The development of my SPY market pressure calculators
My name is Rosh, and like you, I have a story of struggle and success. Creating an algorithm which consistently correlates to then next days marketing open is about as easy as discovering the meaning of life (I’m working on both). I’ve always enjoyed sharing what I learn and helping people improve their lives. This project is no different.
This is a randomly, casual updated history of SPYtomorrow.com development.
Here we go…
In the early 2000’s I taught myself how to buy and sell stocks and options with a small online account. At the time, I learned a lot about the stock and options markets, yet had little success due to the size of my positions and the tax of broker commissions.
Like many in 2008 and 2009, I discontinued my trading as my financial world came tumbling down. Many faced a great recession, and in Detroit, we faced a Depression. To pull my family out of ruin, I refocused my energy on my creative entrepreneurial media career.
Over the next few years, I clawed my way back, yet, I lost everything again in 2013/14. I was forced out of a business I helped build, my teaching positions ended, and I went through a divorce.
Eventually, I married again and rebuilt my entrepreneurial career from the ashes. I vowed never to let outside forces decide my future again.
At the beginning of the new year of 2020, me, my wife, and family are comfortable in our lives and careers. Upon reviewing the past, I realize, I’ve not seriously started to save again for the future. I’m much older now, and if I want to retire, I need to save more aggressively.
I dusted off my 20-year-old account, which, believe it or not, was still active. My plan was to use my stock-picking skills to invest in our future. I made my first trade and noticed something amazing. No commission!
It turns out that during the previous year with the introduction of free trading platforms such as Robinhood (not a recommendation), larger brokerages began to follow suit with no/low commission trading.
The Financial Experiment
So, as an experiment, I started aggressively testing some of my old trading ideas from twenty years ago. With the volatility of early 2020, I made 50% on my new small Roth account in my first month.
However, I had questions.
Some trades didn’t go as I expected, and I wanted to figure out why. I had some theories, so I began randomly using data I found online to create calculations to help me understand why certain things were happening, especially in the options market.
Over time, through testing and observation, I discovered the first market indicators that I trusted. These two indicators work together, one I referred to as the call mood (for a positive market pressure) and the other is put mood (for a negative marketpressure).
I originally implemented QQQ data into my calculators. I also tried using the calculators with other ETF’s with success. However, I realized each ETF was a little different, and the calculators needed to be collaborated. So, I settled on SPY (S&P 500) as my main indicator.
Eventually, through additional experimentation, I noticed another single indicator that is rather accurate. I began using it as a secondary or support market indicator. I referred to it as the Pressure 3 indicator.
After a couple of months, I decided to try to double $1000 in a month using my SPY calculator. However, that is not what happened. I tripled the money, turning $1000 into $3000 in about 3 weeks using directional options trading.
Not long after, I noticed my trades were not going very well. As market volatility began to calm down in the summer of 2020, my readings were more often mixed. I would make educated guesses, however, a flat market is not helpful for some types of options trading.
My calculator, averaging over 80% correct, was now in the low seventies (still not horrible). Still, even when the calculator was right, I could lose due to a lack of volatility and stopping out during market reversals. This is true anytime you trade the stock/options market, yet it was more common during this time frame. I lost a couple of thousand dollars of my previous profit.
After reviewing my summer trades, I realized that when my calculator showed mixed-signal, it was a 50/50 shot as to the next day market direction or profitable trades. Often the market starts off flat, which is valuable information, but not helpful for the style of trading I like.
This doesn’t mean you can’t trade on mixed days, it all depends on the type of stock or options trading you are doing.
Once I disciplined myself not trade mixed-signal days, and the volatility began to increase in the market the profits began to roll back into my account.
During the end of my summer vacation 2020, I decided to backtest and try to find an even better indicators. I did find some promising indicators as a result of all my testing. However, most were not as solid as the ones I have already. Although, one did stick out, and I started to incorporate it into my trading. This is the G 100 indicator (no longer used).
By the end of summer, when all indicators are pointing in the same direction, they average 80%-90% accuracy.
With confidence, I launched spytomorrow.com
Yet, by the end of Sept 2020 and the beginning of October 2020 many of these signals started showing doom and gloom. Black and Red all the time. Although I did OK in the beginning, low volume up days started to take hold, and the indicators were not working well. A lot of opportunities were lost over a three week period. I actually went into the negative.
Back to the drawing board; I started to dig deep. I reviewed, worked on new formulas, and tested old ideas. Eventually, I found a result that met my three criteria. Stabile numbers that didn’t fluctuate dramatically from moment to moment, don’t require interpretation and don’t need calibration.
After a few semi-successes, wins, and failures tough October, I found a better solution. This solution expanded to what I refer to as the SPY Waves. SPY 2 Wave (S2W), SPY 4 Wave (S4W), and SPY 6 Wave (S6W). Now referred to as N Waves.
Volatility continued to lower, and some of the Waves broke down. However, the N2 Wave and The VV Wave held 80% accuracy even during the lowest volatility of my testing. Even better, when I combine parts of the calculations N2VV, accuracy measures about 90% (Read more). Yet, the N2VV eventually pulled back to 80% accuracy like the others. However, when I combined all three, I got the indicator I was looking for.
As of Jan 4, 2021 I started using the 3 Wave indicator for my front page predictions. Additional insight and information from all three Waves were available to members when the membership section opened up again.
By mid-January 2021 I found a better data set that improved the accuracy of the Wave calculations. Also, as of January 14th, 2021, and moving forward, I began to share the result (chart) of the predictions on the front page next to the pressure prediction for a more complete story.
I’ve declared my calculations to be final a few times. Yet, something would break, mostly what I referred to as switches became more common, which means unpredictability. This was not good for a prediction tool.
By late February 2021 I found stronger approaches to capturing and reading the data and I declared my final version of spytomorrow. I knew there was no further I could take formulas and strategy.
Well, March 2021 was a rough month with lots of market-changing news. This, of course, sparked my need for one more shot, the last idea.
It’s true I couldn’t take my formula any further (so I thought), however, I could take all that I learned and apply it directly to the original formula that worked so well in the beginning. I was able to create new waves which were real-world and back-tested with solid results. These waves meet my criteria of no calibration, switches, interpretation etc.
The switch. You may have seen a mention of switches in my writings or social media. This is the down side of the N waves. Randomly they will switch directions. I can catch it, but I don’t want a signal that can randomly switch directions. Is it a cycle or related to volatility? I didn’t know.
Yes, I’d been here before, great results until they weren’t. The way I set it up allows me to add better waves to make the pressure calculation stronger without disrupting the entire process.
I officially launched the C Waves (CBN) March 29th.
As of March 2021, I spent a year working on these formulas. I launched Spytomorrow.com membership in September 2020 to share with the world and then all hell broke loose with my calculations. However, every step I saw the possibilities. However, there is a point at which it’s time to let go. April 2021 was the final exam. If, the final C waves don’t do the job, I would need to let this project go for new opportunities.
Fortunately, April was very positive for spy tomorrow and I reopened memberships (officially) beginning May 1, 2021.
Yet, soon after launch, I shut down the project and refunded members’ subscription fees. Something was still not right.
I spent the summer of 2021 refining and figuring out the switches which were giving me so much trouble. I did find the source using a more macro view of the market data.
This as allowed me to take on the pressure prediction calculator(s) with fresh eyes and new approaches. The results have been very good, plus, I’ve improved my trading skills which made for much more profit on the correct days.
Now that I have a new global data approach, I can reimplement some of the previous highly successful versions of the SPY calculators, because I now know why they eventually failed.
Testing for the third SPY open
Beginning Nov 1, 2021 I ran the 3rd beta for SPY tomorrow through the end of the year using the new calculations with switches implemented. After three months, the everyday calculator is 80%+ and the strongest signals run about 90% correct, although, they don’t offer a complete trading signal every trading day.
Again, I began to post the strongest trading signals on the website’s front page in November. If there is no solid signal, a neutral icon is displayed. When spy tomorrow opened membership again, in the new year, future members could see other signals which can offer direction for the next day.
My goal was to complete 100 trades with successful data and live trades rather than 30-40 as I’d done in the past, before I reopen membership.
Unfortunately, I followed a few shiny objects that lead nowhere which made for a bad December. Once I realized my error, I finalized the calculation approach and delayed the soft-launched site for membership to Feb 2022. The soft launch of the 3rd beta went well and actually performed.
WAR and Inflation
The launch of the new membership option was met with war and growing inflation. Although we had more wrong days than usual, SPY wasn’t breaking. It was more about the changing market mood over night with the news. Fast-paced new cycles are not good for overnight trading systems.
SPY is working the way it should. This means I could focus on improvement rather than fixing a broken theory or system, which is what I’ve been doing over the last two years. The solution to the switch that I began working on in August 2021 was still strong.
During the first quarter of 2022, I was able to improve the speed of my calculations. It would take me 12-15 minutes to cut and paste data. Now, with coding upgrades, I can paste the data, and the calculations are complete in about a minute. This means I have more time to offer more insights. So, beginning mid-February, I began offering QQQ and IWM pressure reading to the members. This is helpful when considering the market’s overall mood, the diversity of ETF’s can helped strengthen predictions.
By April 2022, with even more data, I calculated more pressure combinations. I expanded my triangulation approach, expanded the numbers from 3 columns to 9 for members, in case they want to look for their own patterns. Of course, this can look overwhelming with 3 rows of 12 columns green and red numbers. However, I felt those who wanted to dig in deep, had the opportunity.
Next, to improve readability, I color-coded algorithm sections Purple, Orange and Blue (POB). Blue is the August 2022 combination that performed the best. Then I added the large size prediction numbers to make the system easier to read and understand. I also reimplemented green, yellow, neutral, orange and red arrows for my predictions.
I continued to experiment with the number of results and triangulations I developed from multiple calculations. Generally, there wasn’t much of a correlation between totals and the direction of the market. The more important consideration is positive or negative. So, I started adding the totals of green and red from the 12 columns. I would get results from -27 to 27. I counted how may 1 +/-, 3 +/- , 5 -/+ etc. Adding the plus or minus of a number together as one total.
What I found is 1’s were less accurate, and the higher the number the more accurate the prediction result. Here is a blog post about it. This is extremely helpful to understand why at times SPY does better or worse. I now have confidence level indicators.
This is helpful for position sizing. If you see a 1, you might buy a few options, if you see a 10, you can buy more because the odds are in your favor. Of course, this doesn’t solve overnight news and market mood changes, but it greatly cleans things up.
Even better Triangulation
By mid-April, I improved the visual presentation of the pressure numbers and began using the results of IWM and QQQ to help determine the SPY trade. For example, if SPY was -1, QQQ +3 and IWM + 10. I can determine (cautiously), that SPY will go up. It worked well, and is improving the overall correctness by not taking the lower numbered SPY results at face value.
By June of 2022, I switched the POB from the SIQ (SPY, IWM, QQQ) for the confidence number reading. It also has a solid correlation. Higher the number greater the confidence level. Over the first two months numbers, 9 and above are 100% correct. I’m sure this will stay above 90% over the long term.
I also began to purchase my data (rather than free versions) to expand my services and opportunities.
June/July 2022 I continued to improve the presentation of the numbers breaking out important numbers, like the confidence level, into their own panels and page presentations.
In July 2022, I started to experiment and display results for artificial intelligence (AI) reading of the data and my calculations. It’s an ongoing effort that shows some promise.
Also in July 2022, I begin to do math outside of SPYtomorrow calculators to figure out best way to trade the data offered. This lead me to the SPY-T trading methodology. Directional trading was fine, but the ups and downs can be draining. Developing a directional, yet, hedge system seemed like a good solution for those who want to trade aggressively, yet have some hedging protection.
I launched a webinar July 20, 2022 for those interested in SPY-T trading.
Memberships are growing, and people are holding on to their subscriptions, which signals to me, that they are finding value from SPY-T trading or SPY tomorrow.com signals.
Algorithm Update and Daily Pressure
During the fall of 2022, the spring/summer algorithm began to slip to below 70% correct. This was a time of strong membership growth within the community. Much of the growth began after I started focusing on the spytomorrow YouTube channel and some videos, mostly the SPY-Ta video taking off and driving membership.
However, by December, it was getting frustrating (for all). I had added a lot of support options, however, I wasn’t doing a good job explaining the spytomorrow.com feature updates.
New sections such as hourly pressure updates were being tested. It was a time of fast growth and expansion of membership community and website features. However, I had to let off the gas by mid-December for my own mental health.
Yet, I knew need to make some adjustments and updates to the algorithm. So, that was my holiday project. Ultimately, I discovered the Blue section tended to lean positive, so I removed it for more accurate results as the 2023 algorithm update.
After, the initial solid launch, as I got into January, I soon wasn’t completely satisfied. There was an almost immediate fading of the accuracy of the 2023 algorithm update, I knew I had a bigger job ahead of me.
I decided to restructure the algorithm to the best of my best. Rather than 3, 9 or 27 results to base the pressure conclusion, this one will have 81 results. I call it the heart.
I back-tested it for over 160 trades and it averaged 75% correct. Although, some months were still sketchy. Making adjustments and figuring out the best strategy to apply to the heart took weeks. 50+ hours during one period of time.
Once I released it with my chin up high, it still wasn’t performing as I expected. I liked what I built. I believe it’s the best approach.
I started to question the data approach. There was an evolution away from the original data source approach as I began testing and developing new algorithm solutions over the last couple of years. I started the think more about how I developed the current and past data sources. When I was creating the heart, I went back to some of my best calculations. I also started saving different data sources to test later.
Beginning February 2023 I created a new approach to data using the concepts that kicked off this project. And it seems to work. I figured I would test it during the month, because I didn’t have much data collected. If it did well, I figured I would switch to what I refer to as the short-term data. However, by mid month the short-term data was correct 13/15 times, including up days, down day and passing the switch test. I felt it was necessary to make the adjustment then, because the current data set wasn’t doing as well.
I also noticed, that the short-term data made my based algorithms more accurate too.
As of Feb 16th, 2023. So far so good. I have SPY, IWM and QQQ heart running the short-term data is running live. Including back testing it’s 16/18 correct.
If you’ve read this far, you know I’ve been here before. However, those days of 80% correct often used shorter-term data. As I expanded data sets, I moved out further, it just seem safer and more logical at the time, especially the pre-switch solution days, I was trying everything and that data stuck for about 18 months – 2 years.
What I have running is the best of my best. Small tweaks may be reasonable, however, there is a limit. I think I’m about there (I’ve said that before too).
Heart Fading to Yellow
The concept of the heart is strong. However, the algorithms populating it began to fail soon after launch making for a tough few months. The main reason is the lowering volatility. I’ve realized more over time how volatility affects the various algorithms. Each algorithm works differently in various markets, specifically as it related to volatility. Also, poorly timed data collection updates played a negative role.
During these months, I focused on the SPY algorithms and eventually discounted the QQQ and IWM as part of the official pressure reading. I also updated data collection times and approaches with may be why the algorithms begin to fail soon after.
As the Heart was fading, I also began to try a new coding approach. I started using AI-suggested strategies and experimentation with automating Databased updates and calculations. AI certainly has allowed me to up the game. I first developed the MPI (Market Pressure Index). However, I gave them too much credit early on and ultimately I went astray with my development.
Stepping back again with new insight I reexamined older algorithms again. The Yellow series was working well in the current environment. A group of algorithms that had solid results over 2 years ago, again showed their promise. Especially in combination with the new data collection approaches which seem to be the best approach, but hurt the heart algorithm. I now refer to the heart as the red heart, because I’m working on the yellow heart, using the updated Yellow algorithms which seem to respond to the current approaches much better.
The backtesting through to the last data collection update is almost flawless. Which of course really hearts after a couple of months of disastrous results. Honestly, for the first time, since I started this project, I stopped trading the signals due to confidence. Because of this, I stopped promotion in May and focused on the fix.
All is not lost on the MPI front. I do have a meter which is currently referred to as the composite meter doing well. It’s a combination of three data sets. I’m still testing and collaborating it. I may become the updated version of MPI.
As of the weekend of June 16th 2023, The meters are still available but are now considered experimental and in beta. I will add and subtract the meters as performance dictates. The Yellow boards and Z readings are the break-out support for the current pressure reading.
I have more work to do, however, I do feel confident enough to begin promotion and solid trading again. It continues.
(August 29, 2023)
Unfortunately, the Red and Yellow hearts didn’t work as well (Yeah, I know, you’ve heard this before), the concept is fine, it all depends on what you put in the 81 squares, I’ll update the heart, which will most likely be Purple. The Yellow boards kicked off an entirely new direction which led to the Orange algorithms and the current Purple series.
The MPI era of measurement derived form AI is currently dead (for now).
However, the Purple takes data from the 3PM reading and compares it to the 3:48. Kind of like the N2VV. I’ve tested a few variations. The Purple started off well but needs support because there is a switch that I can’t account for, however, I have other calculations which are also affected by this switch and offset each other. Kinda works like guardrails. The newest Purple Max seems to account for this switch even when the B6 switch is added.
I’ve cleaned the house in the navigation. I really want a few algorithms that are consistent, do what they are supposed to do without an uncalculated switch or just breaking. This is my goal. I feel good, Purple Max seems promising. But, we can say it all together, we’ve been here before.
Yes, there is a point in which the effort must be given up. I’m about 85% (maybe more) there. If this series doesn’t work, I’ll need a new approach. I currently, don’t have many in mind.
(October 14, 2023)
The Purple series is doing well. I’ve developed variations of Purple. The Purple Max I is the public up/down signal. It’s 100% this month. Over the 61 days of tracking it’s 80% and when the core Purple algorithms are in aline, over those 61 days, they are 90% correct. Purple is holding. Let see how we are holding up in the new year.
Quest For 100%
Can the SPY signals be 100% accurate? No. Honestly, it’s impossible to predict a market 100% of the time. Simply, I’m amazed at the current success of the indicators. However, markets can change direction and mood for many reasons, with news being a big catalyst.
News can not be predicted (that is why it’s news). Yet, solid market pressure can minimize, enhance or erase news reactions quickly. It all depends on the perceived value of the news to the markets.
Another issue is I’m currently using third-party data. Sometimes the data is not updated properly, timely, or in a few cases, not available through data source glitches.
We are getting close. But 100% is still a dream.
I knew that many people were losing their jobs and face uncertainty in 2020, they turned to the stock and option markets. I wanted to help. This is why I offer basic indicators for free to the public after market close. There are many ways to use the SPY up or down direction indicator on the front page of spytomorrow.com for trading. My plan is to keep the indicator free for as long as possible.
I also share helpful information on how to use these indicators on my blog (available free and some limited to members), I post predictions on our Twitter @prosperous and support info on the YouTube channel.
spytomorrow.com is not for everyone
Not everyone should be trading stocks and options. Despite these indicators’ high value and accuracy, this doesn’t mean every trade is successful. Not every trader has the skills necessary to make short-term trades. 10 traders can have the same information and create ten different results.
However, if you do well with the free information, you are welcome to take the next step and join the paid membership group (when open), which offers additional insight and more actionable information.
I encourage people who have the skills and know they can make up the monthly membership fee with their trading to participate.
What if people want the data behind my SPY indicators? Well, I do not offer my formulas to the public. There is a lot of data available for free, and I’m sure I’m not the only person with a stock market direction calculator. I’m sure you can make your own calculator through tests and observation.
However, over the years, I’ve already done the hard work. I take the time each weekday to present it to the public for free and for a fee for those who need deeper actionable information.
I know professionals and big firms are investing greatly in artificial intelligence (AI), which may beat my indicators very well. However, I’ve seen many people lose money by giving money to professionals. Still, if you are unsure of your trading skills, and don’t want to actively manage your money, you may want to research to see what information and opportunities are available to you from professional money managers.
My goal is to support as many people as possible while providing additional income for my family, and employees and support the community, charities, and organizations I feel are helping others too.