The Market Charges Admission, and I Forgot to Pay

I once had a strategy that looked genuinely good. The simulated results were solid, consistent, the kind of thing that makes you sit up a little straighter. Then I added the cost of actually trading it — the real, unavoidable price of doing business in a market — and watched the profit not shrink but vanish. Reduced would have been survivable. It was erased, and then some.

The edge had been real, in a sense. It was real in a world with no frictions. The trouble is that world does not exist, and I had spent weeks admiring a strategy that only worked in it.

The frictionless fantasy

The hidden assumption in a naive backtest is seductive precisely because it is invisible. When you simulate trading against historical prices, it is dangerously easy to assume you could have traded at exactly the price you see on the screen, instantly, as many times as you like, for free. Every one of those assumptions is false.

Real trading has costs, and they come in layers. There are fees you pay to the exchange on every transaction. There is the spread — the small but permanent gap between the price to buy and the price to sell. And there is slippage: the difference between the price you wanted and the price you actually got, because by the time your order arrives the market has moved, or your own order moves it. Each of these is small. That is exactly what makes them so easy to wave away, and so dangerous when you do.

Small costs, enormous when multiplied

The real cruelty of trading costs is in the multiplication. A tiny toll per trade feels obviously negligible — who cares about a sliver of a percent? But a strategy does not trade once. A busy one trades thousands upon thousands of times, and it pays that negligible toll every single time.

So the cost does not politely add up in the background. It grinds against you, trade after trade, until it has consumed an amount that bears no resemblance to how small it looked per transaction. The more active the strategy, the more admission it pays to the market — and some of my most active strategies were paying their way straight into the ground without my noticing, because I was looking at the gross profit and the market was quietly charging me on the side.

The strategies that looked best were the most fragile

Here is the perverse twist that took me a while to see. In a frictionless backtest, the strategies that looked most profitable were very often the ones that traded most frantically — the hyperactive ones, snatching at countless tiny opportunities. And those hyperactive strategies are precisely the ones that realistic costs punish the hardest, because they pay the toll the most times.

Which means the frictionless version of my backtest was not a neutral measurement. It was actively flattering exactly the strategies most likely to die in reality. My “best” ideas, ranked by frictionless profit, were disproportionately my most cost-fragile ones. The leaderboard was, without my realizing it, sorted partly by how badly each strategy would collapse the moment it had to pay its own way.

Slippage is worst exactly when you need it least

There is a further sting. Slippage is not a constant tax you can average away. It is small when markets are calm, deep, and liquid — and it balloons precisely in the violent, fast, thin moments when prices are lurching and everyone wants to act at once. In other words, the cost of trading is highest exactly when your strategy most wants to trade.

This correlation is brutal, and a naive model hides it completely. If you assume a single flat cost per trade, your simulation quietly pretends that executing in a calm market and executing in a chaotic one cost the same. They do not. The real toll is worst on the worst days, when the road is most dangerous, which is exactly when a strategy is most tempted to floor it.

Modeling costs honestly is depressing and essential

The fix is not clever, only disciplined: build realistic — even deliberately pessimistic — cost assumptions into the simulation from the very beginning, not as a cosmetic adjustment bolted on at the end. Assume you pay the fees. Assume you cross the spread. Assume slippage, and assume it gets worse when conditions get worse.

Doing this is genuinely depressing, because it makes every result look worse, sometimes dramatically worse. That is the entire point. A strategy that only works when trading is free is not a strategy. It is a daydream with a chart attached. And the cheapest possible moment to discover that is in a simulation, before a single unit of real money has been committed to the fantasy — not afterward, when the market is collecting its admission in cash.

The deeper principle: price is an invitation, not a promise

Underneath all of this is a simple shift in how I read a price. A number on a screen is an invitation, not a guarantee. It says “trades have happened around here,” not “you may transact any quantity you like at this exact level, right now, for nothing.” The entire hard reality of trading lives in the gap between the price you see and the price you actually get when you try to act on it.

Most of what separates a paper edge from a real one is simply whether it survives that gap. A great deal of clever-looking profit is really just the gap, unmodeled — money that was never going to be there once the costs of reaching for it were counted.

What this changed

After that, I started treating costs as a first-class adversary rather than a correction factor sprinkled on at the end. The first question I now ask about any promising result is not “how much does it make?” It is “how much does it make after it pays full admission — at the frequency it actually trades, in the conditions it will actually face, including the bad ones?”

Most edges do not survive that question. They were never really there; they were the frictionless fantasy wearing the costume of a discovery. But the few that do survive it are worth far more than the many that did not, because they are the only ones that were ever real. The market always charges admission. The only useful question is whether what is inside is still worth the price of the ticket.

— No signals, no returns, not investment advice.