Algorithm based trading systems were developed first in U.S.A. The idea, of developing an internally comparable system in India germinated from a series of discussions between Saurabh Sood, Founder and C.E.O. of Marketopper with a renowned expert from the trading field.

The initial discussion group felt convinced that the technology and knowledge were locally available to design and develop such systems. Moreover, the major stock exchanges in India had already initiated on-line trading with live feeds in to broker terminals. Consequently, the external infrastructure was appropriate for testing the viability of such a system. However, the concept was greeted with skepticism and doubt outside the internal discussion group.

The reason for this was two-fold: firstly, there was widespread skepticism regarding a system based totally on technical analysis - 'techno-voodoo' to quote one highly respected investment analyst - and secondly, there were serious doubts about whether such expertise actually existed in India.

In view of such perceptions, it was decided that the first priority was to establish the overall viability and credibility - in terms of design, development and performance - of the conceptualized system.

The Approach
The idea was to develop an initial model for the Indian market, create a live trading track record by trading on this model. To create a knowledge domain which would evolve from such an exercise. To leverage this knowledge to create efficient and scalable processes which then can be employed to create multiple strategies for multiple markets, resulting in robust returns with controlled  drawdowns.

Consequently we were able to mark out the major milestone in our journey which would ultimately lead us to achieve our above stated objective

Major Milestones Traversed
The original objective was to develop a system for trading on the US stocks. However, it was accepted that such a system - designed and developed in  India - would lack credibility and therefore also lack acceptance in the American market.
Consequently, the approach stage was conservatively defined and phased as follows:

Phase 1: Development of a proprietary indicator, based on which the first strategy was created for the Indian Market. While on the whole the strategy was a huge success, it highlighted the fact that dependence on one single model even while trading multiple markets through it can expose the whole portfolio to large drawdowns, if the market undergoes a structural change.

Phase 2: The experience garnered from phase 1 had taught us to focus our efforts towards creating multiple strategies. However creating robust strategies which would consistently beat the market returns, called for huge amount of data crunching and consequently much higher development time.

Phase 3: To meet the twin challenges of high processing requirement and severe time constraint, we realized that we need to refocus our efforts into developing artificial intelligence based Logic Engines. These engines developed on cutting edge technologies like Supervised Learning and genetic algorithm running on parallel networked environment can create millions of condition set each signaling a unique buy or a sell order. The development of these logic engines brought about a paradigm shift in our working style radically cutting down the development time, exponentially increasing the data crunching ability thus helping us create multiple 100 strategies across different markets and based on different time frames. The strategies then only become the outputs of the various building blocks which have been created.

Phase 4:With the help of these logic engines we were soon able to create different products like:

• An Intraday system based on Information Technology stocks
• An Intraday system based on Nifty; the index future
• A Carry Forward system based on Nifty
Interestingly each of the above products in turn were composed of hundreds of independent strategies (trading logics) running simultaneously.

Phase5: While risk management was always of primal importance for us, but earlier our focus was more towards risk management at the strategy level. But with each of our products running on numerous logics, we realized there was an urgent need to apply risk management at the portfolio and product levels also. Further with so many models taking high frequency trades the need to automate the monitoring process had also become imperative.

So for product level control we created overlays like Drawdown Controller and Portfolio enhancer to, offset risk while enhancing the returns.

The Drawdown Controller, tries to predict high risk situations at the product level by creating a market position versus open profit matrix  of the product .It  then initiates action against our main product as and when the probablity of high  drawdown scnario increases.

The portfolio risk evaluator, tries to discover highly profitable situations when  Buy/ Sell trades are generated though dissimlar logics. The notable observation for trades arising out of  such combinations, is that they have a higher percentage profitability as well as a higher profit per share realization. So we execute these trades on a higher exposure level thus ensureing higher portfolio level profitability with theoretical lower minimal increase in risk level.

  We do micro analysis of our products using Portfolio Risk Evaluator, an in-house utility. Portfolio Risk Evaluator (PRE) helps in analyzing the portfolio based on numerous independent parameters. It helps in reviewing the portfolio based on time of entry of strategies, generic nature of strategies (e.g. in trend strategies, contra strategies), it helps us in understanding how strategies based on different time frames have performed, how strategies of different analyst have performed etc. We can also analyze our portfolio based on a multiple of these above mentioned parameters taken together.

The Results
Between April and September of 2001, the first system was put through a dry run for 6 months. The results significantly exceeded expectations and were substantially superior to the entire batch of target comparative systems. The dry run was followed up by 5 months of live trading between September 2001 and March 2002. The major take-outs of both the dry as well as live runs were:

1. The dry run produced an annualized return well in excess of 100% net of costs.
2. The live trading generated a return of 50% net of costs.
3. During the dry run as well as the live trading, the System perfectly captured all major market trends - including the Tech. Stocks meltdown - thereby demonstrating it's ability to perform in both bull as well as bear markets.

The returns of subsequent years proved to be equally good, however they were achieved over a wider product range over a consistently increasing exposure base. We have provided the certified result of last 4 years, which would provide a very fair indication of “The journey so far”.

RESULT VALIDATION
Prudence and predictability determined the key elements of the validation process.

[a] The capital employed is computed on the basis of the most conservative methodology - i.e. the capital and margin requirement of the Exchange along with the highest draw down incurred during the period of trading. For instance, if a gain of $100000 is followed by a loss of $20000, then it is assumed that the Capital Employed is the Margin plus $20000 - the highest draw down may have occurred after a profit of $100000 and then would have again got profitable and moved to a new high of $150000 profit, but still that $20000 is taken into consideration irrespective of what happened before or after.

[b] The system tracks volatility. Consequently, only stocks with high volatility, high beta and standard deviation are selected. The methodology of scrip selection and testing is highly scientific, properly documented and proprietary to the company.

[c] In order to totally understand the behavior of the system and appropriately assess risk - both to the investor as well as the investment - advanced money management and portfolio analysis techniques have been used to analyze the system both individually as well as on a portfolio basis.

MarkeTopper Systems Are Not Available For Public Use Or Purchase.