Your Trading Style Sets Your Trading Bill: Keim and Madhavan

Ask what a trade costs and most models will tell you it depends on the trade: how big it is, how fast you go, how liquid the stock is. All true.

Donald Keim and Ananth Madhavan got hold of the actual order-level records of a large number of institutional investors, covering many billions of dollars of real equity transactions, and found something the models mostly miss. A very large chunk of what a trade costs depends not on the trade at all, but on who is doing it and why.

The problem: cost is not a property of the order

Consider two funds that both buy a million shares of the same stock on the same day.

Fund A is a value manager. It buys stocks that have been falling, that nobody wants, and that it intends to hold for three years. It is in no hurry. If it does not finish today, that is fine.

Fund B is a momentum manager. It buys stocks that are already rising, because they are rising, and it needs to be in before the move is over. It is in a hurry, by construction.

The orders look identical on paper. The costs will not be. And the reason is not that Fund B has worse traders. It is that Fund B's entire investment philosophy commits it to demanding liquidity at exactly the moments liquidity is expensive.

Keim and Madhavan's contribution is to show, with real data across many institutions, that trading costs vary systematically with investment style, and that this is a first-order effect rather than a curiosity.

The key idea via analogy: the shopper and the fireman

Think about buying something at an auction.

The patient shopper turns up, looks at what is available, and if the price is not right, walks away. She might come back next week. Because she is willing to walk away, she is a liquidity supplier: she buys when others are desperate to sell, and she is paid for her patience with a good price.

The fireman must buy the fire hose, today, at any price, because there is a fire. His need is urgent and it is not negotiable. He is a liquidity demander, and he will pay whatever the market asks.

Both may be excellent at their jobs. But the fireman's job structurally requires him to pay up, and no amount of trading skill changes that.

The empirical findings map onto this cleanly.

Costs rise with trade difficulty. Bigger orders, in less liquid stocks, cost more. This is unsurprising and it confirms what everyone believed, but it is important to have it measured on real institutional data rather than assumed.

Costs depend heavily on the trader's own characteristics, not just the stock's. Investment style and order submission strategy matter enormously. A manager whose style demands immediacy pays for immediacy. A manager who can be patient captures the other side of that trade.

The demand for immediacy is the central variable. Strip away the details and this is the finding. Trading cost is essentially the price of impatience. Managers whose strategies require them to act quickly, momentum and technical strategies in particular, are structurally committed to paying that price. Managers whose strategies let them wait, deep value in particular, are structurally positioned to be paid it.

Costs are economically significant. This matters because it is easy to dismiss trading costs as a rounding error next to the alpha you are chasing. The data says they are large enough to meaningfully erode returns, which means a strategy's gross performance and its net performance can tell very different stories.

The paper also compares costs across exchanges, finding that stock-specific factors such as where the stock is listed affect costs too, which was a live policy question at the time.

Why it mattered

• It reframed transaction costs as a strategic variable, not an operational one. If your investment style determines your trading costs, then choosing a style is partly choosing a cost structure. That belongs in the portfolio construction decision, not in a back-office report.
• It explains why paper strategies die in practice. A backtest that ignores costs will happily favour high-turnover, momentum-flavoured strategies, precisely the ones whose real-world costs are highest. Keim and Madhavan's numbers help explain why so much beautiful backtested alpha evaporates on contact with the market.
• It gives an economic account of the value premium's trading edge. Value managers, by buying what everyone else is dumping, are supplying liquidity and being compensated for it. That is a real and often overlooked component of the value strategy's return.
• It set the standard for institutional cost studies. Using order-level institutional data, spanning multiple managers, styles and exchanges, was a methodological leap over the trade-level public data everyone else was using.
• It made capacity a serious question. If costs scale with size and with impatience, then a strategy's viable capacity depends on both. That is now a routine part of manager due diligence, and this paper is part of why.

The honest limitations

• Style and cost are tangled together and hard to separate. Momentum managers pay more. Is that because momentum requires immediacy, or because momentum managers happen to trade smaller, less liquid, more volatile stocks? Both are true, and disentangling the causal contribution of style from the composition of what they trade is genuinely difficult.
• The sample is a set of institutions who agreed to share their data. That is not a random sample of the industry, and the ones who volunteer their execution records may not be typical.
• The market studied no longer exists. This is data from the era of floor trading, wide spreads, fixed commissions in living memory, and no electronic execution algorithms. The absolute cost levels are historical curiosities. The structural finding, that impatience is what you pay for, has held up far better than the numbers.
• It measures cost, not net value. A momentum manager who pays a lot to trade might still be doing the right thing, if the alpha they are chasing outruns the cost. The paper measures one side of that ledger.
• Attribution of impact remains hard. Separating the price move caused by the manager's own trading from the move that was happening anyway, which is precisely why the manager was trading, is the perennial difficulty in this literature and it is not fully solved here.

The one-line takeaway

Keim and Madhavan showed with real institutional order data that trading costs are driven as much by who you are as by what you trade, because a manager's investment style determines how urgently they must trade, and urgency is precisely the thing the market charges you for.