Yeah, I wish I had bought stock back then too.

December 31, 2007 8:52 PM

The Keynote Index Fund. In which Matt asks "Is it possible to make money off the Apple keynote talks given by Steve Jobs?".
posted by Brandon Blatcher to MetaFilter-related at 8:52 PM (41 comments total)

um yeah. I posted it to projects when I was done with it. Dunno if it's worth a metatalk thread.
posted by mathowie at 9:06 PM on December 31, 2007

Yeah, but nobody reads Projects, Matt.
posted by BitterOldPunk at 9:41 PM on December 31, 2007 [4 favorites]

I voted for it! Yay votes!
posted by 31d1 at 9:58 PM on December 31, 2007

The last two years have been amazing, showing 7.3% growth if you held AAPL for 24 hours, and 11.9% if you held it for 48 hours.

How do you get these numbers, Matt? My calculations, for example, show 15% growth if you had employed this strategy (24 hrs) for the past 2 years. It looks like you're just averaging the yearly percentage gain. If so, thats not the way to do it...
posted by vacapinta at 10:14 PM on December 31, 2007

Ah, never mind. I now realize you meant annual perentage gain based on later statements.
posted by vacapinta at 10:22 PM on December 31, 2007

This is really cool, and probably belongs in the blue. Like BitterOldPunk said: nobody reads Projects.
posted by Ryvar at 12:34 AM on January 1, 2008

My bad, didn't know it was projects, I got it from daringfireball.net. Delete away if you like.
posted by Brandon Blatcher at 12:37 AM on January 1, 2008

There's a typo: "In a world where doubling your money ever 7 years is a wise investment..."
posted by TheDonF at 1:25 AM on January 1, 2008

Twitter is what all the cool stalkers are using.

although I admit I was surprised and mildly creeped out to see this in the grey
posted by Phire at 1:25 AM on January 1, 2008

So do the A-listers buy domains for every blog post they write nowadays?
posted by Plutor at 4:15 AM on January 1, 2008 [3 favorites]

"In a world where doubling your money very 7 years is a wise investment, one man seeks to go against the system--in a time of great uncertainty. Forget everything you know; ignore everything you see; and just maybe, you will hear what is hidden in your heart."
posted by Deathalicious at 4:29 AM on January 1, 2008

Well, the page is worth a vote as a cautionary tale of "A-listers Should Stick To Their Blogs" or "Why Fanboys Shouldn't Mix Business With Pleasure". Other than that, it might be best to hold off on that front page post. Because the figures and conclusions are wrong in very substantial ways.

First, there is inconsistency in how the prices are selected. 1998 uses low price of the day; sampling shows most or all of the rest of the years use closing price. Statistical purists would probably say that invalidates the entire dataset, but as it likely does not make a big difference for the point being made, I'll just note that changing the base comparison allows for malicious uses of statistics by far more nefarious types than a mild-mannered #1 admin.

OK, last two year's growth shows 7.3% on 24 hours and 11.9% on 48 hours? Half-right, but the right half appears mostly by happenstance. Growth is a compounding investment. You cannot average stock price deltas between years with compounding investments (look up CAGR). The actual figures are approximately 7.3% and 9.9%. Why can't you average deltas? Because you're not dealing with the same figures each time. If you bought $100K of shares one year and made 3% ($3K), next year took your $103K and doubled it as 100% gain (woo-hoo!, $206K), does that mean you made 3%+100%/2=51.5% over two years? No, CAGR is 43.5%. Calculator here.

Next, the point is made that this is for days and not years of investment, as if you could multiple the days by stock market sessions to get humongous growth rates. Well, no, you can't. Apple doesn't have a keynote every day. For the comparison to mean anything, to get those humongous growth rates, you have to find 50 or 100 other companies just like Apple to fill out the rest of the stock market year. Not only do they have to be like Apple in the formula, the companies have to cooperate to give keynote speeches neatly in a series so that you can conveniently make your money off one, and immediately roll it over to the next company throughout the year. The day -> year comparison has no meaning here, this is a once-a-year-only event and real investment returns cannot be theoretically multiplied by parallel universe events. (Hey, if Jobs gave a keynote every week, you wouldn't find these large fluctuations anyway).

Past five years is not 1.5% and 3.7%, but 1.4% and 2.3%. Past ten years is not 1.2% and 2.2%, but 1.1% and 1.5%. Thank you Google spreadsheet. Small changes, but they add up over the years.

OK, we're at the money shot: if you held the $10,000 of shares bought in 1997, your investment would be worth $525,187 today. Yep, absolutely, with your perfect 20-20 hindsight, you can see this happen, because the last couple years of Apple have shown unsustainable exponential growth of the stock price, unlike most other years. But how does that 10-year strategy work for the chart? What about the years in the chart leading up to that ten year end? Let's pick an investment number, any number, say $10000.00 and see the 1-day and 2-day figures:

        $10,000.00   $10,000.00
 1998   $9,710.34   $11,395.66
 1999   $10,195.26  $9,828.04
 2000   $10,344.46  $9,449.27
 2001   $10,738.00  $10,344.46
 2002   $10,379.92  $10,248.47
 2003   $10,345.09  $10,136.09
 2004   $10,307.76  $10,541.07
 2005   $9,650.07   $9,784.60
 2006   $10,260.41  $10,646.16
 2007   $11,112.75  $11,644.56

Wow, the first eight years look pretty grim. After eight years, you're down a couple hundred bucks of your ten grand, with a CAGR of -.27%. That's not good. But this investor is really smart, right? Smart investor knows that ten years is the magic number, not three, not five, not eight. Ten! Magic! Well, sure, ten is the lucky number. And buy-and-hold is the key after all, don't use that buy-and-sell keynote chart splashed across the page. Just do ten years buy-and-hold. Uhh, unless, for example, you buy in 1993 and run to 2003. January close of AAPL in 1993 was $14.875, January close in 2003 was, oops, $7.18. CAGR of -7%. Hmm, guess it's not just ten years, it's the right ten years.

Then there's the transactions costs and...but enough.

All this isn't to slag Apple, though it's tempting. It's to point out how seductive financial projections are when you use the benefit of perfect hindsight, and to show how those projections can be not nearly as great as they first sound. Heck no, Apple is okay with me, even it all it runs on amazing overhype. As a gamble, I made a thousand bucks on Apple last year selling covered calls before the Jobs keynote (Jan07 85 calls), proving that I'm not immune to the lure of playing the keynote game. And I still have an odd-lot 14 shares in certificates from a buy I made for another fanboy without a brokerage account back when AAPL was a split-adjusted $30-something. Yeaa, Apple! (Would be even more impressive if it wasn't pure dumb luck on my part for not bothering to dump the leftovers.)
posted by mdevore at 7:11 AM on January 1, 2008 [7 favorites]

Wait...what?
posted by danb at 7:16 AM on January 1, 2008

That would be a lot more entertaining if you shriek and run around like on Mad Money. Video pls.
posted by Mid at 7:21 AM on January 1, 2008

Yeah, I kinda felt like I was channelling there for a bit, but without the sound effects, it loses a lot.
posted by mdevore at 7:28 AM on January 1, 2008

I think the more important point, mdevore, is that Matt should get us all iPhones.
posted by cortex at 7:50 AM on January 1, 2008

mdevore, I trust your calculations are right, as I've never done them before and was just guessing by doing simple straight math on it. I figured someone with a real business background would correct me.

In the crazy imaginary world of my own head, I figured the stock must be shooting up 10% every year after the keynote. That's what I thought I'd find in the data, so I was surprised the keynote didn't have much positive affect on things and basically wrote it up as such, stressing that buy and hold was really the way to make any money from Apple in the last ten years.

And I bought a domain for it vs. a blog post as a bit of an experiment. I wanted to see if it was easier to track that way (via blog searches, stats, etc) and I wanted to experiment with a couple small textads on it that were isolated from everything else. After the keynote actually happens in a couple weeks, I'll write a post on my personal blog with a wrapup of stats on the 1-page site. It was only ten bucks in the end and I think it'll make tracking traffic and stuff to it way easier.
posted by mathowie at 7:54 AM on January 1, 2008

The idea brings up another point: if there is a simple formula to make a lot of money on the market, the market immediately discounts it. There are genius-level quants with years of education and experience who do this stuff 60+ hours a week, plus they have a lot more resources behind them than a quad-core box. Why does AAPL not jump $10 on the Jobs keynote? Because if it started to do that, the smart money would notice the trend and make the cash, knocking the difference back down before the little guys could move. Regular investors cannot make big money on short-term predictable trends.

Actually, I'm impressed you can see as much a difference as you did. Like I said, my playing the keynote last year was a gamble, pure and simple. There are pros out there who can spot a third derivative stock market trend in realtime; swat down amateurs like me, you, or about anyone on here without even noticing.

That's why the regular investor books always say buy-and-hold, though many of us don't, and though it frequently doesn't work anyway.
posted by mdevore at 8:25 AM on January 1, 2008 [1 favorite]

An interesting comparison would be a car dealer after a big annual car show, I'd think, or Sony after CES or something. Are there any other companies that have comparable yearly 'events'?

If there aren't, why aren't there? Surely Apple isn't the only company that releases cool new products every year. Why is Apple the only one that gets people to give a shit.
posted by empath at 8:33 AM on January 1, 2008

" January close of AAPL in 1993 was $14.875, January close in 2003 was, oops, $7.18. CAGR of -7%. Hmm, guess it's not just ten years, it's the right ten years."

Were there any splits in that ten years?
posted by mr_crash_davis at 9:17 AM on January 1, 2008 [1 favorite]

I was looking at a split-adjusted chart over the same years for the prices, so it should be the same. Otherwise the chart would be a useless tool for investors if there were any splits. Generally they do say they are split-adjusted, same as with histories.

There are worse 10-year stories, if you cherry pick back to 1985 for highs and lows of various months with 10 years differences. As always, the person who picks the starts has a lot of control over the perceived end results.
posted by mdevore at 9:22 AM on January 1, 2008

Dunno if it's worth a metatalk thread.
posted by mathowie

E-mail a mod about this.
posted by The Deej at 9:24 AM on January 1, 2008 [3 favorites]

There are genius-level quants [mefi] with years of education and experience who do this stuff 60+ hours a week, plus they have a lot more resources behind them than a quad-core box.

I've seen the same "genius level quants," make the same, stupid predictions. Just because you can do math at the level of an electrical engineer does not mean you are better are picking stocks. Granted these are very, very smart people ... but their very intelligence is their hindrance. At least Matt would go, "Yeah I was just fooling around, I guess it does have some great risks," while your average quant goes, "The data exceeded the models! It wasn't suppose to happen! Ah fuck it, give me $200,000, we'll call it even and I'll go start a charity and start spamming popular blogs."

But more importantly let us look as to why this is probably not a good idea. Each time you make the trade you have frictional/transactional costs which are not accounted for. It will totally eat away at the little gains and magnify the losses. But more importantly you're putting $92 into a stock you know will probably not move more than 10% but has the potential, the very real potential, to drop much, much more than that. That's an expensive bet to make.

And more importantly, let us say you are more rigorous, and look at the volatility for the stock on a 24-hour (or is it 48-hour) moving average. Your hypothesis would then be along the lines of, "volatility during keynotes exceeds this average and is misplaced." And let us go a step further and you created a derivative model that bet on the volatility exceeding it during this time period (only when it was cheap to do so) and below it during the time period. I would still say the risks would be too great, but it'd be an interesting model to test. The important lesson is that you cannot know if the trade was good after the trade ended. I just don't think that is possible, even if, say, the trade worked out fabulously. Because what is good and what is luck is too hard to discern, such that you need to have a trade that before hand is so cheap, that you couldn't pass it up. Sort of like buying every Lexus you see for 50 cents. Sure the vast majority of them are probably worth nothing, but there is a good chance the one that you bought for 50 cents is worth $50,000 and suddenly it is all worth it.
posted by geoff. at 10:27 AM on January 1, 2008 [1 favorite]

I just discovered this the other day, and it's somewhat related. But did you know that Nintendo is now a bigger company then Sony? Sony, with all their TVs, Music, Movies, video games, etc. has a smaller market cap then Nintendo. Nintendo has an 85 billion dollar market cap, while Sony has just $54 billion. And That nintendo is the third largest company on the Tokyo Stock market behind Toyota and "Mitsubishi UFJ Financial Group Inc", a bank.

Pretty crazy.
posted by delmoi at 10:31 AM on January 1, 2008

Nintendo has this crazy corporate policy of not totally shitting the bed every time they want to roll out a new product, I guess. VirtualBoy notwithstanding.
posted by cortex at 10:55 AM on January 1, 2008

It's that notwithstanding part that makes this a dangerous bet.
posted by smackfu at 11:18 AM on January 1, 2008

Naturally you hear all the sexy horror stories about the quants (correct link this time) whose models blew up, thereby burning X dollars where X is the GDP of small country. Still, I would think that if there are private companies successfully using quants or advanced modelling they would keep a low profile. It seems a bad idea to advertise the presence of a successful model, because mere market knowledge of that success could affect returns. I am not convinced that all the very very smart people are no better than us relative dummies when it comes to playing the market.

Alternatively, there is Taleb's maverick Black swan theory, which I believe he uses to invest in the market. I'm not sure how or if it could be applied to the concept of Jobs' keynote speech affecting AAPL price, but the idea is an interesting one. Perhaps the application would be purchasing dirt cheap put options in anticipation of a huge stock drop if the keynote news is poorly received. That sounds close to your 50 cent Lexus idea.
posted by mdevore at 11:28 AM on January 1, 2008

Still, I would think that if there are private companies successfully using quants or advanced modelling they would keep a low profile. It seems a bad idea to advertise the presence of a successful model, because mere market knowledge of that success could affect returns. I am not convinced that all the very very smart people are no better than us relative dummies when it comes to playing the market.

Yeah but if you look at the data, even statistically advanced modeling like GARCH or other packages that try to compensate for the blow-ups do not do, on average, better than much simpler methods. Even funds like SAC Capital which are "quant funds" do not really use quant funds in the terms that most people would be using to describe it. What I mean is that SAC Capital and those types of enormous return funds that keep everyone incredibly quiet use networking (e.g., calling up friends at Bear Sterns, Goldman, etc.) and other you-rub-your-back-I'll-rub-yours, more often than most people would probably be comfortable with. Not that there haven't been good quant funds, I have yet to see or hear of one (including Renaissance) that relied totally on developing incredibly complex models and using them as black boxes and putting their feet up, which is what their models imply they should theoretically be able to do. I think they like to hide behind using incredibly complex models instead of the more simple explanation ... natural intelligence and connections. Of course you and I could be talking past each other and have different definition of what we mean by quantitative finance. Of course the main, and glaring exception to this, are the funds that trade by the millisecond or smaller. But they act more in the traditional sense as market makers, and less of a fancy machine that picks stocks.

In practice I think Taleb goes more off of historical aberrations, like when volatility skews indicate mis-pricing, and could probably be summed up better by the Dodd investing method of buying things when they are cheap, with a mathematical bent. But this is at its heart, what a lot of quant funds already claim to be doing. There's a couple of derviativesstrategy.com articles about Taleb's strategy before he put on his philosopher hat that are interesting. It was back when he was a trader's trader and before he kind of went crazy with VAR and Black Swans.

But of course I could be completely wrong, the more I learn about how the secretive firms work, the less I realize I know. The cynic inside me has come to believe there's a lot more, "Hey look over here, quant models, quick look," and while everyone is looking over there they do something quasi-legal that doesn't take an MIT PhD to figure out but nets them those 20% monthly returns. Rubinstein explains some of these tactics, as does Taleb in his earlier textbook. It is not as fun intellectually to think that it is an old boys club, so that's why I keep on reading the journals and keep on believing that anyone with a good model and some capital can tap into the market.
posted by geoff. at 12:20 PM on January 1, 2008 [1 favorite]

"...even statistically advanced modeling like GARCH .."

or, more properly, A-GARCH, to capture the well documented asymmetric behaviour of equity market returns -- e.g., while it's depressingly common to see 100 plus point drops in the stock market, such dramatic moves to the upside are statistically rarer, thus the need to develop GARCH models - "A-GARCH" capturing this asymmetry for modeling purposes. But even so these models, while performing well under certain conditions, don't have 100% predictive power.

Why? Well, I think advanced modeling capabilities, detailed market knowledge and personal connections all play different parts at different times in different market conditions. And to different effect.

A current example: all our models were pointing in one direction last July while Goldman went out and (almost unilaterally it would appear in retrospect) shorted the ABX. Why? Who made all the money after the fact when subprime started to burn off big time?

Goldman. They were short a key index of subprime market, and as the market moved downward this bet sharply increased in value.

They weren't looking at past performance. They weren't solely marking to the model, and blindly doubling down as the retail money does when trades move against them. No, instead a few of their traders had a gut feel and put a big bet on the table.

That's the single factor that would kill a trade like is being suggested here. I'm a long time Apple proponent but even so I know no company, even Apple, will consistently have killer product introductions. Look at it this way - they've had ten years or so of great product intros, moving from strength to strength; what natural law says this trend will continue?

Jobs might walk out onto stage this month and announce what will be widely viewed as his Pippin. Guess what that misfire would do to this trade?
posted by Mutant at 2:09 PM on January 1, 2008

After the keynote actually happens in a couple weeks, I'll write a post on my personal blog with a wrapup of stats on the 1-page site.

I'll look for that with interest. I wonder how much of a traffic bump you'll have gotten (as compared to just Some Random Web Guy) because, you know, it's you, and you've got deep wells of internet cred and goodwill and stuff.
posted by stavrosthewonderchicken at 5:00 PM on January 1, 2008

That is pretty crazy, delmoi. Apparently, this was all news to me, Nintendo has also been around for over 100 years and owned everything from a cab company to a "love hotel" in their earlier days.
posted by Roman Graves at 5:29 PM on January 1, 2008

I knew Nintendo was a 100+ year old company, but I'd always thought of them as a smaller company then Sony. I guess that's changed as Sony has fallen down the well of screaming about IP while producing crap, and Nintendo has been poster child for innovation.
posted by delmoi at 7:39 PM on January 1, 2008

Depends how you define 'bigger'. For instance, Sony employs about 160,000 people, while Nintendo employs only 3000. Market cap isn't necessarily a good way of determining size, if a company's size even means anything.
posted by Burger-Eating Invasion Monkey at 8:06 PM on January 1, 2008

Usually gross revenue is used for size comparisons like the Fortune 500. By that measure, Sony is at $70 billion while Nintendo is not even close at only $7.8 billion.
posted by smackfu at 8:24 PM on January 1, 2008

Also, don't forget that the keynotes fall in January, which is subject to the "January Effect".

But matt's experiment here with apple underscores the ancient market adage "Buy on the rumor, sell on the news." The same study could be applied to earnings announcements (and has).

Heck no, Apple is okay with me, even it all it runs on amazing overhype.

You know, I used the be the flag-bearer of this movement. I still don't get why Apple is so great when every time I enter an Apple store I want to put my fist through the admittedly delicious monitors.

But then I read a story about how popular GPS receivers were this year, and it occurred to me that if Apple tosses a GPS receiver chip into its iPods or iPhones, it will dominate/decimate that industry as it has the audio player industry.
posted by Pastabagel at 8:25 AM on January 2, 2008

A story on Matt's site and this very thread are now on the front page at cnn.com.
posted by ewagoner at 10:06 AM on January 2, 2008

Ha! That CNN site is pretty keen; I wonder if anybody's ever made an FPP about it?
posted by cortex at 10:10 AM on January 2, 2008

I dunno, cortex. I was hoping to keep it my little secret, which is why I didn't make it a real link. I try to visit it at least once a month, and they've got new "stories" there every time.
posted by ewagoner at 10:26 AM on January 2, 2008

Fair enough. I guess it's time I introduce mefi to this one parody newspaper website I found instead. Man, the things they say—it's funny because it's true, you know?
posted by cortex at 10:34 AM on January 2, 2008

Raked over the coals CNN? Oh, they have not seen raked over the coals.
posted by geoff. at 10:40 AM on January 2, 2008

I plan my grocery shopping around Apple keynotes. As soon as Jobs stops speaking, I go heavily into breads and cheese. During the following days I shift my focus to vegetables, and then when the furor dies, it's back to fruit.
posted by blue_beetle at 11:43 AM on January 2, 2008