Sunday, May 9, 2010

New Home!

All future models, strategies, and market analysis can now be found on JeremyBain.com

[site overview]

This site is built from a collection of quantitative models, fundamental models, market studies, and trading strategies all designed to gain an edge in trading options and equities. The models, studies, and strategies found within this site are not based on discretionary price patterns or other subjective measure of technical analysis. Rather, this site is designed around observations that can be reduced to numbers, rigorously tested, and measured.

Further, this site recognizes the advantages that come with combining the analyses of both quantitative and fundamental analysis. Quantitatively, I've designed models that identify momentum and relative strength/weakness, as well as methods that identify when securites and/or sectors have become extended to the point of offering (un)favorable risk/reward ratios. Fundamentally, I've designed models based on earnings, cash flow, ROE, and momentum.

Both techniques currently use a top-down approach to valuation. Broad market indices receive the first level of quantitative/fundamental analysis to identify long and intermediate-term trends. Then market sectors are analysed and ranked according to any number of criteria, including price, earnings, momentum, relative strength, etc. These sectors can then be further drilled into, to identify specific securites that satisfy personal risk profiles. Future models will identify value through a bottoms-up approach.

Tuesday, April 28, 2009

RSI Divergence

DJI - RSI Divergence

Monday, April 27, 2009

Shooting Star - SPY Tested

Several sources today were citing the bearish implications that a 'shooting star' may have.

A shooting star is defined as: A single day pattern that can appear in an uptrend. It opens higher, trades much higher, then closes near its open. It looks just like the Inverted Hammer except that it is bearish.

Of course, we opened lower but we did trade much higher from where we opened. And it did close near its open. While this doesn't fit the description of a shooting star, the pattern still looks similar and I wanted to test its significance. It seems like it would suggest a rejection of higher prices which should lead to lower prices. So let's test that.

The criteria I used for my test was as follows:
1. The close has to be in the bottom 25% of the day's range.
2. The day's range has to be at least 4 x greater than the difference between the open and close. This results in a long 'shadow' and a small 'body'.
3. It sells tomorrow's close (1 day was used but 2, 3, 4, or 5 days didn't help)

Tested over a 5, 10, or 15 year period, this criteria provided almost NO edge whatsoever. It had a net profit close to $0 and won on average 50% of the time.

So all this shows is that with this criteria isolated is it virtually useless. If you'd like to incorporate this price pattern into your trading, you would obviously need to use this as a supplement to your entry criteria.

As a simple example, I used the criteria above but added the filter that the close had to be above its 10 day MA and under its 200 day MA. I just randomly picked these numbers but it does in fact describe the market's (SPY) current situation.

Under this new scenario it is actually profitable to buy the close. Granted it has only occurred 11 times since 1993 so I wouldn't consider this 'properly' backtested but it's just an example. Anyhoo, in this example, buying this scenario was profitable 91% of the time. See below - click to enlarge.
Results

Equity Curve

Sunday, April 26, 2009

Simple System

I stumbled across a very simple system today. It buys when SPY closes in the bottom 20% of the days range. It buys the close (day 1) and sells the open on day 3. It's essentially a one day trade but it sells on day 3's open opposed to day 2's close.

It is commonly known that the sell strength, buy weakness, mean reverting strategies are working right now. It didn't used to be like this though and at some point in the future it probably won't be either but until then that is where your focus should be. A good example of this is, go read some of the older trading books by Williams, Connors, the Turtles, etc. Many of their strategies were based on buying breakouts (buying strength). If you test those systems in today's market (I have) you will not achieve the same satisfactory results.

I don't recommend using this alone as a trading system but the concepts presented should be considered when building your own.
Results on SPY (5 yrs) - click to enlarge:



Here is a screenshot of what happens when 20% is replaced with x% (column A)



The results are surprisingly decent even when tested over a 20 year period. This certainly does not work with all stocks, so backtest accordingly.

TradeStation code:
INPUTS: len(1), pcent(.2);
IF Close < (Low + Range*pcent) THEN buy this bar at close; If BarsSinceEntry=len THEN sell next bar at open;