well below what its payout ratio suggests. And an undercapitalized bank has a negative FCFE: it has to raise equity, and the model prices that future dilution today. A dividend model cannot express any of this, because a dividend cannot go negative.
The mechanics are Damodaran's own. Five forecast years fade return on equity and the Tier 1 ratio linearly toward their year-five targets. The steady state that follows pays out one minus growth divided by return on equity, the retention identity a stable bank has to obey. Two costs of equity are carried, one for the forecast years and one for the perpetuity. And the risk of a regulatory takeover stays out of the discount rate entirely: an explicit probability of failure multiplies the value per share by (1 - p) at the end. The Excel model you download is Damodaran's own bank spreadsheet, filled in with the run's numbers.
After the research steps finish, a plan step reads the reported segments, the reconciliation quality, and the research synthesis, then decides the shape and explains its reasoning. Follow the adaptive pipeline to see that decision step by step.
A preference is a strong default, not an override. dbot adopts your method whenever the company's facts allow it, even when the evidence merely leans another way. Two things still outrank it:
Every report shows the composition plan it actually ran, with the plan step's rationale. If your preference was overruled or downgraded, that is where you'll see why: the rationale names your preferred method, the rule that overrode it, and the shape dbot chose instead.