Speed is essential to backtesting platforms. This is why when I was beginning to think about the new version of TradersStudio, I knew that I needed to create a version that was faster. Real traders and systems developers know that one of the longest parts of developing a new system is the time that it takes to test your hypotheses. Systems development is a little like laboratory work. We know certain basic premises about the markets and we often have a rough idea of the system we want to develop. However, we also know that development of particularly robust systems takes a lot of time. I recently used the new TradersStudio to develop a system in a fraction of the time that it would have taken me with the old version.
With the new multi-core version of the software, we can now develop systems over larger parameter ranges and markets. At the same time, we can also finish the development cycle much more quickly because we’re sitting around less waiting for the system to run and spending more time actually researching and delving into the results.
I enjoyed developing a new system recently in what almost seemed like record time. As you may be aware, I have a bond system for sale called MAR_Bond that was originally released in 2010. While this is a very robust system, it had a natural upward bias assumption in treasury bonds. Given the current economic climate, I now no longer believe this to be true. The problem with my original system was not the core intermarket logic, but rather the fact that we greatly limited short trades from happening and made it easier to get into long positions. To counter this, I wanted to redesign this and create a symmetrical system that is able to adapt to changes in the bond market. Using the multi-core version of TradersStudio, I was actually able to finish my development in less than two days including my intensive robustness analyses. This would have taken me almost 20 days on my six core machine if I only had the old version of this software.
After developing the system it suddenly occurred to me to think of this in terms of money. You probably know the old adage that “time is money”. Consider this: what is your time worth? I estimated that it took me a total of 16 hours to create this system on the new version, but I suspect it would have been about 150 hours using the single-core version. If you estimate your time to be worth $100 per hour, then you would have spent $15,000 (150 * $100) worth of time with the old version. With the new version, your time expense was only $1,600 (16 * $100). Even if you value your time at less, say $20 per hour, you still would have spent $3,000 worth of time compared to $320 worth using the new version. In this business, time truly is money. The more time we spend waiting for results, the less productive we can be at finding incredibly interesting trading strategies.
When developing trading systems we want to first develop a premise, and then develop a number of experiments over different ranges of parameters to test our original hypothesis’ correctness and also its robustness. In addition, markets sometimes change over time and therefore we need to optimize over a large range of values to see if the parameters are stable or if they are “drifting” over time. Sometimes we observe changes in our premise and running a large set of parmameters can confirm our observations.
The recent change in the bond market is a good example of how tweaks in the broad “climate” can have a significant impact on a system. To that extent, this system was redesigned to tweak the filters – in fact, the same core intermarkets are used in both versions. The results in 2013, though, are radically different. The orginal system actually lost a little over $13K, while my new system is up a little over $13K. Short trades that were filtered out in the old system are taken in this one. These changes help over a longer period as well. The original system made a little over $187k since 9/22/1987 (quite a decent return!), while the new system produced over $296k during the same period (an even better return!). The drawdown of the new system is higher at $20,125 versus $17,550 for the old system. However, I think the upside potential definitely warrants the higher drawdown. Finally, I’d just like to note that this new system is totally symmetrical on the long and short side. For example if we use a .5 parameter on the long side we would use -.5 on the short side.
After designing this system I felt this was a good example of why the new TradersStudio will be so powerful for so many people. You can develop systems intuitively using the interface you know and love, but you can also test advanced strategies in blazing-fast time. I encourage you to take a look at the new version and try your wildest ideas out – who knows, you might wind up discovering a system you never thought possible before!
Additionally, if you have interest in my bond systems, I will include both versions in my packages available on my web site.