Over the course of this week I have reviewed my initial analysis of the high probability trade system and have revised how the system stats are derived somewhat based on what I feel is the most likely price action scenario. This decreased the systems probability of success, but I feel it is more realistic. I also went through and optimized each stock in the system and then re-ran the study.

The reworked stats for my system are “correct” except for during the development, which target (sell or buy) was hit first was not a factor. It is simply too time consuming to look at this for each individual stock in the system. As it is, it takes about 30 minutes to optimize one. When I re-ran the study, I was using a basket of 9 stocks which I selected from the 19 I optimized. Now that I know what I am looking for with this, I have gone back through and have an additional ~80 potentials which I will probably narrow down to about 30-40 total after optimizing.

The following is the results for the revised study:

The base stats of the stocks in the system are:

- Average % win = 70.48% (stocks range between 67.31% - 74.19%)
- Average win = 0.81% (stocks range between 0.69% - 1.00%)
- Average % loss = 29.52% (stocks range between 25.81% - 32.69%)
- Average stop loss = 0.96% (stocks range between 0.72% - 1.26%)
- Average maximum % stop out = 12.39% (stocks range between 3.85% - 17.46%)

I have looked at this in three ways, one being using purely the averages, one being using worst case scenario stats (i.e. using the lowest % win, the lowest win, the highest % loss, and the highest loss), and what I (being delusional) believe to be the most likely outcome.

For the average and worst case studies, I considered all losses to be at the stop out % and all wins to be at the win %. The stats I am using for my delusional study which I am using as the target metric are a 65% win rate with a 0.9% gain, and a 35% loss with a 0.56% decrease. The 0.56% decrease was derived from assuming 15% of the time I get stopped out at a stop loss of 0.9%, and all other losses average to a 0.5% decline.

The following table summarizes all results for the outcome probability of just below 50% (i.e. the actual outcome will most likely be that shown in the table or better). The returns shown are assuming compounding.

My basic plan for how to manage this strategy is as follows:

Monthly:

- Search for new potential stocks and optimize them
- Re-optimize stocks that are below the trade basket threshold but still have potential

Weekly:

- Verify the statistics of the stocks in the trade basket have stayed at an acceptable level. Drop them out of the trade basket if they are no longer at the required levels.

Daily:

- Look at the stocks in the trade basket and rank them by the most probable ones to have an up day based on end of day charts.
- Select the top 10 and verify there are no obvious issues with trading them the next day. Basically check it isn’t an earnings report day and verify there are no recent bad news items with them.

Stocks will be put in my trading basket if 3 of the 4 criterion are met (or if 2 of the 4 are met with a third that is not too far off):

- Probability of success > 65%
- Probability of stop out < 15%
- Target gain > 0.7%
- Target loss < 1.00%

In total this means I will have to dedicate about 45 minutes daily, 5 hours weekly, and 15 hours monthly to maintain this strategy. That is approximately 40 hours a month…getting in the realm of a part time job. If this strategy proves it has merit I will look at automating the optimization process for each stock since that is really the only major time sink with this technique.

I believe I have convinced myself that this strategy has a positive expectancy and an edge, but really the only proof of this is actual results. So, I present the actual results thus far of this strategy. I will be tracking this strategy using my actual wins and losses with commission included and assuming pure compounding each time. I have included my metric and worst case results in the graph as well since I expect my actual results to fall somewhere in the middle. Also I know that if my actual results dip below the worst case I need to re-evaluate. Once I have more data I will revise my metric to better fit the actual results and will use that to determine if something is wrong in the system.

Some stats on the actual trades:

- # wins = 4, average win % = 0.725%
- # losses = 3, average loss % = 0.537%
- Probability of outcome based on metric stats = 53.23%

The take away of this so far is that really there is no conclusive proof of success or failure of this strategy yet. More data is needed to validate or refute it.