In a world where the minute, second and even millisecond that you pull the trading trigger can materially affect the return you get, managers of mutual funds too often are slow to buy when prices are rising and too fast when prices are dropping.
That’s the result of a study of thousands of orders in a two-year database of stock transactions in the United States, executed by Pipeline Trading Systems, a supplier of tools to institutional trading desks. The company analyzes high-frequency trading techniques and how they affect mutual funds.
“On average, when the market dictates the rate of your execution, you will tend to trade too fast, when it’s detrimental, and trade too slow when it would be better to trade quickly,” said Pipeline Chief Executive Fred Federspiel.
The study found that trades that had the best return over the first three months of their lives were executed much slower than average. Meaning: Gains were left on the floor, because of slow reaction.
And the trades that ended up hurting the funds the most on that three-month time scale were executed much faster than average. Meaning: Institutional traders took bigger hits than they should have, if they could have a better idea of which way a given stock was likely to move in that time.
What’s driving that effect is the success of high-frequency traders. They are able to make these short-term predictions, Federspiel maintains.
“They deliver liquidity into a market at a high rate, when it is not good for their counterparties,” he said. “And starve the market of liquidity when it would be good to trade quickly.”
This is “adverse selection,” high-frequency style. And Pipeline is trying to put mutual funds on an equal footing, with one new measurement benchmark and, separately, a tool that lets institutional traders project what’s likely to happen during a trading day for a particular stock.
Then, that tool makes a buy or sell recommendation, suggesting how fast to move. Finally, it gives the trader a chance to either “fast forward” the execution of the trading decision as events unfold or, if so desired, swipe all existing liquidity out of the market, in one fell swoop.
The measuring stick is something called a “momentum-adjusted cost benchmark.”
The benchmark estimates the impact on the price of a stock if an institution moves on it, adjusted for the underlying momentum in the stock that already exists. Then it adds onto that a computation of the short-term gain or loss in the stock, after the impact is removed.
So, over time, users get a sense of which trading strategies have worked and which haven’t. “It does help you identify trading techniques that are subject to more of this high-frequency trading adverse selection,” he said.
That allows a firm to weed out either trading methods or traders themselves that have tough times coping with the effect of high-frequency trading tactics.
Its Alpha Pro tool, introduced last year, takes the process and looks at hundreds of statistical factors that influence stocks.
This takes into account whether trading is taking place more heavily on the offer side of the market or the bid side, how much trading is taking place in dark pools, and what kind of news has played into results. It then makes a recommendation on a trading strategy, based on its statistical analysis.
The recommendation can include whether to “front load” the buying or selling at the beginning of the time frame involved, whether to access block trading venues such as Pipeline’s own crossing pool or that of Liquidnet, the leading system for block trades—or not. And what percentage of overall trading in a stock should be taken out of the market at any given moment.
The system flags a trading desk if too much liquidity comes on the market, when a market participant first starts sending sell orders into the market, for instance. The system also can signal a hidden or “iceberg” order, when a thousand shares of a given stock keeps coming onto the market, at a particular price.
“All of these signals are pulled together into an execution framework that minute by minute helps the institutional trader control his execution speed and helps him push back against the starvation of liquidity when well-informed high-frequency traders are trying to pull out of the market,” Federspiel said.
The tool can insist on grabbing liquidity, for instance, by “crossing the spread.” That entails dropping one’s bid price and switching over, by design, to the offer price, before offers go away.
Making adjustments in real-time is what puts institutional traders on the same footing as high-frequency traders, he believes.
This may even give the institutional traders a leg up. That’s because, Federspiel notes, the mutual fund manager or institutional trader knows how big an order he or she has to move and how much that accounts for of a given stock’s average daily volume of trading.
The process is not perfect. “There’s a lot of uncertainty even with the best predictive tools,” Federspiel said.
That’s where real-time adjustment of results comes in, according to Stephen D. Marchini, Pipeline’s managing director for program trading.
Alpha Pro and its other predictive tools can show, for instance, what the expected underlying return or “alpha” in a stock is for a given day, how much of that is coming from underlying market momentum and what the impact on that return will be if the institution gets involved.
That, in turn, lets a trading strategy be customized for the time of day when the biggest gains will come or be missed.
And if the trader knows of some exogenous factor that is likely to come into immediate play that the market has not yet worked into prices—an oil refinery accident, for example—the trader is given “fast forward” buttons to speed up execution of a trading program and a “Big Boy” button to grab all shares available, from all sources.
You may be able to move faster than predatory rivals. But there’s a cost.
“If you have 100,000 shares [on the market] and 5,000 is offered, you have to be prepared to buy the whole 100,000 shares,” Marchini said, when you decide to clean out the market.