Do You Want to Be Right, or Do You Want to Make Money?

The question ends more debates in this office than any other, and it is not rhetorical. Two people can both be right and only one of them gets paid.

A version of it: should stock-based compensation sit inside EPS? One side says yes, obviously — it is a real expense, shareholders bear it through dilution, and excluding it flatters every multiple you calculate. That argument is correct. We agree with it.

And it does not settle what the stock is worth. The pool of institutional capital that sets the marginal price in large-cap technology strips it out. They screen on adjusted numbers, they set price targets on adjusted numbers, and they transact at prices derived from adjusted numbers. If you build a multiple that nobody in that pool is using, you have not built a valuation. You have built an opinion about accounting.

So we carry both. The business gets underwritten on the number we believe — dilution included, because it is real and it will eventually be paid. The security gets underwritten on the number the buyer is using, because that is the number the price is made of. Confusing those two objects is expensive in both directions.

Keynes described this in 1936 as a newspaper beauty contest where readers win by picking not the prettiest face but the face they expect other readers to pick. The joke is that you quickly find yourself guessing what everyone thinks everyone thinks. The serious part is that in the short run, price is the output of that process and not of yours.

The second example is cleaner because it resolves in a single morning. A company is going to beat consensus. Good — and then the question that actually matters: does the stock go up?

Consensus is published. The bar the stock trades against is not. If buy-side expectations have already migrated above the printed sell-side number, then a beat against that number is a miss against the real one, and the stock sells off into a headline that says it beat. Nobody involved has done anything wrong. The analyst was right about earnings and wrong about the stock, which are two different jobs.

Now the part that keeps this from becoming an argument for intellectual surrender, because taken carelessly that is exactly where it goes. Modeling what the crowd believes is necessary. Adopting what the crowd believes is fatal. Every bubble in history recruited its final buyers from people who had stopped separating those two activities.

The market’s model tells you the path. Yours tells you the destination. You need both, and you need to keep them in separate columns, because the moment the second one gets quietly overwritten by the first you are no longer investing — you are extrapolating, with leverage, in a crowd.

This is also the honest reason we care so much about a definable catalyst. Being right eventually is not a strategy; it is a hope with a carrying cost. A catalyst is the mechanism that forces the two numbers to meet on something resembling a schedule — a separation, a refinancing, a change in who is required to own the security. Without one, a cheap thing can stay cheap for longer than the argument for owning it survives.

And it is why the horizon matters. Being early is indistinguishable from being wrong to anyone who has to explain a quarter or meet a redemption, which is why so much capital cannot hold a correct position long enough to collect on it. Not having a fund clock does not make you right. It makes being early survivable instead of terminal.

What keeps all of this disciplined is writing down which number is which before the position exists: what we think the business earns, what the market is capitalizing, and what specifically would have to happen for those to converge. When the trade works, that record is the only way to know whether we were right or merely paid.

So: right, or paid? You want both, and most of the time you can have both. When you cannot, the expensive mistake is not choosing wrong. It is failing to notice there was a choice.