PowerST is a product of RDB Computing, Inc., a software company founded by Bob Bolotin, a computer programmer with a Masters degree in Computer Science from Northwestern University in Chicago and 12 years of industry programming experience at AT&T Bell Labs prior to the PowerST business.
This Product History discussion starts from the early years. To jump to more recent history of the current PowerST software
This PowerST software began it's life in December, 1988 when a trader friend asked Bob if he could write a program to test a trading system idea. That initial project consumed around a year of part time effort. The approach taken was to write a program from scratch, a very simple software program strictly to crunch the number to test the trading strategy against price data. No user interface or elaborate features. Only crunching the numbers.
During this initial project, the idea of offering trading systems programming services as a business emerged. This was not uncommon in that era that there were programmers offering to program trading strategies. But mostly it was someone becoming an expert in one of the available commerical backtesting software products and offering to program trading strategies into that product.
Bob considered that approach but looking around the problem was that all available backtesting software products had limitations. It didn't seem practical to base a software business on a third party backtesting software product. What if a customer needed something beyond what the third party software offered? Or if a limitation was found in the software that would prevent being able to fulfill the customer's needs?
On the other hand, offering to write software from scratch was asking too much of a potential customer to pay for the development of this kind of complex software from scratch.
So Bob decided that the solution was to write his own software. To develop a backtesting software which could be used as a starting point for customer projects.
To that end in December, 1989 a major overhaul of the initial software written for the project with the trader friend was started. Some advertising of custom programming services was done in the Spring of 1990, leading to a custom programming project that was completed and delivered to this first customer during the summer of 1990.
All of this was being done as a sideline business while still working full time as a software engineer at AT&T Bell Labs. The experience of successfully delivering a trading system software project to a paying customer in the summer of 1990 intensified interest in the business concept as well as work on the software.
Full Time Business and Commercial Trading Systems:
In May, 1991 Bob took a two year leave of absence from his job at AT&T to attempt the independent trading software business. The business incorporated under the name RDB Computing. Inc. The original business plan was to offer custom programming services, similar to the project completed while it was a moonlighting business.
Most of the rest of 1991 was spent enhancing the software in anticipation of resuming advertising of custom programming services. When advertising was restarted in late 1991, the first interest in the software came not from someone who wanted a proprietary system programmed for their own exclusive use (as had been the original business plan). Rather, the first interest was from someone who had purchased a commercial trading systems product and wanted a software version of that commercial trading system. This person liked the custom programming software demo, and was interested in having the system programmed into the software. But he only wanted to purchase a single copy rather than paying for the entire programming effort.
This person approached the trading system vendor and recommended that they strike a deal with RDB Computing, Inc. to program the system into the RDB software so he could purchase a copy. It was an obvious match and a deal was quickly made. The arrangement was that RDB Computing, Inc. would provide the software and in exchange the vendor would advertise the availability of the software in their advertising literature. The vendor essentially acted as a sales agent for the software, but the software was shipped to customers directly by RDB Computing, Inc., and RDB was directly responsible for all end user customer support for the software.
This companion software program was first shipped to customers in early 1992. The business arrangement worked out well, and led to the cloning of the concept, with RDB Computing, Inc. falling into the niche of providing software for commercial trading systems. RDB became well known within the commercial trading system vendor community as a company which could provide software for commercial trading system products for a share of profits with no up front cost. These were mostly small trading system vendors who were traders, not programmers, who were happy to contract out the software part of their business in exchange for a portion of the selling price.
Providing software for commercial trading systems became an active business for an extended period of years. The final count of commercial trading systems which RDB Computing, Inc. supplied software for ended up at twelve systems sold by nine different vendors.
The RDB Computing, Inc. software was reviewed in the following publications during the commercial trading system period. These are all primarily reviews of the trading system being sold by the system vendor, but they all also discuss the software which was being supplied by RDB Computing, Inc..
- Club 3000, 92.07, Page 8.
- Futures, April 1992, Page 70.
- Stocks and Commodities, December, 1992, Page 24.
- Stocks and Commodities, December, 1993, Page 83.
- Barrons, March 10,1997, Page 50.
Meanwhile, RDB Computing, Inc. would also do custom programming of proprietary systems for individual traders and trading businesses as opportunities arose. During the commercial trading systems period, all custom programming work came from the same source, which was a paragraph at the end of a letter included with commercial trading system software disks saying that RDB is available for custom programming. In other words, custom programming was strictly word of mouth for people who purchased the RDB software for a commercial trading system. Almost no advertising was done in this entire period of time because even without public advertising there was as much programming business floating in as could be handled.
Throughout this commercial trading system period, considerable time was set aside for enhancing the software. The longer term goal since the earliest years was to develop the software into the strategy testing software package being advertised in this PowerST web site. The first end user programming environment (what is currently being called the Systems Composer environment) emerged in 1994. The portfolio money management feature consumed a full year of almost full time programming, extending from mid 1995 to mid 1996.
In mid 1995, the PowerST software was still a DOS program (DOS was the original Microsoft operating system before Microsoft Windows emerged). It was a debate between adding the portfolio money management feature or working on the transition from DOS to Windows. The decision was that traders were asking for portfolio money management, and it was felt that the DOS versus Windows issue was less important than the strategy testing abilities of the software. So the portfolio money managment feature was added to the DOS program, and that decision did lead to some solid projects where the customer (trading system vendor and proprietary custom programming) wanted and used the capability.
In the late 1996 period there were multiple factors at play.
Until this time the business had completely been the original business plan of offering to program trading systems for customers (with the modification that much of this business unexpectedly became programming software in cooperation with trading system vendors for sale to customers).
However, the commercial trading systems business was getting saturated as RDB Computing, Inc. was already supplying software for all of the trading system vendors who were interested in the software. Jobs custom programming proprietary trading strategies were sporatic, not steady work. It was clear that the business needed to expand to have a longer term future.
It had already been decided that the future was the desire was to expand the software being used for the custom programming into a product where customers can program their own trading strategies. In other words, the PowerST software being advertised in this website. Over the 8 years of programming trading strategies for a wide variety of customers, the software had grown into a very flexible and powerful backtesting software capable of testing a wide variety of trading strategies. This grew from a lot of successful experience working with a wide variety of customers. The desire was to make this power available as a commercial backtesting software product.
By late 1996, much of the current PowerST feature set was ready to start advertising as the strategy testing product described in this web site. However, Windows had become so popular that early conversations with potential customers led to the conclusion people would not view a DOS program as serious software. On Wednesday December 18, 1996, three different people in separate conversations referred to the PowerST software as a "DOS program". It seemed hopeless to attempt to advertise the current software as a DOS program. Besides, a commercial trading system vendor who was selling a significant number of copies of the RDB software, and was therefore the main source of current income, requested that work on a Windows version of the software be intensified. Further development of the DOS program was dropped in favor of full time effort on a Windows version of the software.
Over eight years of software development, the DOS software had grown into a big software program with a complex user interface. The transition from DOS to Windows ended up consuming 2 1/2 years of almost full time work. It became almost a complete rewrite of the user interface, but much of the core number crunching migrated without change.
The Windows version of the software was first shipped to a paying customer in Spring, 1999. The earliest users of the Windows version were custom programming customers, where RDB Computing, Inc. was still programming the customer's systems into the software and delivering fully programmed software.
The first public announcement of the software now named PowerST as a strategy testing product with the Systems Composer environment that gives users the ability to program trading strategy rules themselves was made in January, 2000. This announcement led to the first sale in March, 2000. The first user is the hardest, and considerable time was spent working closely with this first user.
Subsequent sales to additional customers followed. The early customers were termed early adopters where the deal was that in exchange for being an early user they would get personal support directly from the software developer and considerable influence on product direction. Much time was spend working with each new customer, taking on customers one at a time.
Slowly the software started gaining momentum and some recognition in the marketplace for backtesting software.
Life for the PowerST software changed in late 2004 when the PowerST developer made contact with a startup hedge fund seeking backtesting software. They wanted to start a hedge fund trading a large scale stock trading strategy. They had done some initial testing of the strategy in Excel and were convinced of the value of the strategy, but needed backtesting software to research the strategy exhaustively before launching it as a hedge fund. At the time contact was made with the PowerST developer they had already tried other backtesting products and had failed in their goals repeatedly.
After a series of conversations it was decided that the PowerST software could fulfill their goals if one significant enhancement was made. Specifically the needed enhancement was the ability to scale market positions based upon portfolio level money management considerations. This was a feature that other prospective PowerST customers were asking for, and working with this hedge fund was a way for the PowerST developer to have someone to work with and to be the first customer for the feature. So a deal was made that the PowerST developer would add this capability and in exchange the company would purchase a copy of the software and also to hire the PowerST developer to program the trading strategy into the software.
The company decided to take a chance with making this deal to work with the PowerST developer. The research director said at the time that this was his last try. After so much failure that if it failed with PowerST they would have to give up on the hedge fund. Ultimately the hedge fund was successfully launched on schedule in October, 2005.
This led to a friendly relationship between the hedge fund and the PowerST developer where the hedge fund took an interest in helping the PowerST developer expand the PowerST software business. Before this the direction of the PowerST product was as much towards individuals as businesses. It was this startup hedge fund customer who first suggested that the product should instead be directed towards business level customers. They felt that the capabilities of the software warranted a much higher price and that trading businesses were the prospective customers who could pay a higher price for the capabilities.
That became the future direction of the software. The pricing was raised drastically to $25,000 purchase price plus monthly maintenance fees, and all subsequent customers were trading businesses rather than individuals. This business model was quite successful.
Not large quantities of customers. Rather the business has been a small quantity of elite customers very advanced in the speciality of the software. The business always was and remained a one person software company where the same person did software development, sales, and customer support. Some customers also hired the PowerST developer to program their strategies into the software (and other customers would prefer to do their own strategy programming). Often customers would request new software features as part of the purchase.
What are the capabilities of the software that attracted customers?
Part of it is powerful general backtesting capability as discussed in detail throughout this PowerST website. However, to discuss this from a historical perspective which is the accent of this current discussion. After many years of experience working with a wide variety of customers, as discussed in this Product History above, the software had evolved into a powerful and flexible backtesting engine.
A significant part of the software business had from the start been the PowerST developer programming trading strategies for customers. Capable software was necessary for this because it was only with successful completion of a project that completely fulfilled customer expectation that the developer would be paid. Also, these were not easy trading strategies. For easy strategies people could instead purchase a commercial backtesting software and do the strategy programming themselves. So it was always challenging trading strategies, often doing something different than the capability of other software. Sometimes the trading strategy might require adding new features to the software. Parts of the backtesting engine were written and rewritten multiple times as new requirements emerged.
All of this experience was invaluable in the development of the current software. The experience of working with a wide variety of customers. A wide variety of perspectives from a wide variety of customers. A wide variety of customers giving feedback and helping to shake out new features. After years of this evolution and refinement, by the late 2005 time period the software had evolved into a very powerful and flexible backtesting engine.
Remember that the goal of PowerST is the ability to backtest even challenging trading strategies. Furthermore, not only current trading strategies but to be flexible and extendable for ongoing strategy research into the future. Helpful to this goal is the experience of working with a variety of users who will challenge the software in different ways, and the feedback from that process. The experience of running an ongoing software business supporting customers, programming trading strategies for customers, and supporting customers doing their own programming. It is all valuable experience. As a result the PowerST software has had tremendous ongoing success at being able to handle any trading strategy which customers have thrown at it.
Also, the software has been amazingly stable. New software versions would be produced as needed, mostly for new customers who had new requirements. Some older customers would go years without upgrading because the version they had did what they needed and they never found a bug that needed to be fixed. Users would extensively spot check the backtesting calculations by hand, with a spreadsheet, and as actual trading progressed. Never once did a user find a calculation error. This would be a frequent comment from users. For example one customer said that repeatedly he would think the software was calculating incorrectly and generating an incorrect trade. So he would logic through it and find that actually it was his mistake and the software was accurate. The software was written very carefully, because think about it if you are charging $25,000 for a software product a bug free and accurate product is expected in return.
Not simple to use. The emphasis (and namesake) of the software is on power (backtesting capability) rather then simplicity of use. The software does require learning curve and some programming ability. But in exchange is software with a high level of capability.
Other backtesting software of the era was advertising the ability to backtest trading strategies with a small amount of simple code or even drag and drop of mix and match components. Of course those approaches are ultimately going to be limiting. On the other hand, the PowerST goal was always the exact opposite of power over simplicity. That is a smaller and specialized market of trading strategy developers needing capability not available in lower cost easier to use software. But it was enough of a market to nicely sustain a one person software business.
In addition to the general backtesting capability discussed above, to highlight some specific capability that has attracted business level customers:
- Advanced portfolio level money management testing capability including the feature discussed above for scaling of market positions based upon portfolio level money management considerations.
- Large scale testing. PowerST has the ability to test unlimited amounts of market data in a fixed amount of computer memory, and to do this with good performance. For example, a customer wanted to do a large scale monte carlo type simulation for a portfolio money management strategy. When they first approached the PowerST developer they were working on coding this into a well known name software, and although interested in PowerST were confident they could do what they needed in the other software. So they continued with the well known name software, got it programmed, started running simulations and discovered that it would take 1 1/2 years of constant calculation to accomplish what they sought. So out of desperation they reluctantly paid the high cost for PowerST. After programming the strategy into PowerST they successfully accomplished the testing they needed in a few weeks of computation.
The current RDB Computing, Inc. business is both selling the PowerST software and custom programming of proprietary trading strategies. Experience is that some customers want to do their own programming and there is no way to talk them into custom programming, where other customers want custom programming services and there is no way to talk them into doing their own programming. Both types of customers are equally appreciated.