tilt: simple human psychology with regard to recent loss/gain vis a vis risk.
The claim that tilt is an "emotional state" in the original article, may not be a wholly accurate one.
Rather, simple human psychology regarding recent loss or gain vis a vis risk.
(from 2+2's magazine originally,
here):
The Science of Tilt
By Jay McCauley
The author would like to thank Jared Lunsford for his extensive and very helpful commentary on an earlier version of this article.
Caught in the steel trap of a recalcitrant tilt, one feels motivated by an otherworldly force - a demonic body-snatcher, maybe - to pursue the humiliating anti-goal of losing as much as possible as quickly as possible. Tilt is powerful and strange, yes, but it is
not mystical. Many of the central behaviors that we associate with tilt can be readily understood in light of the Prospect Theory of Daniel Khaneman and Amos Tversky.
Prospect Theory is one of the most important and well-established contributions to economics in history. In fact, Khaneman earned the Nobel Prize in 2002 for his part in developing and extending Prospect Theory1. Given the significance of Prospect Theory to economics, behavioral finance, psychology, and related disciplines, it is somewhat surprising that gamblers are not generally aware of the theory. A 'prospect' is a possibility or gamble, and any serious gambler should be familiar with a '
gambling Theory' worthy of the Nobel.
What is 'tilt'?
Ask a poker player to talk about tilt and he will probably mention emotion as an intermediate cause. He might tell a story along the lines of "Tilt is when a player takes a really bad beat, and then spews chips out of anger. He tends to play more marginal hands, he makes weak calldowns and hopeless bluffs, and he just makes bad decisions in general. Sometimes he is angry about getting bluffed and lets his desire for revenge overwhelm him. Other times he arrives at the table upset and begins playing poorly because of it."
This description gets just enough right to make it a dangerous misunderstanding.
Yes, it is true that emotion can interfere with one's ability to make a good decision. Yes, it is also true that bad beats, getting bluffed, etc. tends to make a person emotional. So it would seem to be a relatively simple ABC chain of events: Bad beats lead to anger leads to bad decisions.
This line of thinking ignores the possibility of a direct "bad beats lead to bad decisions" causal chain that requires no emotion. The distinction is subtle, but there are dramatic consequences to whether emotion is a necessary aspect of tilt. As I will show, Prospect Theory provides substantial reason to think that bad beats (among other things) will in fact lead directly to tilt. This is because bad beats involve a large perceived loss, and
perceived loss tends to make people more risk-seeking in an effort to recover the loss (irrespective of emotion).
Why would it be bad to assume that emotion is both a sufficient and
necessary cause of tilt?2 First, if emotion went hand-in-hand with tilt, then introspecting about one's emotional state could provide reliable information about tilt potential. However, while a gambler might legitimately infer that he is on tilt because he is angry, he cannot know that he is tilt-free simply because he is unemotional.
A second reason to avoid focusing on emotion as a cause of tilt is that this could erroneously lead one to believe that if emotions like anger were controlled, then tilt might be prevented. Again, while it is true that controlling anger might reduce tilt, it is by no means a guarantee because the natural tendency to tilt after a large perceived loss will remain. I read on a well-known poker professional's blog that he is taking adrenaline blockers in an effort to prevent tilt. While this might be useful (and it might be harmful), the natural tendency to tilt in certain circumstances will likely be unchecked.
A final reason it is bad to think about tilt strictly in terms of emotion is that by missing a fundamental insight of Prospect Theory - namely, how
subjective perceptions of loss and gain influence decisions in particular ways - a deeper understanding of tilt is impossible.
Tilt defined
As I have argued, a bad definition of tilt would assume that a person on tilt is necessarily emotional. By contrast, a useful and encompassing definition of tilt will stick as closely as possible to the observable behavior of someone on tilt, without making unnecessary assumptions about the underlying process. I offer the following:
Tilt is when a gambler who wants to make the most money he can over the course of a gambling situation employs a wagering strategy other than the highest expected-value pattern of wagering he generally has the ability to adopt for that situation.
Prospect Theory
Now I will introduce Prospect Theory and the connection to tilt by way of an example. Consider the following problem from Khaneman and Tversky's research:
PROBLEM I: Imagine the US is preparing for the outbreak of an Asian disease, expected to kill 600 people:
- If program A is adopted, 200 people will be saved
- If program B is adopted, there is one third probability that 600 people will be saved and two thirds probability that no people will be saved
Think for a moment about which program you would prefer.
When research participants are asked this question, 72% prefer program A and 28% prefer program B. Since the expected-value of program A and program B are equivalent, participants who choose program A can be described as
risk-averse, while participants who choose program B are
risk-seeking. Thus, on average people seem to be
risk-averse by a 3:1 margin for this particular problem. There is nothing wrong with any of this --
yet. Risk-seeking is not inherently more rational or less rational than risk-aversion.
Consider a different problem:
PROBLEM II: Imagine the US is preparing for the outbreak of an Asian disease, expected to kill 600 people:
- If program C is adopted, 400 people will die
- If program D is adopted, there is one third probability that nobody will die and two thirds probability that 600 people will die
Faced with this problem, 22% of research participants prefer program C and 78% prefer program D. Thus, in problem II research participants seem to be
risk-seeking by a 3:1 margin. Take a look back. Program A is
structurally identical to Program C, just as program B is structurally identical to program D. Only the framing of the programs is different (i.e. stating that 400 of 600 will die rather than stating that 200 of 600 will be saved). So why do the vast majority opt to gamble in the second problem but not the first?
According to Prospect Theory, as people regard themselves to have made gains, they tend to be less willing to risk losing those gains in exchange for further gains (i.e. they become more risk-averse). This is because subsequent gains bring less satisfaction than previous gains. Conversely, as people regard themselves to have accrued losses, they tend to be more willing to risk further losses in exchange for the opportunity to have lost less (i.e. they become more risk-seeking). This is because further losses don't hurt as much as the initial losses did. In many research studies, the effect of losses on decisions is about twice as strong as the effect of gains. The figure below is a graph of utility (i.e., happiness/satisfaction) by money that illustrates the classic S-curve predicted by Prospect Theory.
So how does Prospect Theory specifically explain the Asian-disease problem? Participants who were asked to choose between 200 lives saved and the gamble to save everyone (or nobody) were evaluating the decision relative to a sure gain of 200. Since every extra life saved beyond 200 meant a little less than the previous life saved, participants generally declined to gamble (i.e. for the majority of participants, saving 600 lives is worth less than three times as much as saving 200 lives).
Participants who were asked to choose between 400 lives lost and a gamble to save everyone (or nobody) were evaluating the decision relative to a sure loss of 400. Now, every extra life saved meant a little MORE than the life before, and subsequent lives lost meant less and less. This explains the stark reversal in which the majority of participants now risked losing everyone for the sake of 'getting even.'
As a little test before continuing, use the graph above to see if you can determine mathematically if someone who regards himself to have made $50 would gamble that money on a coin-flip that pays even money. Now, what about someone who perceives herself to be down $50? Would she take the coin flip?
In addition to illustrating the elements of Prospect Theory, the Asian-disease problem holds a few other lessons that are important for gamblers:
- Emotion. I think it would be far-fetched to assume that the participants who gambled everything when 'down 400' in the example were more emotional than the participants who were making the decision with respect to a gain of 200. The perceived loss is what counts (there are countless other research examples that do not seem to involve emotion).
- Perception of gain and loss is subjective. The Asian-disease problem illustrates that an objective set of facts can be variously framed so that a gain or loss is perceived. The accrual of gains and losses reside in the mind, not the wallet. This is a powerful concept, and a key to understanding and manipulating opponents.
The contrast between real dollars and Sklansky dollars
A naïve gambler doesn't know probability theory - she doesn't understand the mathematical relationship between risk and reward that allows one to make a decision that is profitable over the long-run. A more sophisticated gambler, by contrast, will have some understanding of how to 'get the best of it.' But while the sophisticated gambler might be good - perhaps very good - at
making decisions in terms of "Sklansky dollars" (long-term profit), she is sometimes very bad at
perceiving results in terms of Sklansky dollars (note that perceiving results and evaluating results are two very different things!).
For example, suppose you flop the nuts in no-limit Texas hold 'em ($1000 effective stacks), get all-in on the flop, and lose to someone who needed to hit specific cards on turn and river. How would you perceive your gain/loss? Do you feel "up $1000," because that was your Sklansky profit? Or, as is probably much more common, do you primarily perceive your real $2000 loss?
Now you might be thinking "OK, but if I get all-in preflop with AK vs. QQ in no-limit Texas hold 'em, I'm not nearly as likely to tilt as if I lose to a bad beat like that. I lost the same amount in either case so why should the former be more likely to put me on tilt?" The answer is that just as someone who sees that he is a huge favorite has already 'mentally pocketed' the real dollars as his own, a person who knows that he is 50/50 cannot reasonably think 'this pot is mind'. That is, he often remains focused on the Sklansky profit (0) even when he loses, because it follows directly from how he first conceptualized the situation.3
While many gamblers can absorb a single large real (but 0 EV) loss like this in stride, something curious often happens with a string of such losses. At some point a gambler's focus shifts from the accrued Sklansky loss ($0) to the real monetary loss (say, $4000). Think about what this means: A gambler goes immediately from perceiving he has lost nothing to perceiving a loss that is many times greater than anything he could lose in a single hand. Prospect Theory predicts that a dramatic shift toward more risk-seeking behavior would generally follow.
Gambling tendencies, tilt potential, and ways to control tilt
Consider a gambler who makes every wagering decision on the advice of a magic eight-ball. His perception of gains and losses - and, thus, his tendency to risk-seeking or risk-averse behavior - is irrelevant, because he is effectively making no decision at all beyond the initial choice to use the eight-ball in this way. Similarly, to the extent that a gambler's system of wagering is rigid and rule-based, he is less likely to tilt. This is because he is effectively following through on a decision previously made (i.e. the decision of which system to adopt).
Conversely, the more flexible and intuitive a system of wagering, the more opportunity exists for perceived losses and gains to influence decision-making in accordance with Prospect Theory. Consequently, this sort of gambler should generally be very wary of tilt, and take concrete steps to mitigate it (particularly if tilt has been a problem in the past).
One way to diagnose potential tilt is to introspect about your perceptions of how much you have made or loss (rather than introspecting about how you feel about your gains/losses). To the extent you perceive yourself to be down, and to the extent that your gambling methods give you room to let risk-seeking tendencies influence your wagers, you are more likely to tilt.
Other ways to control tilt besides merely introspecting about tilt potential include strategies to take the focus off of money altogether, such as playing a certain number of hands (perhaps make a wager with a friend from 2+2 on who can put in more hands in a given day/week/month), or keeping a log of mistakes and great plays. If tilt is a big problem for you then I suggest a 'hard stop-loss' (i.e., keep a limited amount of money with you at the casino or online and limit your ability to access further funds beforehand).
Conclusion
Poker players have long known that gamblers tend to sit on gains and chase losses. Mike Caro, for example, provides a discussion of such behavior in his Super System article. I hope, however, that you now understand that this tendency is simply the exaggerated form of a behavior pattern that is common when people make decisions under uncertainty across many different domains. Though poker often feels like a universe unto itself, it is really just a microcosm of the decision-making situations we generally face in life.
1 Tversky would have shared the prize, but he died in 1996 and the Nobel is not awarded posthumously.
2 For the purpose of these articles I'll assume that emotion can be a sufficient cause of tilt, though this may not be the case. For example, Damasio's research concerns ways that emotions can improve decision-making and he has used gambling situations to make his case.
3 Interestingly, I think when people win coin-flips such as these their attention tends to go to the real gain, so that they often become more risk-averse after winning.