Paper Versus Reality: Perold and the Invention of Implementation Shortfall

Every quant has had this experience. The strategy looks glorious on paper. You run it live and it is mediocre. Somewhere between the idea and the money, a large chunk of the return has evaporated, and it is maddeningly hard to say where.

In 1988, Andre Perold wrote a short, brutally clear paper that named this gap, measured it, and broke it into pieces. He called it the implementation shortfall, and it is now the standard benchmark used by essentially every institutional trading desk on earth.

The problem: the paper portfolio does not pay for lunch

Perold's setup is elegantly simple. Imagine two portfolios.

The first is the paper portfolio. The moment your model says "buy," this portfolio buys, instantly, in unlimited size, at the price currently on the screen, with no commissions. It is the portfolio your backtest owns. It is a fantasy, but it is a well-defined fantasy.

The second is the real portfolio. This one has to go out into an actual market and buy the stock over hours or days, paying commissions, pushing the price up as it goes, and sometimes failing to get the whole position on at all.

The implementation shortfall is simply the difference in performance between those two portfolios. That is the whole idea. It is the total price of turning an idea into a position.

Framed that way, it becomes obvious that the shortfall is not just about broker commissions. Commissions are the part everyone sees, and they are usually the smallest part.

The key idea via analogy: the true cost of a concert ticket

You see a concert ticket advertised at fifty pounds. What does going to that concert actually cost you?

• The fifty pounds on the sticker. This is the explicit, visible, easy-to-measure part. In trading, this is your commission and fees.
• The booking fee you did not notice until checkout, and the fact that by the time you got to the front of the online queue the cheap seats were gone and you had to buy a dearer one. In trading, this is market impact: the price got worse because you were trying to buy.
• The fact that you dithered for two days before deciding, during which the price went up. In trading, this is delay cost: the market moved between the moment the decision was made and the moment the order actually hit the market.
• The seats you never managed to buy at all because they sold out. If the concert turns out to be legendary, the value you missed is a real cost even though you spent nothing. In trading, this is opportunity cost: the part of the order you never filled, on a stock that then ran away from you.

Perold's insight is that a serious accounting of trading cost must include all four, and the last three are usually much larger than the first. He decomposes the shortfall into exactly these components: explicit costs, execution costs, delay costs, and the opportunity cost of unexecuted shares.

The opportunity cost piece is the one that catches people out, and it is the sharpest part of the paper. It means you cannot game the benchmark by simply not trading. A trader who only fills the easy orders and quietly abandons the hard ones will show wonderful execution statistics on any naive measure. Under implementation shortfall, the orders they walked away from show up as a cost, because the paper portfolio would have owned them. There is nowhere to hide.

Why it mattered

• It gave the industry an honest scoreboard. Before this, trading desks were often measured against benchmarks that were easy to flatter. Implementation shortfall measures against the price at the moment the decision was made, which is the only benchmark that captures what the investor actually lost.
• It aligned the trader with the portfolio manager. A trader optimising implementation shortfall is optimising exactly the thing the fund's investors care about: how much of the manager's idea actually made it into the account. Other benchmarks can pull the trader in directions that look good on their own report card and cost the fund money.
• It named the transaction cost problem, and naming it made it a research field. Almost the entire optimal execution literature, from Bertsimas and Lo onward, is in effect trying to minimise Perold's number. The framework told everyone what the objective function was.
• It birthed the "IS algorithm." Walk onto any modern trading desk and you will find an execution algorithm literally called Implementation Shortfall, which trades aggressively at the start to reduce delay and opportunity costs, then eases off. It is Perold's paper turned into software.

The honest limitations

• The decision price is slippery. The whole measure is anchored to the price at the instant the investment decision was made. But when exactly was that? When the PM had the idea in the shower? When they typed it into the system? Firms make different choices here, and the choice materially changes the number.
• It punishes the trader for things they cannot control. If the market rallies hard for macro reasons between decision and execution, the shortfall on a buy order looks awful, and the trader had nothing to do with it. Untangling market drift from genuine execution quality is genuinely difficult, and a single bad print can swamp a quarter of good work.
• It is noisy on individual orders. Shortfall on any one trade is dominated by ordinary price volatility. It only becomes a meaningful signal about skill when averaged over a large number of orders, which many desks do not have.
• It says what to measure, not how to trade. The paper hands you an objective. It does not tell you the optimal schedule for hitting it. That took the next twenty years of the literature.

The one-line takeaway

Perold defined the implementation shortfall as the gap between the portfolio your idea deserved and the portfolio you actually got, and by insisting that commissions, market impact, delay, and the orders you never filled all belong in the same number, he gave the industry the one execution benchmark that cannot be gamed by simply trading less.