Progress Update 13
It has been 4 months since the last progress update and I want to highlight what has happened since then since as of writing this I completed a major milestone in making a nice trading algorithm. Since writing the last update, I have:

That is the basics of RSI divergences and obviously, I chose those pictures because they showed a nice divergence before a large trend reversal. But these divergences can and will ping on the way up/down in a strong trend. So you don't want to make trade decisions based off the first divergence you see, otherwise you'll miss out on large moves. You can see this phenomenon in the 4-hour charts of the SPY below. This is also the results of my python algorithm which finds the divergences for me. Blue line is the SPY price and the orange line is the RSI for it. Red lines are bearish divergences and green lines are bullish divergences. Pretty cool huh?
-Jamie
- Completed my 3rd semester of college
- Met Scott Juds (author of the book I was talking about in previous blog posts: Conquering The Seven Faces of Risk) and learned a lot about true sector rotation theory
- Read a lot of books on creating technical trading systems such as Quantitative Trading Strategies by Lars Kestner
- Created a key part that can be applied to any strategy I would want to test: a RSI divergence model all in Python code
The main thing I want to talk about on this progress update is the RSI divergence model that I completed late last night. For those of you that don't know what a RSI divergence is, it is where the RSI (Relative Strength Index) diverges either positively or negatively from price action. RSI divergences usually indicate a potential turning point in the trend but can also keep signaling in strong trends. For example, a positive/bullish divergence is where price makes new local lows while RSI has higher local lows. To show an example of a bullish divergence, here is a picture of Apple stock from March of 2020 where price made new lows while RSI made higher lows:

A negative/bearish RSI divergence is where price makes new local highs while RSI makes lower local highs. To show an example of a bearish divergence, here is a picture of Apple stock from late 2018 where price made new highs while RSI made lower highs:
That is the basics of RSI divergences and obviously, I chose those pictures because they showed a nice divergence before a large trend reversal. But these divergences can and will ping on the way up/down in a strong trend. So you don't want to make trade decisions based off the first divergence you see, otherwise you'll miss out on large moves. You can see this phenomenon in the 4-hour charts of the SPY below. This is also the results of my python algorithm which finds the divergences for me. Blue line is the SPY price and the orange line is the RSI for it. Red lines are bearish divergences and green lines are bullish divergences. Pretty cool huh?
This RSI divergence algorithm will be a part of my main algorithm that I want to try out. Specifically it will be the "setup" part of the algorithm which precedes another indicator which will act as a trigger. This use of setups and triggers within trading strategies is something I learned from reading Phil Erlanger's whitepaper on it seen here.
So that's all I wanted to update on. I will be working on this a lot until my 4th semester starts up in late January so there will be more blog updates to come. I will be busy with making this trading algorithm as well as I am starting an investment fund with my good friend Craig who is really good at fundamental analysis and will be trading all month as well. I'll be busy, but I'll still update my progress on this project of mine!
-Jamie



That's cool that meet Scott Jud. I just stared using his sector surfer algo in January. I'm 60 years old but still have alot to learn. I programed my own python algo that I ran last year, but after 9 months got back to $0. I feel into trap of the backtesting fallacy that Scott mentioned in his book. For now I'm sticking to Scott's algo , but may goback to mine in the future. Good luck on your algo
ReplyDeleteThanks! Glad to hear it! The SectorSurfer models are pretty much fool-proof once you get the hang of it. Make sure your models are including one of each of the economic sectors (consumer discretionary, consumer staples, tech, energy, etc.) except the utilities sector (since utilities sector messes with the model because of its performance during bear markets) to avoid the hindsight bias in your SectorSurfer model.
Delete