Nerds on Wall Street: Math, Machines, and Wired Markets

5.7. Jumping the Trading Cost Hurdle

The next step is to turn the raw material of potential value added into real value added. Lots of investment ideas look great on paper, but don't work out very well in practice. There are a variety of reasons why this is true. In the next chapter, we see the dangers of data mining, a leading cause of disappointment. But even for carefully designed strategies, the cost of trading often eats whatever theoretical alpha might be there. Simple simulations often assume that you can buy or sell all the stock you might want to at your decision price, which is far from accurate. Many stocks trade in very low volumes. I saw one astonishing simulated strategy, disclosed only after tedious legal work, that involved simulated trading of billions of shares of stock in a single firm that traded only a few hundred thousand shares a day in the real world. One consequence of basic supply and demand is that it often costs more to trade in size than the idea is worth. Trading costs, in institutional-size portfolios, are not just the broker's commission. The biggest piece of the real cost of trading is in the market impact, the price movement caused by the trade itself.

Here's how market impact works. A typical market quote will be along lines of "1,000 shares offered to buy at $50, 1,000 shares to sell at $50.25." Anyone can buy or sell at those prices in the advertised quantities. This is how specialists and market makers make a living: buy low, sell high.