Trade Using Market Regimes?
Interest in trading multiple strategies via predictive market regimes, henceforth referred to as regime trading, is exploding since the teaser Quantivity post on a Market Regime Dashboard.
A quick poll of readers to help tailor future Quantivity posts: are you interested in regime trading, including any of the following: analysis, algorithms, full systems, and/or daily trading lists—along the lines introduced in the following posts and papers? If so, please comment or reply privately (all replies will be kept strictly confidential).
Interest in regime trading is building across the blogosphere (echoing Quantivity readers’ comments), such as:
- Creating the Adaptive Time Machine by CSS Analytics
- Multi-Strategy Trading with Regimes by Hack the Market
- On the Drawing Board (Revisited), including Combining Strategies and Timing the Strategy ideas, by MarketSci
Along with a bevy of fairly pedestrian recent academic work, such as:
- Multi-Strategy Trading Utilizing Market Regimes by Mlnarık, Ramamoorthy, and Savani (including slides and video from Advances in Machine Learning for Computational Finance Workshop, London, 20-21 July 2009)
- Dynamic Regimes of a Multi-agent Stock Market Model by Yu and Li
- Market-Neutral Portfolio of Trading Strategies as Universal Indicator of Market Micro-Regimes: From Rare-Event Forecasting to Single-Example Learning of Emerging Patterns by Gavrishchaka and Bykov
Given the notable algorithmic and quantitative challenges of regime trading, this is a topic tailored made for Quantivity. Speak out now if you are interested.
Count 1 for interested. This seems to be a popular topic in the blogosphere these days.
Very much interested.
It is the future of quant trading.
Bob
definitely
count me in…
Yes, interested! But more in principles and techniques than ‘daily trading lists’ — learning to fish…
Very interested! But whatever happened to the promised posts regarding the Market Regime Dashboard? It’s been over a month now since the original post surfaced.
I’m definitely not trying to be critical — I appreciate any and all posts on this subject, if and whenever they come. I guess I’m just impatient, but I’ll try to control that. 🙂
@CarlosR: this is indeed one-in-the-same to the original Regime Dashboard post. Delay has been due to several factors, including: (a) parsing significantly larger-than-expected reader interest and calibrating content to audience; (b) triangulating how this expanded topic overlaps with numerous private collaborations on similar topics; (c) a life event. Thanks for your enthusiastic comment and support.
Very much interested.
John
Thanks for the explanation, sounds very reasonable. I didn’t realize this was the same as the Regime Dashboard, so that’s good news.
Definitely
I’ll check back often to see what you find to be worthy of the effort.
jc
I have been checking in periodically, and am concerned that the blog has met an untimely end. Any updates for your past readers? Thank you.
Back in the saddle, after a bit of hiatus mandated by the day job.
I had so much interest and hope for this blog. Where did you go?
@John: bit of hiatus mandated by the day job. Looking forward to resuming regular posting.
can’t wait to see you get started on this!
My trading has become increasingly mechanical, and I’ve been thinking about the whole subject of non-correlated trading for quite some time. It seems to me that since the meltdown, it’s much harder to really achieve non-correlated positions by choosing among different trading instruments. Instead, the key is with non-correlated trading strategies. My hope is to develop a market regime sensing algorithm and use it in conjunction with counter-trend strategies and trend following strategies to have a portfolio of positions at any given moment that are as non-correlated as possible. Each strategy in and of itself should have a consistent edge, but combine all of them together and the profits to drawdowns ratio should be exemplary.
@Michael: thanks for your comment; indeed, we are thinking along very similar lines. I am drafting several posts on this topic now, to which I hope you will similarly comment when they are published.
I am doing more and more reading on complexity and thinking of ways to apply to my trading systems. I have become a big believer in taking a base signal generation strategy (can be a simple RSI2 type strategy to AI neural based). The signals are then filtered by regime (I a simple two state model). Based on the regime state, I’ll only take the long or short signal. I then separate the exit algorithm and use the opposite signals not being used to indicate exits, also using an RSI2 or time limit. All of a sudden these rules and systems create a fairly complex system which trades mean reversion in the direction of the trend very effectively. In other words when the market is trending up it only buys dips and when trending down it sells rallies. Since the exit is separated from the entry signal, you can get out quicker or slower depending on current market conditions.
@quant.this: thanks for your comments. Very interesting multi-layer strategy. Would be interesting to chat with you further privately, or link to a blog if you have posted further details. Of particular interest is your regime modeling, which sound to be via Markov switching over state space?
I’m also investigating this strategy for Forex trading (RSI(2) but only in the direction of the trend). It seems to work even with a simple good averaging function.