Algorithm Results
Above is my August spreadsheet tracking the three algorithms for SPY tomorrow.
The first (B6) is the current algorithm in which I based the August predictions. E6 is the previous algorithm I use for predictions and D6 is the newest algorithm update. All the algorithms are the same core mathematical process except for the type of switch they use. The newest, D6, actually uses the original switch that I updated earlier this year, which as not been applied to the current core algorithm.
To help you understand my chart, let’s start from left to right. Obviously, the first column is the date. The next is the up or down result. A 1 or -1 means the market opened positive or negative and stayed that way for at least the first half hour of trading. A 2 or -2 means the market opened one direction yet crossed over during the first half hour allowing for a profit in either direction. A 2 is automatically considered a correct call based on our scoring system. We had seven such dates counted in our list for August.
The third column is the daily range for the day.
The fourth column is the Grand. I’m still trying to understand the full impact of the Grand. When the algorithms are positive and the grand is negative is something I’m watching. Lower numbers seem to have less impact, however, when it starts hitting the 40’s, 50’s and above, it’s a solid indicator.
Next is the confidence level. This month you can see how a confidence level 1 is often the sign of a flat start to the day and results in a 2 (both sides of the positive/negative line). As the confidence numbers move up the stronger the signal. This continues to correlate, I’ll have an updated post in the coming months.
We did have our first loser after a strong signal . 8/17 was a 13, which is a solid confidence number. However, UK inflation data seemed to spook and change the market mood overnight (my speculation). Another day to note was a confidence level 3, however, the Grand was a solid 79. Although technically correct, it was a flat open starting from positive waiting for J Powell to speak, who then crashed the market.
Last are the algorithms, each column shows the best two out of three core numbers either positive or negative, with the total results at the bottom.
As you can see D6 performed extremely well with 95.65%, only missing one day. The old algorithm was 86.96% correct and our current algorithm had one less correct landing at an 82.61%. So, does this mean I’ll use D6 for predictions in September? No. At least not yet.
I have limited backtest data on D6. During July it did about the same as the other two algorithms. I have more data on B6 and E6, and over the long term with over 125 trades B6 is 99 correct out of 125 and E6 is 97 out of 125. Both are rather consistent through positive and negative switches.
The good news is, as a member, you can use any of the algorithms displayed and not depend on my interpretation of the data. If D6 continues to do well, you can use it as your key signal. Especially on the days the three don’t agree.
If you have any thoughts, let me hear about them in SPY chat.
