Displaying items by tag: algorithms

(New York)

One of the big problems in our growing era of algorithmic trading is herd behavior. For instance, when many trading algorithms are all geared to trade on the basis of momentum, then you tend to get a ton of it at the same time. Well, the problem might be set to get worse as UBS is debuting a new product to help active managers with trade selection using AI. UBS is launching an AI-based product which recommends trade ideas to active managers, something being referred to as the Netflix of asset management. In other words, UBS’ AI recommends a trading strategy which it thinks will suit the manager.

FINSUM: So now even active managers are trying to be enticed into using AI-recommended strategies. The problem with this is that many managers will end being recommended the same strategies, leading to more trading in the same direction.

(New York)

No this is not an article about a liquidity mismatch between ETFs and their underlying products, well at east not entirely. The FT has published a new article by an asset management industry insider arguing that to understand the implications of passive investing, one needs to look more broadly than ETFs themselves. In particular, the piece contends that it is the rise of algorithmic trading which is the true danger, as the technologies which now dominate market trading are agnostic of human-based warnings and insights, and instead simply trade on momentum. This means there are and will be dangerous run-ups and losses in shares. The article points out that only 10% of equity trading now occurs from traditional discretionary human traders. Overall, the piece warns that the current market structure runs very large risks of volatility getting out of hand, and ETFs being forced to dump way more shares than the market can absorb, compounding losses.

FINSUM: This argument is what we would refer to as a “snowball” risk, as it basically discusses the multiple levels of knock-on effects from an initial jump in volatility, which would then be followed by algorithmic selling, then ETF selling, and the cycle continues.

(New York)

As if there were not enough worrying indicators out there, Market Watch has featured a new worrisome measure. The paper interviewed a famous Wall Street quant who says that algorithms which track broad social media sentiment are showing significant risks of a serious decline in equities. The big worries on the public’s mind revolve around the escalating trade war between China and the US. The indicator also informs sector picks, to which strategist Yin Luo said “With U.S. stocks, we are bullish consumer discretionary, technology, and industrials over the medium horizon, and are negative on consumer staples and telecom services, where fundamentals remain relatively weak and momentum has been negative”.

FINSUM: We are always skeptical of these kinds of views because what people say on social media is not a very good reflection of what they are doing in their investment account. Further, there are likely mountains of people being assessed by the algorithms that have no trading/investment account, so their impact is nearly non existent.

(New York)

The markets had a wild day yesterday. Big loss at open, almost back to even, then a really steep fall, and finally, a little rally to close. Bloomberg says that the trading activity has all the telltale signs of algorithms wreaking havoc. For 15 minutes just after 3 pm, the volume of sell orders was so quick and so voluminous that nothing alive could have possibly executed them. The market tanked, plunging to a 1,597-point loss. Interestingly, the involvement of algorithms might help to assuage some fears, as brokers are using that dimension as a way to calm human investors that this was not an all out emotional panic, but rather technology gone wild.

FINSUM: So we know they are deeply ingrained and certainly going nowhere, but why, in principle, are non-human agents allowed to transact in markets? Market-making firms would say they add liquidity, but they certainly exacerbate, or even cause panic too.