DecisionGuru: Stocks-vs-ETF Decision Engine

Most investing tools tell you what a position is worth today. DecisionGuru answers a harder, more useful question: am I actually better off holding this individual stock, or would the same money — invested on the same day — have done better in a plain ETF? It works that out for every position and for the whole portfolio, fully after Swiss tax, dividends and FX, in a CHF base.

It is a decision-support tool, not advice — it lays the hard numbers out plainly and lets me draw my own conclusions. But being disciplined about those numbers is exactly what has made it, quietly, one of the most profitable things I have ever built for myself.

Why I built it#

I was making investment decisions the way most people do: on a gut feeling, a headline, and a vague sense that a stock was "doing fine." What I had no honest answer to was the counterfactual — whether fine was actually good, or whether a boring index fund would have quietly beaten me while I congratulated myself.

So I built the tool that would tell me the truth I kept avoiding.

The counterfactual engine#

For every franc that ever went into a stock, DecisionGuru mirrors that outflow into a purchase of a benchmark ETF on the same date, rolls it forward, and shows the CHF opportunity-cost delta — the signature azure-vs-gold chart with the gap between the two shaded green when I'm ahead and red when the index would have won.

The verdict engine that sits on top of it is deliberately conservative: underperformance against the benchmark alone never triggers a Sell. A sound, undervalued laggard stays a Hold. That restraint is what stopped me panic-selling good companies on a bad quarter.

Swiss tax, done properly#

Numbers that ignore tax are fiction. DecisionGuru models the real Swiss picture: capital gains tax-free, dividends taxed at your marginal rate (including the income component of accumulating ETFs), withholding-tax reclaim, and wealth tax — all configurable, with a pre-tax / after-tax toggle everywhere. That is the difference between a number that looks nice and a number I can actually act on.

Value analysis, not vibes#

Beside the counterfactual sits a Graham/Buffett-style read on any stock: an intrinsic-value estimate blended from the Graham Number, a two-stage owner-earnings DCF and an FCF DCF, a margin-of-safety verdict against the live price, bear/base/bull scenarios, a reverse-DCF implied growth, and a six-point quality scorecard. Every figure is explicitly an estimate — but it turns "I think this is cheap" into "here is the range, and here is where today's price sits in it."

The price chart is then shaded with Buy / Fair / Overvalued / Sell zones reconstructed as they stood at each past date, rebuilt from the fundamentals that were actually known then — no hindsight, no look-ahead.

What it has actually done for me#

This is the part I care about most. Since I started letting DecisionGuru — rather than my mood — decide when to add, hold or rotate a position, the difference has shown up directly in my own portfolio. Trimming a benchmark-lagging holding I'd have otherwise babied, topping up the right position instead of the obvious one, and staying in undervalued names the tool told me not to sell have together added up to somewhere in the [[low four figures — into the CHF 5,000]] range of gains I would otherwise have left on the table.

And unlike a one-off product sale, this compounds: it is a decision every week, and a better decision every week keeps [[generating returns to this day]]. FlickClean was the first thing I built that earned money directly. DecisionGuru is the thing that quietly makes me money by making me a measurably better investor.

Who it's for#

• Long-term investors get the honest, after-tax verdict on whether their conviction holdings are actually beating the index they could have bought instead — and the discipline to sit still when the numbers say sit still.
• Shorter-term investors dipping into a name get the point-in-time fair-value zones, the margin-of-safety read, and the "who is catching up?" competitor watch — enough to size an entry and know when the thesis has broken.

Same engine, two speeds. In both cases the pitch is identical: stop guessing, and let the after-tax counterfactual make the call.

Status#

Actively developed. It runs local-first — every position and note lives in a local SQLite file, and only market data and FX are fetched (and cached hard). Current work is on deeper scenario workbenches and tightening the reinvestment checks that decide where my next franc should actually go.