include_once "include/js/common.js.php";

Why Autopilot Fails ? - part one

by R.Ajay Kumar on Sunday 19 June 2011, 5:28 AM | Category: Forex Market| View: 1350 views

Business models that are up and coming in this age –regardless of the state if  they are models for analyzing market developments or running a main auto manufacturer; characteristically take for granted that history offers a funnel to future upshots. Such an conjecture is more often than not reliable, but at whatever time events fall remote to historical medians, the outcomes can be calamitous. In opposition to this backdrop, we should pause to mull over the preface of computer-based program trading.


Quantitative analysts trained in mathematics and physics have used refined data analytics and modelling proficiencies to appraise securities and expand portfolio-managing theories. The arrival of Quantitative analysts has set aside firms of all bands of colour to trade larger than ever degrees of securities and to lengthen their activities to novel and out of the ordinary instruments.


It is only by using both mathematical and statistical models that organizations have as well been proficient to buy and sell huge dimensions of securities internationally. Within many cases--the computers not only did present advice, but also in point of fact executed stock trades.


By the time it was October 2008, the international stock exchange NYSE accounted that the purported programmed trading, wherein computers implement trades anchored in programs expanded by Quantitative analysts without explicit human interference and that it constituted  almost twenty one percent of all trade (i.e more than 800 million shares on the days the market was open)


In view of the fact that the data that they provide for these analytical formulas arrive from the precedents- it should come as no surprise that the models can have difficulty reacting to bizarre or unprecedented occurrences. When credit markets began to grind to a halt in middle of 2008 and the securities markets went into plummet, the models tried to make out an apposite reaction.


It was known to one and all that they were only programmed to steer clear of precariousness by opting out of securities and into hard cash. Evidently, when a lot of models trading  millions of shares ,each and every one tried to close their business investments and shift into cash, they merely augmented the stock depreciation- showing the way to additional precariousness and as a consequence to further selling.


The models unfortunately were not in the least bit- programmed to comprehend a development in which all and sundry may try to move about to cash at the same time. The outcome was akin to a alarmed guinea pig franctically running on the wheel , afraid of the very bulb that it is helping to be lit.


We are acquainted with the importance of opportune bailouts due to the follies of Wall Street and its regulators, who did not lift up the sound coming back from LTCM in next to no time as they should have (* that's what they get paid a fortune for right?).


Long-Term Capital Management (LTCM) was a hedge fund management firm based in Greenwich (Connecticut) that employed absolute-return trading strategies such as fixed-income arbitrage, statistical arbitrage, and pairs trading and combined them with high leverage. The firm's master hedge fund, Long-Term Capital Portfolio L.P., failed in the late 1990s, leading to a bailout by other financial institutions, under the regulation of the Federal Reserve of United states of America.


That aside it is contemplated that most of the executives botched to make out the susceptibility to precariousness of trading models such as those that LTCM engaged. Primly, they attributed the organization's crumple to a flawed trading tactic more willingly than to a venomous muddle up of a belligerent scheme funded by far above the ground -influence.


Computer models have 3 innate hitches. The 1st  hitch is that the people who fashioned the models didn't quite appreciate the markets. Modellers are proficient in math, computer science, and physics. but most of them are not by and large experts in stocks, bonds, markets, psychology and the minutiae of the system of trade. Modellers are fond of reflecting of markets as competent notions, but these notions can under no circumstances totally report for the muddled and absurd strokes that humans take for expressive raison d'êtres. What is more, at the same time as we have seen, they assemble their models or programs derived from a study of historical market data and test them by showing how well the model would have performed in a given historical situation. Because their programs must have some strictures, modellers inevitably have to leave out unprecedented conditions like the contemporary synchronized unpredictability in worldwide liability, equity, currency, and service markets.


Coming to the 2nd hitch : managers never really try to understand the modellers. The majority of the present generation of higher-ranking executives on the Forex tradi9ng firms lack the technical backdrop to comprehend the models or the algorithms that they are based on , although they know fully well that it is only through those programmes that they cabn effectively control their own firms' trading strategies. For the reason that they are unable to speak the same language as the technical experts creating the models- the management have to bare through complicatedness while framing the queries essential to understand how the models might act in response to diverse conditions.


The predicament at this time goes further than their intellectual capacity. Even if the executives were Quantitative analysts, they may very well not identify with to the extent that they would like with reference to the programs running their businesses. The models themselves & predominantly the interface in the midst of models – which has developed so compoundly that long before the recession came to be , it may as well  have turned out to be impracticable for any human to completely take hold of the kinds and quantities of derivatives traded in this manner or to foresee how the models will interrelate to one another.


The 3rd  hitch is that the models don't identify with one another. Every model carries out its own strategy founded on its calculus for maximizing worth in a particular market, other than that individual models are never till now been able to take into relation the function other models play in driving the markets. Consequently, every program acts in response almost in true time to the strokes of other programs, prospectively compositing precariousness and escorting to uncultivated market swings. The same as we have seen, this has come about recently when a set of models analyzing market data led their own firms to shut down assets and make the most of their cash positions. The snowballing effect intensified the effecting selloff.

Print-Friendly Version Email to a Friend
Related to Why Autopilot Fails ? - part one
Total articles: 7500
    Total Categories: 22
    Updated in Last Week: 1
   Login Id
  Forgot Password?
  Enter Keywords
    Banking and Finance
    Campus Articles
Invite your Friends, your Colleagues, your Group Members, your Seniors, your Team mates...and others.

spacer gif
Management Paradise
spacer gif
arrow gif About Us
MBA Projects
Kartik Raichura
spacer gif
spacer gif
spacer gif
spacer gif
spacer gif Terms & Conditions
spacer gif Privacy Policy
spacer gif Disclaimer
spacer gif Copyrights
spacer gif Contact Us
spacer gif
spacer gif
spacer gif
spacer gif
spacer gif Zeitgeist
Our Network

MBA B School Blog

MBA Buy Entrance Exam Forms



MBA BMM Network



Copyright 2004 - 2018 Management Paradise.