Motivation:
A trading system should have as few tuning parameters as possible: simple is better than complex. Simple systems work best when suited to the current price regime. If we can identify N price regimes, we can build N simple trading systems. If current price behavior can be seen as a synthesis of the different price regimes, then our strategy should be the same synthesis of the different trading systems.
I am trying to be as general as possible, above. This could be as simple as the following Regime / System:
- Trending / Enter at price extremes in the direction of the trend only
- Cycling / Reverse position at both price extremes
- Noisy / Hold cash.
So, qualitatively, if the market is 10% trending, 40% cycling and 50% noisy our strategy might be to commit larger positions (50%) in the direction of the trend, smaller against the trend (40%), remainder in cash.
Thesis behind this project:
Price series reflect all the information that matters. Furthermore, information includes the inefficiencies acting in the market. Therefore, price series can inform trading strategy IF we can extract the information.
At some point in the future, I may discover that additional information is contained in volume, open interest and option price data. For now, KISS.
Objectives:
- Explore universe of statistics used to characterize price series.
- Categorize price series behavior into actionable regimes.
- Investigate predictive connection between statistics and regimes.
- Investigate the persistence of different regimes.
- Compile minimal list of regime-defining statistics.
- Determine if asset classes have tendencies towards particular regimes.
Share
If you found this post informative, please share it!