more risk = bigger losses

Friend of the blog and Cornell grad student, Kyle Siler, has research that has been covered by Time magazine. Let that sink in for a moment: Time. Friggin. Magazine. In his research, he studied gazillions of internet poker games and found a fascinating result: the more hands you win, the less money you win.

The reason for the paradoxical results was straightforward enough: the majority of the wins the players tallied were for relatively small stakes. But the longer they played — and the more confident they got — the likelier they were to get blown out on one or a few very big hands. Win a dozen $50 pots and you’re still going to wind up far behind if you lose a single $1,000 one. “People overweigh their frequent small gains vis-à-vis occasional large losses,” Siler says.

Siler applies this risk taking to everyday life, too:

Investing, driving, buying a house and merely crossing the street are all acts that involve discernible risks and uncertain rewards. The more small returns you get from your small investments in stocks, the likelier you are to make — and lose — a big investment. The more times you get behind the wheel and speed a little bit, the likelier you are to speed a lot — with deadlier consequences.

“These kinds of calculations are made every day,” says Siler. “Adultery is another good example. People get away with it countless times but they get caught just once and they lose everything.”

Nicely played, Kyle.

h/t: Contexts Crawler

20 thoughts on “more risk = bigger losses”

  1. Hey Tina,

    Thanks for the kind words; I guess our “15 minutes (kilobytes?) of fame” occur on the internet now! The academic in me is dealing with the oversimplifications necessary to communicate research through the media. Despite the years I spent trying to pull off that project, a bunch of stuff I said about adultery, Enron and Iceland’s recent bankruptcy (the latter two examples in particular may have been stretching the analogy a bit too far) that I came up with on the spot while on the phone with a journalist asking me to “riff on the real-life implications” ended up being a big focus in what diffused through the media. I guess that’s the nature of the beast.

    …and now back to Stata, Excel, Word and my regularly scheduled dissertation…


  2. The analogy with risky driving (while texting, while drunk, etc.) may be better than analogy with adultery (I claim no first-hand experience in either area). It’s not just that the odds eventually catch up with you. Most drivers learn that they can get away with driving while texting or after drinking. Their diminished ability becomes crucial mostly when something unexpected comes up suddenly.

    In online no-limit hold ’em, apparently people get a lot of experience playing hands that aren’t very competitive and don’t involve big sums. When the game suddenly jumps to the unusual hand — big money, competitive cards — their lack of the necessary experience causes them to crash. At least, that’s my guess.


  3. That’s an interesting point, Jay. There are several threads to it. Is there any value in distinguishing diminished capacity, inexperience and lowered sensitivity to risk? In your analogy the drunk/texting drivers have all three, but the poker players only have the second and third.


  4. Carl, Here’s what I had in mind. Drivers do things that they’ve heard are risky – use their phones to talk or text, drive after drinking. Most of the time, they do so without incident (or accident) because the drive is not demanding. No surprises. Phones and liquor detract from the driver’s attentiveness, but most trips don’t require high levels of attentiveness. Each successful trip makes them overrate their ability and discount the possibility of events that do demand more attentiveness.

    You’re right that in my model, the poker players don’t have “diminshed” capacity. They never did have the chops that the better players have. But like the drivers, their experience in less challenging games leads them to overrate their abilities. I don’t play much poker and have never played online, but I’d guess that in no-limit games, the crucial hand pops up on them unexpectedly. The trip starts just like the others, with low amounts, but then suddenly, some other player is pushing in a huge stacks of chips. Evaluating your hand when the bet is $500 may require a different calculus from what you use when the bet is $50. And the player who lacks experience in big-pot hands may not be able to make the correct evaluation.

    That’s my guess. But Kyle Siler would be the one to ask.


  5. @ Carl & Jay

    Interesting stuff. In my study, I theorized that 1) the asymmetric incentive structure (frequent small wins, outweighed by occasional big losses) and 2) not understanding the existence and logic of Pyrrhic victories, were what appeared to hurt losing players.

    Crimes like shoplifting and DUI might fall into an analogous incentive structure, because not only do you get away with it more often than not, you are positively reinforced with each success, which leads to further entrenchment of that strategy/viewpoint. This can lead to confidence, sloppiness and further exposing your strategy to shrewd observers.

    Behavioral economics has shown that people have difficulty weighting occasional large losses vis-a-vis frequent small wins, in part because humans tend to perceive risks and gains differently. So, it appears that for whatever reason, people (or at least, losing poker players) tend not to evaluate their risks and rewards appropriately or proportionally. This also dovetails with Zelizer’s work on the social construction and perception of money. Further, you get into Kruger and Dunning (1997) territory, where people are not only incompetent, but people also lack the skills to realize that they are incompetent (here’s where you can imagine a certain colleague or neighbor).

    Risk and uncertainty are both really complex things, especially when brought into ‘real’ social realms. Poker provides a relatively pure example of people making strategic decisions under conditions of risk and uncertainty. The stock market doesn’t seem to create the ‘behavioral text trail’ footprints that exist in online poker, so I figured I’d take a gander at poker. I know economists are beginning to study crime and punishment using these frames, and as a sociologist, I tend to be skeptical of the over-application of economic behavior into social realms. However, these sorts of calculations (as ideal types, at the very least) infuse our lives daily, so I think they’re very interesting and worth considering.


  6. So this is an extension of Kahnemann and Tversky? Is there are paper I can access? I’m not allowed to read such “magazines”, they aggravate my Tourrette’s syndrome….


  7. Just shooting from the hip before I click through to the new links, would it be fair to say that risk assessment is a particular kind of skill subject to learning (and mislearning)? It seems to me when we talk about size of risk we’re not talking about absolute size but about the magnitude of possible costs and benefits in relation to affordability. I would expect an expert to be aware of this, calculate accordingly and devote a certain amount of her attention and effort to managing risk, for example through insurance, proxy, or becoming too big to fail.

    Oddly enough I have a different sort of example from my recreational tennis league. Inexpert players are far more likely to go for broke with big low-percentage shots early in points. Better players in contrast hit a series of quality high-percentage shots designed to work the situation toward forcing an opponent error or creating a high-percentage winner. (This crafty old guy told me recently that he counted to three shots in a rally and hit a deep, central floater because he knew his opponent would go for something stupid and usually miss at that point.)

    Now the thing is, you might say that the weaker players should adopt the stronger players’ strategy. But if your game is not reliable enough to choreograph a five-shot sequence and carry it off, you may be facing low odds either way in which case going for it early isn’t automatically stupid.


  8. Carl (@12), your tennis example echoes the old experiment by David McClelland (I think Gladwell mentions it). He had children play a ring-toss game. Each kid could choose how close to the target he stood. Kids who scored high on McClelland’s “achievement” measure (presumably the more entrepreneurial types) chose a middle distance. Low achievement kids stood either so close they couldn’t miss or way beyond the 3-point line.

    In my tennis-playing days, I used to notice (and sometimes play against)guys who went for the devastating winner on nearly every shot. When one of those screamers went in, it was beautiful. But I’d think, “Stick with this kid. He’s a loser.” And most of the time, they wound up beating themselves. Then they’d get angry at having lost to an inferior player.


  9. I was like a children’s beauty pageant contestant there. I had been smiling for the past five minutes and my face muscles were exhausted, so the photographer told me to give him my best “poker face.” In the end, he chose that photo.

    I just hope the first impression I make of future Googlers of my name or image isn’t one of ‘sourpuss’, ‘robot’ or ‘curmudgeon!’


  10. Kyle, maybe this is a dumb question but is the whole Geertz ‘deep play’ dynamic relevant to your findings at all? When I’ve played hold’em, which is not a lot and always ‘socially’, there have seemed to be two kinds of players – those who were using the game to signal things about themselves to each other, and those whose whole focus was winning. The former regularly made extravagant bets on poor odds, made a big show of going all in, and bought back in cheerfully when they lost; the latter folded hand after hand until they saw cards they liked, then pounced, all with the grim intensity of an accountant looking for a tax loophole.

    If the play is deep, that is, there’s a whole social dimension to it in which the money is a means of signaling rather than an end in itself, it may be that the patterns you’re looking at are not at all suboptimal with respect to those different ends?

    Btw, how could someone with the screen name Isildur not lose in a big splashy way?


  11. @ Carl

    Interesting ideas. I think even casual poker games can reveal a lot about people’s personalities/social inclinations on the one hand, and attitudes towards risk and uncertainty on the other. When the stakes are low, I’d say the former is more present; when the stakes are higher, the latter.

    One of the things I found that didn’t diffuse through the media, is that ‘wild’ ‘hyper-aggressive’ play starts to become more profitable at higher stakes games ($1000US buy-in was the highest I looked at). This may be due to the fact that players at this level are more skilled, and can successfully play this riskier and more difficult strategy. Alternatively, when stakes get higher, games of chicken between players become scarier, and conservatively swerving away from profitable risks becomes more tempting and exploitable.

    In casinos and online, it has surprised me how cavalierly some “ATM Machines” keep playing loosely and buying in repeatedly. As a sociologist, it makes me wonder if these people are 1) wealthy, 2) happily spending their discretionary income, or 3) have a problem. It’s hard to tell on the surface. However, that cavalier attitude is also a part of what makes most great poker players or gamblers in general. The line between ‘degenerate gambler’ and ‘wizard of odds’ can be a thin one, especially when that line is being pushed around by the vicissitudes of luck or “noise” in a game involving randomness and chance, like poker.


  12. Kyle (@19). The “ATM Machines,” whether they can afford it (#1 and 2 in your typology) or not (#3) are like the tennis players Carl and I mentioned (@12, 13). They also probably are thinking about the big wins they’ve had, and they imagine that they’ll have more. They don’t think so much about the overall losses. There’s some cognitive distortion here.

    But there’s also accurate cognition. Once a player falls significantly behind, high-risk bets may be the only way to get even. Logic, past performance, and probability all say that his best strategy would be to walk away from the table. But if he does that, he gives up all hope of getting even. On the other hand, if he stays in the game, yes he might lose some more, probably will lose more, but he also has the possibility of catching a killer hand. And he knows it’s possible because he’s seen it happen.


Leave a Reply

Please log in using one of these methods to post your comment: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.