Selling Strategy (2-10-2010)

Last month's losses served to stress the importance of establishing a better basis for profit/loss points. Stretching my profits points worked out very well the first part of the month, but the lack of a good stop loss ended up costing the account a lot of money.

And now, a week and a half into February, and a couple of mistakes later, the account value is down another ~5k.

Two factors - the stop loss and, perhaps an even bigger factor - the psychological: buying outside of the tested strategy, getting carried away with big profits, and the lack of a proper perspective for the trade profit/loss calculation interval - combined for some very poor decisions.

I have been analyzing the data that I have been collecting on the stocks that have passed screener criteria. Unfortunately, almost all of the metrics that I have saved, - or thought of calculating - have pretty much washed out in terms of predictive value. I haven't been able to establish any type of relationship between collected/calculated on the day of purchase to future price - everything resulted in uniform scatter plots with an occasional outlier.

I decided to take the stochastic approach and examine performance of the stocks that passed screener criteria by way of the difference between a day's high and the close of the stock on day of purchase. Day 1 of ownership was the most consistent. The figure below plots the difference between the high of the first day of ownership and the close on day of purchase, with the day of purchase (Day 0) on the x-axis:

Pretty cool huh? Some crazy variance on the DOW's down days, but it is pretty easy to see that the majority of stocks have a high that is higher than the close on the day of purchase.  Here is a histogram of the day 1 data:

Using some statistical analysis, I fit the total Day 1 data to a distribution and calculated values at 50% probability and 75% probability and came up with gain values of approximately $0.20 and $0.04 respectively. Filtering the Day 1 gain for positive only yielded approximately $0.20 and $0.09 respectively. For those who may be interested, the unfiltered data was fit to a Burr (4p) and the filtered data was fit to a Pareto distribution, with respectable goodness of fit tests:

 

I next filtered the Day 1 gain data for negative values only, and checked the corresponding difference between the high gain on Day 2 versus the close on Day 1. The value for a 75% probability was approximately $0.01, indicating that for the most part - it is best to sell the losers on the first day as well.

All of this lends itself to the formation of a new strategy - set a profit point at about $0.20 above Day 0's close, and close all positions at end of Day 1.

Easier said then done. Take Monday and Tuesday of this week for example. Monday was a losing day. I did not sell at end of day, deciding to wait until the morning of Day 2 (Tuesday) (I did however enter two new positions). On Tuesday the market saw a lot of bullish action and I postponed the sale,and ended up reaping a decent return for my trouble - on all but one stock (which I sold today at an even greater loss). The psychological factor - at one point during the day I was up just under $300 on that stock, but still down overall. Why is it so difficult to keep the day's gains in perspective??

I want to be mechanical about my trading - I think that is best probably. But - it seems there has to be room for interpretation - everything was down on Day 1, so itI thought that it was probably was best to wait until Day 2's morning to sell. Futures indicated a strong opening and I held on, which happened to be the correct thing to do. But - what if they wouldn't have - e.g., week three last January?

After the steep loss on January 22, a lot of people were trying to guess at what the market was going to do the next day. I don't think it was at all possible to have any confidence in anyone's opinion on what was going to happen. What if I find myself in that position again? I think if there is any doubt I need to sell. Chances are that I will be wrong sometimes. Cost of doing business.

So that is my stop loss - hold on to the position for a day. If everything is down, wait till the next morning, then try to gauge market sentiment; if the futures market is mixed, sell right at open.

And the last thought - I have been tuning into Don Miller's blog which I have found offers a lot of practical advice. He talks about the idea of outliers and how those will be the big reward days. My implementation of this will be the lowering of the sell point during normal trading hours, but prepped for the possible gap open by bumping up the sell during extended hours. Slow and steady with the occasional home run. That's me.

Lets see what happens.