The presentation demonstrates how cognitive biases is detrimental strategic decision-making.
In particular - it illustrates how poker players perform better by avoiding the availability and representativeness bias.
Finally it illustrates some advice on how to avoid these cognitive errors and improve your decision-making! :)
How Cognitive Biases Impact Poker Ability and Decision-Making
1. Lasse Ringstad The Costs of
Irrationality
The cost of irrationality
Cognitive biases – the difference between good and great
decision-makers
How poker players perform better by avoiding the availability and representativeness bias
For those who prefer the full academic article:
http://www.slideshare.net/lassebringstad/the-costs-of-irrationality-v17-article
2. -9
-8
-7
-6
-5
-4
-3
-2
-1
0
1
2
3
4
5
6
7
8
9
10
0 1 2 3 4 5 6 7 8 9 10
Self-assessed poker ability on a scale from 1-10 (10 being best)
Propensity to cognitive biases (negative more biased)
N=338
R=0.18
Individual propensity to cognitive biases is inversely
correlated with poker ability...
4. Since you are still here – I assume I have your attention!
Let's kick it off with a small task to keep you focused...
At some point in the presentation you will be asked some questions
about the following participants at a recent London fundraiser. Please
pay attention to the attached list of celebrity attendees:
• Angelina Jolie
• Brad Pitt
• Brian Cox
• Cameron Diaz
• Emma Watson
• Jeremy Hunt
• Jim Broadbent
• Kate Middleton
• Kiera Knightley
• Karl Simmons
• Kim Kardashian
• Meryl Streep
• Noel Clark
• Prince William
• Rihanna
• Rufus Sewell
• Stephen Fry
All finished? Good – let's move on!
If you don't feel like thinking – you can skip directly to slide 8. (Although I do believe
the presentation is more powerful if you take the time to perform the tasks properly...)
5. Let's talk about Linda
Linda is thirty-one years old, single, outspoken, and very bright. She majored in philosophy. As a student, she
was deeply concerned with issues of discrimination and social justice, and also participated in antinuclear
demonstrations.
Please rank the following statements by their probability, using 1 for the most probable and 5 for the least
probable.
A) Linda is a teacher in elementary school
B) Linda is a bank teller
C) Linda works in a bookstore and takes yoga classes
D) Linda is an insurance salesperson
E) Linda is a bank teller and is active in the feminist movement
Linda
Come on – give the
question a good go
before moving on!
6. Let's talk about Linda
Linda is thirty-one years old, single, outspoken, and very bright. She majored in philosophy. As a student, she
was deeply concerned with issues of discrimination and social justice, and also participated in antinuclear
demonstrations.
Please rank the following statements by their probability, using 1 for the most probable and 5 for the least
probable.
A) Linda is a teacher in elementary school
B) Linda is a bank teller
C) Linda works in a bookstore and takes yoga classes
D) Linda is an insurance salesperson
E) Linda is a bank teller and is active in the feminist movement
64% of the
respondents in my
survey think that E
is more probable
than B. Did you
agree?
If so – there is something inherently wrong
about that answer. E is a conjunction of B (B
and something more) – so B is true
whenever E is true. And hence B is more
probable than E.
Probability that
Linda is
active in the
feminist
movement
E) Probability that Linda is a
bank teller and active in the
feminist movement.... As
you can see the probability
of E must be smaller than
the probability of B...
B) Probability
that Linda is a
bank teller
Linda "sounds more like" a feminist bank teller than
simply a bank teller. So we mistakenly mix
representativeness with probability. It is an example
of suffering from the representativeness bias.
E)
Linda
Representati-
veness
bias
Why we
are fooled
7. All right, but with Linda I just misunderstood the question!
Cool – let's try two more questions
Without looking back at the celebrities that attended
the fundraiser in London.
Please give your best indication of the number of male
and female celebrity participants at the event:
_____________________
Which cause of death is most common in the United
States?
a) Lightning b) Tornado
Linda
8. All right, but with Linda I just misunderstood the question!
Cool – let's try two more questions
Without looking back at the celebrities that attended
the fundraiser in London.
Please give your best indication of the number of male
and female celebrity participants at the event:
_____________________
Which cause of death is most common in the United
States?
a) Lightning b) Tornado
9 men and 8 women
Did you find yourself thinking that more women than men attended the event? So did 55% of the
individuals in my survey...
And did you think Tornado's were the most common cause of death? So did 84% of the respondents in
my survey...
We are fooled by a heuristic we call the availability heuristic.
- It is easier to recall the female participants since the made-up fundraiser had several famous women, and
several not-so-famous men.
- Similarly it is easier to imagine and remember Tornados killing people than Lightning. So we think they are a
more common cause of death....
It is an example of suffering from the availability heuristic.
• Angelina Jolie
• Brad Pitt
• Brian Cox
• Cameron Diaz
• Emma Watson
• Jeremy Hunt
• Jim Broadbent
• Kate Middleton
• Kiera Knightley
• Karl Simmons
• Kim Kardashian
• Meryl Streep
• Noel Clark
• Prince William
• Rihanna
• Rufus Sewell
• Stephen Fry
Celebrity attendees
Linda
Availability
bias
How we
are fooled
9. So what are these heuristics?
They are the mental shortcuts we apply in our daily lives
...so we simplify our choices by applying mental
"rules of thumb"
In our daily lives we face an incredible amount
of choices....
"I always start my night-out with a beer"
"I never play for an inside straight"
"I don't drink coffee after 6 pm"
10. And these mental "rules of thumb" usually work very well
"I always start my night-out with a beer"
Usually a good choice since
you have experienced that
wine makes you tired and
spirits makes you drunk too
early
"I never play for an inside straight"
Usually a good choice since
chasing that inside straight
will normally not pay off
"I don't drink coffee after 6 pm"
Usually a good choice since
that coffee will make sure
you don't sleep well
11. But sometimes heuristics lead to systematic biases in our
decisions...
"Individuals tend to pick the default
option"
Tom does not save enough for
retirement since the default savings
rate is too low to cover his future
needs
"Individuals tend to love whatever is
for free – even if it is not rational"
So we get tricked into buying two
bananas when the grocery store
states "Buy 2, get 1 for free!", even
though we never eat more than 1
banana...
"Individuals think that events that can
be easily remembered are more likely
than less vivid events"
Mom is mistakenly more worried
when I go out for a flight then
when I go out for a drive
12. So why does all this matter in real life?
Awareness of our cognitive biases can improve our decision-making!
My thesis: Individuals suffering from cognitive biases make poorer choices in
strategic situations than individuals not suffering from these biases. And these
biased individual are consequently perform worse in strategic-decision
making situations, such as negotiations, business or poker.
We should not always question our intuitive judgment. Because as we
can see heuristics can be very useful!
But I believe we should be aware that our heuristics and judgment
sometimes fail us. If we become more aware of how heuristics fail us –
we will become better decision-makers. Which in turn means we can
become better businessmen, negotiators or poker-players!
Heuristics
can be
useful...
...but I think
awareness
will make us
better
decision-
makers...
...thus I
formulated
my thesis
13. Method: To test the thesis I mapped individual propensity to
cognitive biases & correlated the results with poker ability
...and compared each individual's propensity to
cognitive biases with his/her online poker
performance
I administered a psychological survey to online
poker players...
• 338 poker players provided complete responses!
• 11 questions on Linda, Tornados and celebrity
fundraisers created a mapping of the degree to
which individuals suffered from cognitive biases
• Specifically 5 questions on the representativeness
bias – or the Linda bias
• And 6 questions on the availability bias – or the
fundraiser bias
• Poker has been compared to the strategic
dilemmas individuals face elsewhere in life and
business
• Player´s success depends upon chance, his risk-
return strategy and his social judgment
• So poker is very similar to all other forms of
strategic decision-making – be it in business or
life in general
• Tracking of online poker performance available
from sites like sharkscope.com!
14. Definitions: The representativeness & availability bias
1. In essence I can actually measure the poker ability of each individual in my sample based on online-ranking sites such as sharkscope.com. For further information on my
methodology please refer to the methods section in my final paper (link can be found at the end of this presentation).
Note: this is not an exhaustive list of biases, and I do believe that
testing for more biases would have led to even stronger results
The representativeness bias – or the "similarity
heuristic"
The availability bias – or the "ease of
recollection or imagination bias"
The availability heuristic, or the celebrity bias, is our
tendency to assess the probability of an event by
how easily it can be remembered or imagined
Example – tornados & celebrities
• Most people incorrectly believe tornados to be a
more common cause of death than lightning,
since tornados are easier to imagine and
remember
• Individuals consistently believe that more women
attended a male-dominated fundraiser after being
showed a list of participants; including very
famous women and not-so-famous men
The representativeness heuristic, or the Linda bias,
can be thought of as a similarity heuristic, and is our
tendency to assess the probability of an event by
how similar it is to its parent population.
Example – Linda the feminist
• Individuals consistently rate the statement “Linda
is a bank teller, and active in the feminist
movement” as more probable than the statement
“Linda is a bank teller” after reading a fictional
personality sketch of Linda, depicting her as
someone representative of a feminist
15. Result I: Individual propensity to cognitive biases is inversely
correlated with poker ability!
-9
-8
-7
-6
-5
-4
-3
-2
-1
0
1
2
3
4
5
6
7
8
9
10
0 1 2 3 4 5 6 7 8 9 10
Self-assessed poker ability on a scale from 1-10 (10 being best)
Propensity to cognitive biases (negative more biased)
The results unequivocally confirm that suffering from
cognitive biases makes you a poorer decision-maker!
Due to data limitation on online poker tracking1 , self-
assessed poker ability is used as a proxy for actual poker
ability for most of the graphs (the correlation between self-
assessed poker ability and actual online ability is ~0.4)N=338
R=0.18
1. Pokerstars decided to ban online tracking tools on their site – thus effectively eliminating 2/3 of my sample. For further information on my methodology please refer to the
methods section in my final paper (link can be found at the end of this presentation).
16. Result I – backup: The results are supported by smaller
subset where actual online ability could be measured
Backup
-9
-8
-7
-6
-5
-4
-3
-2
-1
0
1
2
3
4
5
6
7
8
9
10
50 52 54 56 58 60 62 64 66 68 70 72 74 76 78 80 82 84 86 88 90 92 94 96 98
Sharkscope assessed poker ability on a scale from 0-100 (100 being best)
Propensity to cognitive biases (negative more biased)
N=89
R=0.17
The correlation is roughly the same as for self-
assessed poker ability at 0.17 vs. 0.18
17. Result II: Suffering from either of the cognitive biases
negatively affects your poker performance
The inverse correlation between the availability
bias and poker ability is 0.11...
...but the representativeness bias is more
detrimental to ability with a correlation of 0.16
-10
-8
-6
-4
-2
0
2
4
6
8
10
0 1 2 3 4 5 6 7 8 9 10
Self-assessed poker ability on a scale from 1-10 (10 being best)
Propensity to the availability bias
(negative more biased)
-10
-8
-6
-4
-2
0
2
4
6
8
10
0 1 2 3 4 5 6 7 8 9 10
Propensity to the representativeness biases
(negative more biased)
Self-assessed poker ability on a scale from 1-10 (10 being best)
N=338
R=0.11
N=338
R=0.16
Note: The correlation between poker ability and the availability bias is statistically significant at a 5% level, and the correlation between poker ability and the
representativeness bias is statistically significant at a 1% level
18. Result III: Individual propensity to both biases is more
strongly correlated with poker ability than any single bias
0.15
0.20
0.10
0.05
0.00
Representativeness
bias
0.16
Availability
bias
0.11
Correlation between individual propensity to biases, and poker ability1
Combination
of both biases
0.18
1. The difference in correlation between the combination of both biases and the availability bias is statistically significant at a 10% level, whilst the difference between the
representativeness bias and the combination of both biases is not statistically significant
For those interested; the correlation is significant at a
1% level (5% for availability) – which means that I am
99% certain that my results are correct.
N=338
19. Result IV: Professional poker players are less prone to
cognitive biases than amateurs
...and the professionals are undoubtedly less
prone to cognitive biases than amateurs!
I segmented my sample based on their "main
source of income"...
• 73 individuals stated poker as their main source
of income, and I count them as professionals
– The median poker professional had made
$18k from online poker (usually the pros have
accounts on several online sites – so the real
number is probably much higher)
• Whilst 265 stated other professions as their main
source of income – such as engineering, teaching,
IT or logistics
– The median poker amateur in my sample had
lost 16 dollars playing online poker2
0.0
-0.5
-1.0
-1.5
-2.0 -1.94
Representativeness
bias
-0.92
Combination
of both biases1
-0.37
-1.17
Availability
bias
-0.55
-0.77
Amateurs
Professionals
Differences in propensity to biases between
professionals and amateurs
(Negative numbers indicate a more biased individual)
-11 11"Cognitive Bias Scale"
The "non-biased" individual –
always getting the correct answer
would score 11 (1 point per
exercise)
The "biased" individual – always
getting the biased answer would
score -11 (-1 point per exercise)
1. And the combined difference in biases between pros and amateurs is significant at a 1% level. (The results are correct with 99% likelihood) 2. Skewed towards good players
N=338
20. Result V: For elite-decision makers, such as professional
poker players, it is even more important not to be biased!
...so cognitively biased professionals are
seriously harmed by their errors!
Cognitive biases are more correlated to the
performance of professionals than amateurs...
0.3
0.2
0.1
0.0
Combination
of both biases
0.26
0.13
Representativeness
bias
0.22
0.09
Availability
bias
0.19
0.09
Professionals
Amateurs
Correlation between poker ability and propensity to
biases for professionals and amateurs
• Statistically this means that amongst this group – my test
can reveal ~6% of the performance differential between
individuals!1
-10
-5
0
5
10
5 6 7 8 9 10
Propensity to cognitive biases (negative more biased)
Poker ability on a scale from 1-10 (10 being best)
1. By running an adjusted R squared analysis for those interested... And yes – the p-level of the test is very good at 0,025
Details matter more for pros (they all know the basics). Just as Northug is more dependent
on good skies to win Olympic races than your kid is dependent on good skies to win his
local race, being unbiased is more important for top-decision makers than for amateurs
21. Actions I: Individuals should develop strategies to overcome
cognitive biases to become better decision-makers!
"Firms with dispersed power will commit fewer strategic
errors than firms with an "all-might CEO" – since
individuals are more prone to cognitive biases than
groups
A CEO contemplating launching a new product might
make a better decision, if she realizes that her
assessment of the likelihood of success is probably
coloured by her recollection of the success or failures of
similar products in the past (awareness is key)
"Billy Beane improved the Oakland Athletics by overriding
the ingrained heuristics scouts used when looking for
baseball talent, and replaced it by statistical analysis"
(Watch Moneyball people...!). The biases stemming from
heuristic reasoning might not be crucial if you are picking
out the firm softball team, but amongst professionals
details are key. And eventually other baseball teams had
to replicate Bean´s analytical approach to remain
competitive...
"The next time you see that young Swede bets me, I will counter attack!" Don't jump to the conclusion that he is
aggressive and bluffing a lot (he is representative of a bluffer)... Make sure that you understand the base-rate – how
often is any poker player bluffing in that particular situation? And adjust your estimate according to this.
"This guy has only played 2 hands for the last hour – he must be super tight!" Don't jump to this conclusion (availability
bias)... How often is it that a normal (non-tight) player ends up only playing 2 hands for an hour? Adjust your estimate
accordingly...
22. Actions II: Some pointers to overcome these biases and
become better decision-makers
"Awareness is key"
"Expose yourself to statistical thinking!"
Consultation or debate with others – as
groups are less prone to cognitive biases
than individuals
"Poker players should engage in game
analysis with other professionals!"
Avoid engaging in several mental activities
at the same time!
"Cognitive biases are more likely to occur
under time-pressure"
23. Did you like this? Here are some killer books to learn more!
A Beginner's Guide to Irrational Behavior
A great online course on the topic from Dan Ariely
Check it out at www.Coursera.org
24. For now – thanks for paying attention!
(And a big thanks to those who participated in the survey!!)
"I'm so excited to learn
about biases and making
better decisions!"
http://www.slideshare.net/lassebringstad/the-costs-of-irrationality-v17-article
Here is a link to the actual article I wrote if you would like to investigate it in
further detail! (The full survey is also in the appendix of the article for those keen
to check it out)
PS. I will be in Dublin for the Norwegian championship
main event – so please reach out if you would like to
chat about the results (naturally – I think this is a super
interesting topic for discussion).
25. Bonus: The LSE professors / markers also found my findings
interesting and plausible
" Lovely clear finding that professional gamblers
suffer less from CBs than amateurs. "
"An excellent small quantitative study, with valid
and reliable results."
The study is described very clearly, with a good sized n and good justification of what
was done. The statistical analysis looks very thorough and is reasonably
sophisticated. The discussion, conclusions and limitations sections all flow logically
from the findings.
"...you have found something very interesting
indeed!"
"While I retain some doubts about the
generalisability of games phenomena to
decision-making generally, the author clearly
understands these general reservations and
goes out of his way to justify his approach"
"My only query would be about generalisability,
from poker to strategic decision-making
generally - but this is a well-trodden path and the
writer puts forward a reasonable case. He
expands upon the justification later (section 2:3)
and argues his position well"