04 January 2011

Inside an Impossible Prediction for the S&P 500

Sometimes, the biggest challenge we face in forecasting how stock prices will change over the course of the next month is simply keeping ourselves entertained as we do it.

Impossible!

We really ran into that problem last December when, instead of posting a prediction of where stock prices would average during the month, we entertained ourselves by mocking a couple of hapless article commenters on Seeking Alpha, who were clearly of the opinion that they knew it all as they offered their "insights" on our November forecast for the S&P 500.

In a way, the exercise was every bit as fun as shooting fish in a barrel with a Gatling gun. But that wasn't where the real fun for us was to be found.

For us, the real fun meant taking on the challenge of crafting an impossible prediction. Something that we would do because it just couldn't possibly be done, unless.... Well, we'll get to that later.

We began making our impossible prediction by first closing the door on our previous prediction for the month of November 2010, which we had uncharacteristically left open-ended for that month:

We told you exactly where we expect the average of the S&P 500 would be going forward, which we expect should hold until we say otherwise since we didn't specify when our forecast range would no longer apply.

Let's stop the clock on that prediction here, shall we? Here's the running average of the S&P 500's daily closing prices for each day starting with 25 October 2010 through 30 November 2010:

Average S&P 500 Price Level, 25 October 2010 Through the Indicated Date (Final Date is 30 November 2010)

Hmm. The average of the S&P 500's closing prices for each day since 25 October 2010 and ending on 30 November 2010, has fallen to be within a level between 1182 and 1218. A quick check of the historical stock price data over that time reveals that the lowest closing value was 1180.55 (30 November 2010) while the highest closing value was 1225.85 (5 November 2010).

And here we are now, some 26 trading days later (that's 37 calendar days, MikeyLV) and only with the exception of just five of those days did the S&P 500's daily closing stock prices even fall outside our forecast range during all that time, while the average (ATTN Trader14: that would be the thing we actually track) was well within our predicted limits.

Gosh, it's almost as if we just used our model to predict the S&P would move to where it is right now!

Let's take a closer look at one part of that excerpt: "Let's stop the clock on that prediction here, shall we?" "Here" would be at 5:35 AM on 1 December 2010, which was hours before the stock market would even open that day. The last closing value that anyone would know about at that time would be the previous day's close of 1180.55.

Let's next see what happened to stock prices after that point, shall we?

Daily and Rolling Average S&P 500 Closing Values December 2010

What we see is that the month of December 2010 started off with a bang for the S&P 500, as the index leapt upward by 2.2% on 1 December, the biggest one-day change for the stock index during the entire month. After which, investors never looked back.

Checking more closely over the daily market action, we see that after the S&P 500 jumped to 1206.47 on 1 December 2010, the market then largely rose steadily through much of the rest of the month. By 31 December 2010, the S&P 500 closed at a value of 1257.64, with the average of daily closing values for the index during the month set at 1241.53.

We clearly picked the right day to stop the clock from running for our previous prediction. But believe it or not, that's not the clever bit for our impossible prediction.

That bit begins at the end of our post, in which we declared that we would not offer a public prediction for where the S&P 500 would go during the month of December 2010:

As for where stock prices will go next, well, you'll just have to wait and see. We're taking this December off from making public predictions for stock prices.

And we certainly didn't offer any public predictions for where the S&P 500 would go in the month of December 2010. That doesn't mean however that we didn't make a public prediction of where stock prices would go for December 2010.

We just did it before December 2010. Specifically, 21 days before December 2010.

And to point to that public prediction, we hid a clue in plain sight, in the second paragraph of our 1 December 2010 post:

But unlike the posts here at Political Calculations, which doesn't allow comments at all*, the posts at Seeking Alpha allow for comments to be made. Which brings us to the two "individuals" who we'll describe as being perhaps the two stupidest commenters on Seeking Alpha. Yes, we're going to get all Barry Ritholtz on your sorry posteriors!....

Honestly, we have no idea what it would mean to get all "Barry Ritholtz" on someone's sorry posterior (you're welcome to use your imagination if you're so inclined), but what Barry Ritholtz offers, that we don't, is an active public comment forum. Which would be something that we might expect the kind of industrious commenters who like to tack on their insightful "insights" to articles posted on Seeking Alpha to spend hours of their lives searching through for the sake of sharing their insightful "insights" with Barry's readers.

And of which we took advantage on 9 November 2010 at 4:13 PM in responding to Barry's post titled Late to the Party: Giving Stocks a Green Light:

Maybe that counts as daring on a day when stocks are headed in the opposite direction, since they're down by nearly 1% (at this writing). Then again, if nothing breaks (hear that Ben Bernanke/Congress/banks?) we would see stocks rise to 1225-1257 next. For now though, we'll stick with average prices during this month falling between 1182 and 1218, so a falling market now does represent a bit of a buying opportunity.

"For now" referring to the prediction we opened on 25 October 2010 and closed on 1 December 2010, with "1225-1257" being the range of into which the S&P 500 would go "next."

The S&P 500 got to the low end of that range just a week after 1 December 2010 and reached 1257 on 30 December 2010 before holding near that level through the last trading day of 2010. For the entire month of December, the S&P 500 averaged 1241.53, just above the midpoint of our forecast range (1241). That's a closer margin than what we were off by for our previous prediction, where the S&P 500 averaged 1198.55 in November 2010, just shy of the midpoint of 1200 for our forecast range.

Quoting ourselves again:

Gosh, it's almost as if we just used our model to predict the S&P would move to where it is right now!

Well, we're happy to close out that "next" prediction today, which finally lets us cut loose of those two commenting chumps who started this whole impossible prediction thing off.

But before we do, let's give the two chumps something to think about, since what we just described doing, we couldn't possibly have done. What with all the billions of transactions among millions of traders that occur each month in the stock market, especially if you believe that how stock prices change is fully random. Here are the possibilities:

  • We're very, very good at predicting where stock prices will go in the future.
  • We're very, very lucky in predicting where stock prices will go in the future.
  • We have the means and ability to manipulate the market to make it match our "predictions" of the future.
  • We're very, very sneaky and placed hundreds of predictions for where stock prices would head "next" among the comments at Barry Ritholtz' blog, selecting the only one that matched reality after the fact for this post. But because it's Barry's site, over which we have no control, they would all have to still be there - an industrious commenter would just have to go find them to prove us to be frauds....
  • We have a time machine and are willing to use it for the sake of performing cheap tricks involving stock market predictions.

On that last note, we'll close this post by linking to a discussion of time and how it works from a noted expert on the topic.


To which we'll cryptically add: 1268-1313. In the absence of an excessive amount of noise or a change of fundamental outlook.