Saturday, December 29, 2007

Expanding One's Trading Framework: Building on Strengths

I've written before on my basic framework for trading. I calculate various price targets, including the prior day's high, low, and average points as well as the R1 and S1 pivot points, and then handicap the odds of hitting those levels as the day unfolds. My major tools for handicapping those odds are relative volume (how volume compares to a 20-day average), the adjusted NYSE TICK (how TICK compares to its 20-day average), and Market Delta (volume transacted at market bid vs. offer).

This framework has proven sufficiently successful that I am experimenting with rolling it out to longer time frames. What this means is that, instead of using the prior day as the basis for calculation of the price targets, I will calculate the targets associated with any N-day lookback period. A simple example would be to calculate the price targets based on weekly bars rather than daily ones and use the indicators (as well as other ones, such as advances/declines for common stocks only) to handicap the odds of hitting those targets over the next five trading sessions.

Time is completely malleable in this framework. You can create bars and targets based on 45-minute periods, 3-day periods, or monthly periods--and everything in-between. The swing trades boil down to making bets on which of the price targets will be hit first based upon unfolding price action and volatility. By trading periods of very different lengths, it is possible to diversify bets even as you might be trading highly correlated instruments.

When developing new trading frameworks, it is important that these not only build on existing strengths, but also match your own needs and interests in the markets. My personal trading takes place entirely during AM hours, as I wish to free my days for writing, work with traders, and family life. I know from experience that sitting in front of a screen all day every day leaves many of my needs and interests unfulfilled.

By expanding the trading framework to an N-day level, decision-making and placing of orders can take place during the morning hours, even as the trades may last several days. This leverages a trading method without requiring additional screen time. It also keeps the number of trades I place down to a manageable number, minimizing commission overhead and allowing trades to be well-planned and thought through.

That, in a nutshell, is my trading project for 2008.

6 comments:

ainkurn said...

will you be using this N period strategy only on ES futures or will you also expand the project to other futures contracts and possibly even stocks. Also, have you ever used your current strategy for ES on SPY? I would be interested to know how your win/loss ration would change when trading two different instruments that are both based off the underlying index.

Brett Steenbarger, Ph.D. said...

Hi Ainkurn,

Great questions; you're anticipating my next steps. I'm actually focusing on ETFs for this strategy and will start with SPY and QQQQ and then roll out to sector ETFs, international ETFs, etc. No reason individual equities couldn't fit this framework, though that's not my primary interest.

Brett

Adam said...

Brett ~

One more attention-grabbing post!

A bit on a framework I deployed for individual equities, sector ETFs and index ETFs earlier this year (keep in mind that I am not an intra-day trader):

Making use of price EMAs and SAs of different lengths (look-backs), the model measures inter-day change in the slopes of these curves, watching for consistent acceleration (growing steeper) or slowing (growing flatter). The same method and time frames are simultaneously used to follow net-up-down volume.

Since what one cares about is not absolute change but velocity of change, the metric derived is the degree by which one SD increases over time. A graph of this would be a single horizontal line (SD) with single vertical lines of greater of lesser length projecting above or below that axis.

Because of this format, the price and NUD graphs can be laid on each other. One could “eyeball” correlation; I do the math.

As looking at screens and charts all day tends to make one break rules, the model output is: +1, 0, -1.

The model can be used at different degrees of resolution (time) in different sectors. In biotech, a volatile, event-based sector for which I have reliable fundamental information available, the look-back is half that for energy.

The method has proven to be moderately robust (flexible) in giving entry/short points. Sensitivity of those is easily adjusted. The model is weak on exits. It worked nicely until it stopped working abruptly in mid-November.

Further refinement is indicated. Your post offers a few obvious hints.

Adam.

Brett Steenbarger, Ph.D. said...

Hi Adam,

Thanks for the very insightful comment and for sharing the unique methodology. Reminds me just a bit of some work I've done with an indicator I called the Power Measure, which quantifies "trendiness" as a function of the correlation between directional movement and volume (or volatility). The idea is to enter trades in the direction of increased trending. I also found that I needed other tools to aid with exit timing--

Brett

Andre said...

HI, Im the writer for the new blog Trading Addiction (http://www.tradingaddiction.blogspot.com) and I really like the quality of content that your blog produces and was wondering if you would like to exchange links with me?

Brett Steenbarger, Ph.D. said...

Hi Andre,

Thanks for the interest. I don't do link exchanges as a matter of principle, but if you have some posts that you feel would be of special interest to TraderFeed readers, by all means feel free to email the URLs to me for inclusion in one of my link posts.

Brett