Risk controls at machine speed

Graduate · enrolled learners

This lesson opens with Latour Trading, 2014.

What happened

Latour Trading was a high-frequency trading firm, owned by Tower Research Capital, that traded ETFs and their component stocks and futures at machine speed, at times as much as 9 percent of all US equity trading volume in 2011, and in September 2014 it agreed to pay 16 million dollars, the largest penalty ever for net capital rule violations (the previous high was 400,000 dollars), to settle SEC charges, without admitting or denying them. The rule requires a broker-dealer to hold enough net capital against the risk of its positions, but throughout 2010 and 2011 Latour calculated the required haircuts with a flawed method, offsetting positions against hypothetical positions it did not hold, using wrong index data for some international ETFs, netting futures it should not have netted, and missing some positions through a programming error, so it understated its risk and fell short of its required minimum net capital on 19 of 24 reporting dates, by about 2 million to 28 million dollars. Nothing dramatic announced the problem; the firm traded enormous size continuously, and a control that was quietly wrong, the calculation of how much capital its risk demanded, let it run beyond what its capital supported for two years. A market maker takes on risk at machine speed, so the controls that measure and limit that risk have to be correct and automatic, because when they are wrong the firm accumulates exposure faster than a human could ever notice, and the miscalculation is discovered only after the risk has been carried for a long time.

The decision point

A market maker that trades at machine speed accumulates risk at machine speed, so the controls that measure and limit that risk, capital, position, and loss limits, must themselves be correct and automatic, because a control that is wrong or absent lets exposure build faster than any human could watch, invisibly, until it is discovered. Latour Trading is the case: a high-frequency trading firm used a flawed calculation of the capital its risk required, understating it, and ran short of its required capital on 19 of 24 reporting dates over two years, at a 16 million dollar cost when the SEC brought the case. This closes the market-making course on the theme running through it: market making is a risk business, and at machine speed the risk is only as controlled as the automated systems measuring it, so a firm whose control is quietly wrong is not safe merely because nothing has gone wrong yet. The danger of a bad control at machine speed is that it fails silently: the firm keeps trading, the exposure keeps growing, and because the control that should raise the alarm is the very thing that is broken, there is no alarm, only a gap that widens until an outside check finds it. So the discipline is to treat the risk and capital controls of a fast market maker as critical systems in their own right, to verify that they correctly measure the risk being taken rather than assume they do, to make them automatic because human oversight cannot keep pace with machine-speed accumulation, and to have them independently checked, because Latour shows that a market maker can run for a long time on a control that is wrong, carrying risk it cannot support, precisely because the broken control is the one that was supposed to say so.

Recorded loss

$16,000,000

Orientation and Year One are open: anyone can read them without an account. From Year Two onward the lessons are for enrolled learners, because progress through the later years only means anything if it is tracked against a record.

It is free. We do not sell the list and there is nothing to buy at the end of it.