Hello All. I am looking to backtest some equity strategies over a very long period of time based on some fundamental and technical data. I am willing to pay someone for their work if they have the ability to do this and also have access to fundamental and price data from S&P compustat database which has quarterly data going back to 1962. As I know this is a very broad description private message me if you have an interest for more information and have the S&P compustat database required. Excelsior
Over a very long period of time you do not need to backtest because the best system is buy and hold by far. Only for short-period, 10 years or less you want to backtest to see if you can outperform.
Yeah I know it is expensive...but I don't want the data for free. I am willing to pay someone who has the data and the knowledge...might help them cover their data nut for the year
Thanks for the reply but I would have to disagree. Disagreement is what makes a great market. Have a good one!
Your strategy might have some holes in it. If your new system has for example a process that uses let's say uses PCA, Neural Networks or similar to analyze say data component relationships or something like that you will find an almost miraculous occurrence. YOu will be amazed how well your system performs in the early 1960s and 1970's when these concepts had yet to be invented and wide scale use of computers was not possible and then how poorly it does in the current era were those previously unknown tools and methods are commonplace and hence any edge of insight they would have provided in the distant past has now evaporated.
Excellent points. Only a tiny portion of what worked before up to 2000s worked now. A review of all trading books that had backtesting results will convince any skeptic. Data-mining has destroyed most of edges. My guess is that oif before the 2000s existed X number of edges, now the number id down to 0.001X. This is one of the most important articles I have ever read.
Sergio, Excellent point. However we need to remember that new edges can be created. Modern data mining, knowledge discovery in data applications can only find things that are defined to the application as data. Someone had the idea in the past to define the concept of a "trend line", the RSI or a triple smoothed moving average. Once those are defined mathematically they can be used for data mining. So the creation of "New Edge" requires new conceptual development on what might influence market dynamics. We can assume that from the point of view of physics or philosophy the causative energetic element in market behaviour is not the existence of the candlestick chart or the OBV indicator. It well might not be what is said this morning on Bloomberg either. New Edge is likely to be found in the application of ideas from strange places like quantum thermodynamics, fluid mechanics and how some species of firefly in the tropics can synchronize their flashing over many miles of riverbank down to the millisecond without the use of Twitter.