SPY Data - Predictive Modeling

I’ve been running data on spy and am finding some interesting results. I’m not sure how much past results will predict future results, but I’m intrigued and would like to get others’ thoughts on this. If anyone would like to see the actual data in Excel, I can provide it individually.

Based on my analysis, I’m considering selling about 45% of my investments around the first of the year and waiting till March-June timeframe to reinvest, or when the data groups tell me it’s time. Most of my investments are in spy currently.

The Data:

  • S&P 500 daily close figures from yahoo finance (S&P 500 (^GSPC) Historical Data - Yahoo Finance). This goes back to 1929.

  • Calculated historical 1 year and 3 month returns per day so I could gain some perspective on each day’s close value

  • Calculated future 30 day and 1 year returns so I could attempt predictive modeling

  • Created arbitrary return groups (5 groups) to minimize data complexity. The return groups were

  1. less than -12% per year

  2. -12% to 4% per year

  3. 4% to 12% per year

  4. 12% to 24% per year

  5. 24% + per year

Findings:
There were clear monthly trends that I’ve already heard discussed on various forums and CNBC (i.e. September is the worst month of the year).


image

Based on this data, I conclude that it is best to buy in June, July, and Dec and sell in Feb, Aug, and Sept.

Findings - Part 2
This is the part that I find most interesting and would like additional input. It’s complicated to look at, but I’ll explain.

The first pivot table provides a % likelihood that the next 30 days return will fall in each grouping (1 is bad, 5 is terrific). The row labels show the historical 1 year (1st number) and 3 month (2nd number) return groupings. Think of this as the historical data in slices. For reference, today we are in the 5,3 slice. So if you go down to the 5, then find the 3, you see the % likelihood the next 30 days’ return will be in each grouping. I’ve labeled this pink because over the next 30 days there is a 46% likelihood that SPY will be negative or just barely above breakeven (adding together groups 1 and 2). There is only a 47% likelihood of significant gains (groups 4 and 5) which would generate over 1% for the month. This is compared to the slices such as 5,1 4,1 2,4 and even 4,3 and 4,4 which have likelihoods exceeding 50% for significant gains with less risk of loss.

The next pivot table shows similar data but with 1 year future returns. Notice that the 5,3 slice still does not look desirable. I did not color code it because it falls middle of the road within the data set. However, several other slices look much better.

The 3rd pivot table shows average returns for the slices for 30 days and 1 year which is also helpful, but fails to show volatility/risk. Notice here again, the 5,3 slice shows a lower than average 30 day return of 0.64% and a 1 year return of 6.64%.

Here is a screen shot of some of the data for reference. You can see to the far bottom right is where the slice is assigned. I do find it interesting that only a day before we were in the 5,4 slice, however, results for those two slices are similar.

Given the billions spent by hedge funds and by banks for analyst salaries, I’m sure there are models a trillion times more sophisticated than this. But I do see some potential value in this for regular folks like us trying to make some extra money as a side hustle. I look forward to some input from this tremendous group of people.

25 Likes

Sold all my spy shares today, Santa rally over. Now will see if the analysis can predict future results. Will post updated data and results next week.

2 Likes

Update: so far the predictions are holding. When I posted this originally on 12/16, the S&P 500 was at a 5,3 meaning the 1 year returns were high (5 is the highest on my scale) but the 3 month returns were only mid-range (3). Based on past performance, the data indicated this is bearish. It’s not quite 30 days later yet, but right now, the return is negative. the S&P 500 was at 4719 and is now at 4697.

I did hold my positions in SPY till the 31st since I assumed a Santa rally was imminent, which did occur. It ran up from 4719 to 4775. I will post again once we get to 30 days from the original post to see where things stand.

3 Likes

Here is the updated data.

Since the end of day trading this past Wednesday the 5th, the S&P 500 has been at a 4,4. This is bullish. Depending on where Monday opens, this is probably a great time to buy back into SPY. There is a 55% likelihood of future 30 day returns of over 1% and 53% likelihood of future 1 year returns over 12%.

Here is the chart showing the breakdowns for the different levels (i.e. 5,3 or 4,4, etc.).

image

I’ll check back periodically to update the results.

2 Likes

Now that we are 30 days past the original post, I can report the original call to sell on 12/16 was correct (30 day return of -2%). The recommendation to hold till the Santa rally was over on 12/31 and then sell was even better (return of -6% so far this month).

The recommendation to buy on 1/7/2022 so far has resulted in a return of -4.15% which is not good so far. But we’ll see what the 30 day return looks like on 2/7/2022 to know for sure how this will turn out.

Data as of today suggests it is NOT a good time to buy. The results for today were a 4,2 which is neither bullish or bearish, it’s middle of the road. I would rather wait to buy when the data suggests a stronger likelihood of positive returns. It has been at a 4,2 five days now so we are in a holding pattern.

3 Likes

I have also been running advanced analysis to determine the highest likelihood of success plays (HLSPs) if held for 30 days. This restricts the amount of recommendations to the very few times when the data is in alignment 60% of the time or higher. In some cases, as high as 88%. I simulated the recommendations every day the market was open since 1/1/1980 to determine how often the recommendation would have been accurate. Here are the results.

image

So these only occur about 12% of trading days and then they are usually lumped together over a period of days in a row. In 2021 there were only 8 of these for the entire year and they occurred in successive days from Jan 15 - Jan 26 and on Feb 3. They were all in the 4,4 return group. The results were positive with 6 of the 8 recommendations being correct. Had one played each day, the result would be a total of 9% return (4%,2%,1%, 2%, 0%, -1%, 1%, 1%).

I believe these could utilized for options plays. Playing these recommendations with just ITM or ATM call options expiring 30 days later could be extremely profitable. I’ll continue monitoring the data and report the next HLSP and begin purchasing calls with screen shots for real-world tracking. My hope is that these HLSPs will be reliable enough to invest a higher % of my portfolio in these when they occur and achieve significant returns over market.

Does anyone have any data related to the return % of calls vs shares for every incremental increase in SPY for just ITM or ATM calls?

I believe for every $1 the stock price increases, the calls typically increase $1 for ITM calls. So if the spy call price is $10 and the stock price increases $8 over the course of the month, the call price should increase to $19 which would be an 80% return from a share price increase of 1.86% (average monthly return since 1980 for these recommendations). This assumes a spy stock price of $460 and 30 day ATM call option price of $10 for 30 days out.

4 Likes

After Friday’s close, the return group moved to 4,1. This is important because 4,1 is a HLSP or High Likelihood Success Play according to the advanced analysis I discussed in the prior post. Not only is it a HLSP, but it is the second greatest HLSP, so this is a very unique opportunity. I’ll be purchasing SPY calls Monday unless it runs too hot pre-market.

6 Likes

Keep in mind you have to buy delta 1.0 calls for $1 underlying increase to be equivalent to $1 option price increase. Those are probably much deeper ITM and not cheap.

Edit: took a look. For February 25, a delta 0.9 option would be the 380C, which costs $60x100

1 Like

Thanks for that clarification Dakk. So buying just ITM or ATM options will not rise dollar for dollar with the stock price. If the delta is 0.50, does that mean it should rise $0.50 for every dollar increase in the stock? If so, I calculate the return as 13% for the option if SPY increases 1%. This is based on current price for 429s of $16.

I purchased 1 spy call at 429 with March 4th expiration at $16.65. See screen shot below.

2 Likes

I averaged down here and got another 429 for 13.42.

Yes, delta 0.50 means that the option should rise $0.50 per $1 increase on the underlying. Although that’s localized, so as the stock is moving, the delta is also changing due to gamma (which measures the change in delta per $1 change of the underlying). But for smaller dollar movements, it’s probably pretty close.

Got it, thanks Dakk.

1 Like

Spy made a wicket turn in the right direction after I purchased calls yesterday. Up 40% at end of day. By the time I noticed, the market was closed so I couldn’t lock in any gains. I didn’t expect to day trade this, but I’ve learned from past losses to lock in gains when possible. Since pre-market was very red, I sold both calls at open to protect my gains. They sold for $17.99 each which was a 20% gain ($15 avg cost). Later in the morning I bought back in, but could only buy 1 since I failed to ensure I had sufficient cash to day trade. Cost was $17.19. I’ll look for re-entries over the coming days.

The return group is still at 4,1 (second highest HLSP), so I’m still actively looking for entries to add calls. Since this is my first experiment with this new analysis, I’ll probably buy no more than 3 calls total and limit my exposure to under $6k.

With the gap up this morning and candles breaking recent lows, I sold to secure another 20% profit on the 1 remaining call I had. Looking for re-entry today. As of now, with spy at $439.50, the return group is still 4,1 so overall a great time to buy. If/when it finds a bottom, I’ll likely buy two or three calls. I like the 429s exp 3/4 since it aligns with the orignal signal. Since we are still at a 4,1 and the signal is still strong, I may look at higher just ATM calls with expiration closer to 3/10 which would be playing the most recent signal.

1 Like

When the market dropped during power hour, I bought back in to replace the call I sold earlier and saved 25% by missing the drop. See screen shot of the sale and re-purchase below.

On my closed positions, I’m up $946.94 so far (20%).

I then bought a second 429c exp 3/4 to replace the other one I sold yesterday at $17.99. I bought the replacement for $15.35.

I also decided to buy a few OTM calls for 3/18 exp since the buy signal is still strong at 4,1. This is the max I want to invest since this is still a beta concept.

Here is the updated data showing 4,1 return group.

2 Likes

I wish I understood it, but I will still say this is amazing work you’ve been sharing.

1 Like

Here is an update. I closed all my calls except for 1 during market run ups last week. Here is a summary of the closed positions. So far the signal has generated for me a 25% ROI for a total of approximately $2,000.

I do find it interesting that I’ve had to keep a close watch on this to ensure profitability. Had I not averaged down, taken gains, and reinvested, I would be sitting on a slight loss at this moment based on the first trade. Still lots of time before March 4th, so we’ll see what happens.

The return group is still a 4,1 so should still be a great time to buy SPY calls. I’ll be looking for possibly entries tomorrow. Here is the updated SPY data.

It did change to a 3,1 on Thursday, but flipped back to 4,1 on Friday. 3,1 is also an HLSP. I am planning to keep the 429 March 4th call I purchased last week since I’ve more than covered my cost basis and I’d like to see what happens to that call between now and expiration. It’s currently up 30%.

image

Great day for SPY. I did not buy any additional calls. Below is a screen shot of the current unrealized performance of the remaining call. I probably should have sold it before market close but I was busy at work today and didn’t see SPY up till after 4pm.

The return group is still at 4,1 so I’ll still be looking for buying opportunities when time permits. If you buy SPY calls tomorrow, buy them with expirations 30 business days out (after 3/14/2022) if you are trying to follow this predictive data model.

Below is the updated data after close today showing the 4,1 return group.

3 Likes