When I’m not making games, I’m still cranking away on my day job–Piqqem. Eric Schonfeld did a review of the site on TechCrunch last Friday, so I thought I’d respond with my own two cents on the company. Clearly I’m a little biased since I work there (and wouldn’t be if I thought it was going to fail), and it certainly doesn’t hurt that I know where the site is headed. I should also point out, though, that these are my own thoughts on the site and not the “company line” (assuming we had one). I could paraphrase like crazy, but I think it’s way easier to just quote the original article and put my own thoughts on the subject.

If nothing else, Piqqem is certainly a good place to get ideas for stocks to invest in. But does it really have any chance of ever beating the market? Like any social investing site, its picks are only as good as the people who contribute to it.

Maybe I’m a dirty hippy, but this just seems elitist to me. I already put way more faith in my dad’s/uncles’ stock recommendations than I do in anything I read on Yahoo Finance. None of them have finance degrees (though, to be fair, a couple of them are accountants). If the monkey at the zoo gave me good advice, I’d at least listen to it. I think we can make some assumptions about who would even sign up for a site like Piqqem (educated enough to use the internet, some interest in the stock market) that I’d trust any of our users at least as much as some Wall Street analyst with an Armani suit and a who-knows-what hidden agenda. I’m not going to run out and buy something based on what one single Piqqem user says, but if 5,000 of them are saying they believe in a stock I might just give it a second look.

When it comes to stocks, the best prediction market out there is the stock market itself. It is the biggest prediction market out there, with millions of people predicting the future price of stocks every time they buy or sell shares. All of those predictions are aggregated together in the form of the price.

This is probably the number one complaint we get from people, but I don’t think it’s 100% true. Doing a very unscientific survey of a few companies (AAPL, MSFT, GOOG, YHOO, JAVA) shows that each of them has at least 60% of their outstanding shares owned by either institutions or mutual funds, meaning average trader has very little to do with the movement of the price. Average is a relative term, of course– clearly I’m not talking about Warren Buffet, but at what point does a trader become average? Less than $5 million? $1 million? $500,000? I’ve only got four grand invested in the market (the max contribution to my Roth for the one year I’ve been out of school), so clearly I’m not swinging the market one way or the other. Even if I did feel like a market mover, there are two ways that I as a consumer can affect a stock’s price: by purchasing the stock (a vote of confidence in the company) or by purchasing whatever it is that company produces (or using its service). I use Google Ads instead of Yahoo Ads. I bought a Honda instead of a Ford. Being from Atlanta, I’m a die-hard Coke fan and an avid Pepsi hater. Earnings may be released quarterly, but public sentiment is something that’s constantly evolving. Piqqem is one way for people to gather that type of data.

The votes should be weighted by money. If someone is willing to put down a million dollars on the shares of Apple going up, that should count more than someone who just puts $100 down (or worse, is just clicking on a Website). Money not only makes the predictions of a market better, it also makes it harder to game the system because there is more to lose if you turn out to be wrong.

We are Crowd Technologies, and while we don’t necessarily stick to the Wisdom of the Crowds a hundred percent of the time, we do at least try (one of the reasons you won’t see a “Best Piqqer” anywhere on our site). If I had a million dollars, I would’ve put it all on Apple when it was below $90 a couple of weeks ago. The fact that I have closer to $100 doesn’t make that feeling any less strong, so to say that my sentiment is meaningless goes very strongly against the idea of crowd intelligence. Other than an occasional Super Bowl pool, I’ve never placed a bet on a football game. I still submit my game picks every week, though, and it’s fantastically rewarding when I get more right than my buddies. That may be comparing Apples and Falcons (har har), but both tasks require me to take what data I have and apply it picking which entities I feel have the best shot at “winning.” The stock market is infinitely more complex than a football game (especially considering there’s not a winner and a loser and 10,000 teams are playing against each other at the same time), but I think the analogy holds. There’s a joy to be found in “being right,” regardless of whether or not I have a financial stake in the outcome.

[Assuming] Piqqem is perfect and it does produce better stock picks than any one individual could, investors will quickly discover and then it will become a reflection of the market. Piqqem does not exist in an information vacuum.

This is a problem we’d love to have, which brings me to my biggest question: who says the crowd has to be right? With enough data from enough users on enough stocks, I can pretty much guarantee we can slice the data so that someone is always “right”. And I don’t mean someone as in Steve from Tuscon (apologies if there actually is a Steve from Tuscon). I mean subsets of the big crowd. Maybe men aged 18-25 are better at predicting which way EA’s stock will move. Maybe doctors and pharmacists are great at choosing biotech stocks. Then again, maybe middle-aged housewives from Texas are fantastic predictors of semiconductor stocks. Highlighting any one of these makes for an entertaining story (couldn’t the market use a little of that?), which is enough to make the idea of Piqqem a fun one to work on. As a country, we love our pop culture indicators–from small towns that predict major elections to groundhogs that predict the weather. Who’s to say that crowd sentiment can’t be another one of those?

When I originally tossed the Stockmoose up a couple of months ago, it was mostly a prototype–one that had taken a single evening to produce, and one that was based on a single request to my artist girlfriend: “Can you draw me a moose with a tie?” Well, now it’s finally back with a new coat of paint, some basic anti-gaming measures, and a few other things to spice it up. The “borrowed” Yahoo stock charts have been replaced with our own proprietary charts, and each stock now has a sort of miniature info card so the choice isn’t based solely on name-recognition. We also created a list of 25 Silicon Valley stocks that most people around here have probably heard of–just to make it a little more engaging. Some of the early results are actually a little surprising. In our SV25, TiVo is actually pretty close to the bottom while Netflix is near the top. Based on what I know of the two, I would’ve actually assumed this to be the opposite of what would happen.

The most interesting result, though, has been the clickthroughs. As an experiment, I tossed $50 into Project Wonderful to point some traffic at the moose. PW grew up around web-comics, and you pay for time instead of clicks. The resultant CPC is incredibly low ($0.04-$0.05), but it’s extremely low quality traffic–totally untargeted, and often times people just click on the ads because they think it supports the comics they like. Roughly 1 in 8 unique visitors visiting the Stockmoose have been clicking through to Piqqem, meaning the $0.05 CPC for the StockMoose turns into a $0.40 CPC for Piqqem itself. Because our niche is in the financial sector, most CPC ads on Google cost anywhere from $.60 to $1.20 for even moderately relevant keywords.

It gets better. The bounce rate for traffic coming off of the Adwords was in the neighborhood of 70%, while those coming in off of Stockmoose are closer to 40%. I don’t know if we have enough data to be statistically significant or not, but the concept seems sound. Instead of trusting Google to target our ads to relevant “customers” (they’re good at what they do, but there’s thousands of people out there who make a living scamming AdSense), we instead cast a much wider net and do the targeting ourselves. Those that are interested in the stock market, interested perhaps in the wisdom of the crowd, click through to the main site. Those that don’t either bounce or hang out and pick a few stocks in the Stockmoose game. Either way–our “crowd” results get better and we have an opportunity to pitch the main site.  Win win.

Another day, another kooky web idea. This time, it’s the Stock Moose. We’ve had a lot of debate at work on how to gather data, how to present that data, and how to make that collection/presentation process engaging enough that someone might actually enjoy doing it just for the sake of doing it (rather than tying it to some future promise of “Oh, we’ll give you an edge on the trading room floor”). Back when the college football season was in full swing, Yahoo! introduced what they called the Team Ranker. The concept is incredibly simple: pick two teams out of a hat and display them both (with perhaps a few bits of useful info such as a win-loss record or… a stock chart). The user simply has to click on the one they think is better. Period.

Whichever team (or in this case, stock) has the best win percent is rated as #1, and the rest are sorted accordingly. The system was far too simple to game, at least for football. What this usually meant is that earlier in the day (when the East Coast is awake and voting) highly ranked teams in the SEC and other eastern conferences dominated the rankings. When it got later in the afternoon, teams with East Coast fan bases slipped in the rankings while PAC-10 teams rose into the top spots. This is unavoidable for something like football, where fans are fiercely loyal to their own teams over all others–but is the same true for stocks?

Would users on the East coast sway the list towards East-coast stalwarts like Coke and Home Depot while the West coast might favor silicon valley darlings over all others? It’s hard to say–especially considering that I limited the field to the Nasdaq 100, which is primarily dominated by tech stocks.

The beauty of today’s technology, though, is that these debates don’t have to remain purely academic. The Stock Moose is thin. It’s very, very thin. But rather than spend a few hours swapping emails back and forth with my bosses on whether or not something like that would even work, tools like Heroku make it possible to just go build it. I spent a grand total of around four hours kicking that thing out, and probably half of that time was spent trying to decide on a name for it (and then trying to explain to my girlfriend why I needed her to draw me a moose with a tie). The other half was spent cooking up the round-robin logic, which is probably way too complicated. Building something like this within our own code base–or even hosting it on our own servers without asking permission–would be unthinkable.

If we ultimately decide it’s a failure (which most little web projects are), what’s the cost? There’s always a chance, though, that something so simple (and arguably, so ugly) might go viral. If that happens–and that’s a big IF–what do we get? A little free advertising. If it doesn’t happen, we’ve still profited: we can now cross something else off our list and move towards a more targeted product.

Besides, Stock Moose has a nice ring to it.

The sign-up process is a major chore.  I’ve never done it at another broker, so I don’t know if it’s this bad everywhere or this is a “you get what you pay for” scenario.  I actually filled out all the paperwork back in August, but I never got around to mailing it off and faxing them my info (yes, MAIL and FAX–I didn’t realize anyone still used either).  I got a little more motivated last week, so I checked back in with them.  Because I hadn’t mailed my paperwork in by their 30 day deadline, I had to fill everything out again from scratch–there was no record anywhere of the application I’d previously filled out.

Once I got everything filled out, I headed over to Fedex/Kinkos, figuring I could make copies of my driver’s license, fax them the sheets they required, and mail everything in one stop.  No such luck.  Their mailing address for paperwork is a P.O. Box, so after standing in line for half an hour (December, go figure), they told me I’d have to take it to the post office.  After standing in line again at the post office, I finally had everything sent off properly.

I got the email saying the account had been a few days later–and that I could go to the site and log on using a combination of random personal info and the last 4 digits of my account number.  I tried to do so, but I didn’t actually have an account number yet.  When logging on to their site, it just said “Roth IRA–Account set up in minutes.”  It said that for a few days.  After it finally went through, I was able to log in and take a look at their trading platform.  In a word: ugly.  It looks like it was designed by programmers, with no real care paid to user experience.  I’ll hold further comment on it until I’ve actually used it some (though if the ACH-setup interface is any indication, it’s going to be a major chore).

Then my next issue: management fees.  Somewhere in the paperwork I filled out, I remember reading  that they charge you $30 a year to manage your retirement account (though, as far as I can tell, it’s still up to me to make all my own trades and pay trading fees of $4.50 until my balance hits $2500).  The yearly fee hits in February, and if you fund your account after February they charge you immediately.  It’s now December.  I called their customer support (and, to their credit, actually got a real person within 5 minutes) and asked if this fee was pro-rated or I was really going to be charged $30 so they could manage the couple of hundred or so dollars I’m likely to put in before February.  Their response: full $30.  I asked them if it was a specific date in February (I’d hate to put money in Feb 1st only to get charged again Feb 10th) or if I could safely put money into the account any time in the month.  He didn’t know.  He said it was usually a certain week in February, so if I wanted to be absolutely sure it was best to wait until March 1st.
So I’ve got a Roth IRA now.  I’m just not going to put any money into it until March. That’s not so bad, I guess.  I’ll just have to do a good job of mentally earmarking money in my checking account as “IRA money” and not “shiny new toy” money.  The game I mentioned in my Thanksgiving post got a really great sponsorship offer, so once I polish it up to their specs and finalize the deal I should have a pretty nice chunk of change to act as seed money (though I still don’t see myself maxing it out for 2007).

TypeError: Error #1006: getInstance is not a function…

Figuring that there was some sort of reference to something internal, I started pulling code out. A lot of code out. Eventually I stripped out all of the Meta, all of the Script code, and was left with pure MXML… with anything not core (i.e. function calls) also dumped. It still wouldn’t work. Eventually I had it stripped to the following, and that’s when the “DUH” gods struck:

Of course it should throw an error–I hadn’t wrapped my Panel in any sort of application. I reverted back to the stock Adobe code, made another mxml file with an application in it to load the hacked-RTE, and it worked like a charm the first time out. Because I went into the conversion thinking it would take some nudging to work, I missed my own silly error.

I guess it could be worse, though–it’s not like I have to write my own RTE component from scratch.

The selection quirks are indeed fixed in the new version, but there were two more bugs in the ToggleButtonBar code. The logic for whether or not to select or deselect a particular button happens in the function “hiliteSelectedNavItem,” which is called after the clickHandler. So what? This means, from an itemClickEvent perspective, that the ToggleButtonBar will NEVER generate an event where the selectedIndex = -1. Why don’t I just check the TBB’s selectedIndex manually instead of trusting the events? Well, the handler for the event is actually getting called before the TBB’s hiliteSelectedNavItem fires, so as far as my event handler is concerned the selectedIndex can never be -1.

As I want the deselects to turn off paragraph styles, this is a bad thing. Luckily, it’s easy to fix. In the clickHandler function, you need to change it so that if you’re turning off a button it dispatches that event rather than letting the parent NavBar fire of its simpler “You clicked on Button #4″ event:

Previous version:

if (_toggleOnClick && index == selectedIndex)
hiliteSelectedNavItem(-1);
else
hiliteSelectedNavItem(index);
super.clickHandler(event);

if (_toggleOnClick && index == selectedIndex)
{
selectedIndex = -1;
hiliteSelectedNavItem(-1);
var newEvent:ItemClickEvent = new ItemClickEvent(ItemClickEvent.ITEM_CLICK);
newEvent.label = "none";
newEvent.index = -1;
newEvent.item = null;
dispatchEvent(newEvent);

event.stopImmediatePropagation();
}else{
hiliteSelectedNavItem(index);
super.clickHandler(event);
}