A friend texts me a thesis maybe once a week. Dollar is going to roll over, or this sector is quietly bottoming, or the market is underpricing some event. Good ideas, most of them. And then the very next question is which coin, or which stock, like the idea and the instrument are the same thing. They are not. The idea is the cheap part. Picking the expression is where most of the money is made or quietly bled away, and almost nobody spends time there.
Here is the thing that took me too long to internalize. The same view can usually be worn as spot crypto, a perpetual future, an equity proxy, an FX pair, a commodity, or a prediction contract. Six wardrobes for one body. Each one has its own fees, its own liquidity, its own leverage, its own way of paying you or charging you just to hold the position overnight. Two traders can be right about the exact same thing and one walks away up while the other gives it all back to funding and slippage. The difference is not the call. It is the vehicle.
The idea comes first, the instrument comes last
Before I think about any venue, I try to write the thesis in one plain sentence with no instrument in it. "Risk appetite is rising into the back half of the quarter." Not "I want to buy a leveraged long on this token." The moment an instrument sneaks into the sentence, I have already made a routing decision without meaning to, and usually the wrong one, because I chose the thing I happen to trade most rather than the thing that expresses the view best.
Once the sentence is clean, I list the expressions that would profit if the sentence is true. Rising risk appetite could be a long on a high-beta equity index, a long on a major crypto asset, a short on a safe-haven currency, a long on an industrial commodity, or a prediction contract on some outcome that only resolves yes in a risk-on world. All of them win if I am right. They just win at very different costs, and they lose very differently when I am wrong.
The four things that actually decide the venue
Once I have the candidate list, I run each expression through four filters. I do this roughly in order, and I stop early if something disqualifies itself.
- Carry. What does it cost me just to hold this, per day, doing nothing? A perp charges or pays funding, and in a crowded trade the funding can quietly eat more than you expect to make. Spot crypto has no funding but ties up capital. An equity proxy has a bid-ask and maybe a borrow fee if you are short. A prediction contract has no carry at all, which is one of the underrated things about it, but it has a hard resolution date working against you. If my thesis needs weeks to play out, a high-funding perp is often the worst possible way to wear it even though it feels like the obvious one.
- Liquidity at my size. Not headline volume, depth at the price I actually need. A market can trade billions a day and still have a thin book two ticks off mid. I ask what my full position does to the spread, and whether I can get out in a hurry on a bad day, because the day I need the exit is always the day the book is thinnest.
- Leverage and its cost. If the view needs leverage to be worth doing, that is often a sign the expression is wrong, not that I need more of it. But when leverage genuinely helps, a perp gives it natively, an equity proxy needs margin with its own rate, and a prediction contract gives you a kind of built-in convexity because you pay a small premium for an outcome that pays multiples. Same view, wildly different leverage math.
- Basis and tracking. Does the instrument actually move with my thesis, or with something adjacent? An equity proxy for a commodity view can drift because of the company's balance sheet, its hedging, its management. You can be dead right on the commodity and lose on the stock. The tighter the instrument tracks the actual thing in my sentence, the fewer ways there are to be right and still lose.
A worked example, roughly
Say the thesis is "this metal is in a supply squeeze that lasts a couple of months." The lazy expression is a miner's stock, because it is easy to buy and it is on the exchange I already use. But a miner carries operational risk, hedging books, and general equity beta, so on a red day for stocks it can fall even while the metal rips. The commodity future tracks the actual thing far more tightly, though it comes with roll mechanics I have to understand. A prediction contract, if one exists on a related outcome, costs me nothing to carry but forces the whole thesis to resolve by a fixed date, which may or may not line up with a squeeze that has no schedule. None of these is correct in the abstract. But if my sentence says "a couple of months" and "this metal," the future usually wins on tracking, the prediction contract usually loses on timing, and the stock is the trap that only looks good because it is convenient.
The failure mode I see most often is exactly that. People route to the venue they are comfortable with instead of the one the thesis calls for, and then they blame the thesis when the comfortable vehicle fails them. The idea was fine. The wardrobe was wrong.
A short checklist I actually use
When a thesis survives long enough that I am ready to size it, I run it through this before I touch anything that executes:
- Write the view in one sentence with no instrument in it.
- List every asset class that profits if the sentence is true.
- For each, estimate daily carry over my expected holding period.
- Check real depth at my intended size, not headline volume.
- Ask whether the instrument tracks the sentence or something adjacent to it.
- Pick the cheapest, tightest-tracking expression that clears my size, and only then think about how it gets filled.
None of this is exotic. It is mostly the discipline to separate the call from the container, and to admit that the container I reach for by habit is rarely the one the trade deserves. A router can find the best venue for an order, but it can only route the order you hand it. If you hand it a perp when the thesis wanted a future, the smartest routing in the world just gets you a slightly cheaper version of the wrong trade. The work that matters happens before any of that, in the boring step where you decide what the idea should even be made of.