Drunkards walk - completed book review

[gazuty]

The Drunkard's Walk - How randomness rules our lives - Leonard Mlodinow

Could a random sequence of 0s and 1s produce 30 million 0s in a row? Could a roulette wheel be fair if 3000 reds showed up in a row? And would you place your next bet on 1 or black?

This book looks at the history of probability theory and the characters behind it, from the Greeks and the Romans, through to (probably the greatest mathematical gambler in the history of the world) Pascal and then into more modern times. Just a few words on Pascal - imagine how much money you would make when you are the first person in the world to develop probability theory and apply it to gambling, with everyone else - especially those setting the odds with no idea!!!!

Reading the book can make one feel a little sad in that so many decisions in modern business are made in ignorance of big data and the probability and statistical methods that could be applied but are not.

As some of you know i have a degree in maths and the beginning of the book felt to me like a bit of a repetition of high school maths. However, there is some story telling and many gambling examples to assist those for whom maths does not come easy. (Even though I am very good at maths I have never been a great teacher - because to me it all makes sense and my brain says "of course", I'm not much good when someone says - I don't understand). This book may help teach some of the theory given the use of the historical context and examples.

I found the second half of the book more interesting, perhaps because it had more practical examples.

Overall I would recommend this book to those with a mathematical bent with an interest in mathematical history. I'd also recommend it to those who want to be challenged in their ideas and assumptions. There is plenty to apply to trading and BA techniques if one puts the time in.

[Wyndon]

I have a small collection of books on randomness, uncertainty and probability in general - this is my favourite and one of the few which I tend to go back to. Another thought-provoking book is Ubiquity by Mark Buchanan, which shows how power laws (that's mathematical powers) crop all over the place. The subtitle is "Why Catatrophes Happen" - how little events trigger much bigger ones - and how the temptation is to try and work back to find a single "cause" that made the process inevitable, whereas whilst the end result was inevitable, there are many possible causes.

[gazuty]

The subtitle is "Why Catatrophes Happen" - how little events trigger much bigger ones - and how the temptation is to try and work back to find a single "cause" that made the process inevitable, whereas whilst the end result was inevitable, there are many possible causes.

Yep. "Accident theory" also briefly covered in Drunkard's Walk.

Also so called "intelligence failures". It all seems so obvious in hindsight and yet at the time it is not, or at least not obvious.

I am always intrigued by statements, my side would have won if we had scored that goal, the ref made a different decision etc. of course if the goal had really been scored, or the ref made a different decision, the whole course of the game would have been altered and no one would know what the outcome from that point would have been.

[PeterLe]

Just off topic slightly; You guys obviously have a love of mathematics, as I do.
Most of my formal qualifications were in electronics. I wish I had done my degree in mathematics as most of what I was taught is way out of date now.

I just wondered if you had used anything other than basic maths in your excel formula with great effect? I once used the golden ratio (http://en.wikipedia.org/wiki/Golden_ratio) in a formula (I also used it when I built my patio!, my missus thinks Im a bit geeky!)
Just curious as to what level maths you had used and whether it has posed a challenge in getting it right!

regards
Peter
PS There are some good programs by marcus du sautoy (maybe on BBC Iplayer - worth a watch. it opens your eyes that mathematics are all around us)
PS I read the book mentioned above too (in audio) and I tend to reread thm now and again to refresh my memory..

[NileVentures]

Hi Peter,
My first degree is in Civil Engineering, which is basically applied mathematics. For the first two years of my degree we did lectures with the maths majors and other engineering undergrads.
You should consider looking at Markov Chains and Stochastic Processes in the theory of probability and randomness - it is not as difficult to understand as it sounds Markov Chains are particularly useful in the modelling of the probability outcome of tennis matches. There is lots of material on the web.
Enjoy!

[PeterLe]

Thanks NV
I have come across Markov Chains before although I haven't used them in any strategy.
Ill have a look at them when I get time
Regards
Peter

[Wyndon]

Hi Peter,

My degree was in Mathematical Sciences, but I don't think I've used anything above A level standard in my betting. To be honest, I don't remember much of my degree course - I was a bad student, which I regret today. Occasionally I do something that I consider really clever in Excel to get the result I want - but the actual maths may be fairly basic. The trick is to apply lateral thinking to a problem in the hope that you come across an aspect that nobody has considered before. I have now given up trading (in the sense of offsetting back and lay bets) as I consider that I'm temperamentally unsuited to it - I don't have the self-discipline required. I have an automated in-running backing programme which is showing promise. I'm intrigued by your use of the Golden ratio - no obvious application to betting springs to mind. Something that is really exciting me at the moment is the use of the Kelly Criterion in determining bet size (see Fortune's Formula, William Poundstone - a good read, as well as providing the background to the mathematics). A lot of people claim to use the Kelly criterion, but tend to use an average of their long term advantage, rather than looking at the advantage in individual bets. How do you determine that? You need to analyse past results of the selection process you employ. There is another thread where someone is asking 'what is value?' - it depends on your selection process and past results. For instance, consistently achieving a 10% premium on Betfair SP would be value. How you achieve that is another matter...

[PeterLe]

Hi Wyndon,
In hindsight; i probably got carry away with that statement about the golden ratio, I did use the figure, but in all honesty, it would have worked with + - 0.5

Ive read that book and Im familiar with Kelly Criteria (KC).
A question you may be able to answer then...
Imagine the scenario: A race is in running, and the back price of the horse is say 3's and the lay price is 7's (it happens). Which odds would you use to determine the true chance of that horse as an input to the KC equation? (or would you simply choose the midpoint?)
regards
Peter
PS As an alternative to KC; you could have a simple calculation that is constantly monitoring the P&L of each horse. If the lowest profit is say greater than £10, then add 5% or 10% etc of that value to the stake. Every now and again when you have the perfect race, your stakes can grow proportionally to the lowest P&L (my own term for this is ratcheting) Ive had some of my biggest wins like this

[gazuty]

Just off topic slightly; You guys obviously have a love of mathematics, as I do.
Most of my formal qualifications were in electronics. I wish I had done my degree in mathematics as most of what I was taught is way out of date now.

Always loved maths and have a degree in applied mathematics. Also love physics.

From the day I finished my degree until I discovered BA I had never again used my abilities much beyond high school maths let alone university maths.

I don't use any complicated formula, but i do consider maths and physics made me open to experimenting and testing, adjusting and tweaking. And thinking, thinking, thinking. And looking for patterns. Then exploiting them. Which is what my automation does. It's not fancy and not complicated, it took time and observation, I started with a theory and refined it.

My brain approaches everything in life as an equation - it's the way I was made or my brain was wired. If x then y. This made it come naturally to learn some simple computer programming skills and other skills. Composing logical arguments made it easy to be a lawyer. And that is where I earn my money.

But I'd always dreamed of using my maths skills to plunder the financial markets with some friends who also did maths (some of them also did engineering). Only one of our group stayed in a technical role - all of the rest ended up in the financial services industry (with me as a service provider to that industry). Even the guys at the banks, are not using their maths every day, although at one stage two of them were doing Monte Carlo simulations. But they are now in higher management.

Ps. None of this makes me any more or less likely to succeed and I certainly don't think you need a strong maths background to win at trading or BA generally. IMHO you do need a willingness to experiement to find an edge.

PPS. I'm reading a book called the Element - about looking for what makes you truly happy and then trying to focus on that, getting in the zone. Solving problems is what makes me happy. Doing that in the company of friends and other really smart people makes me very happy.

[Wyndon]

Imagine the scenario: A race is in running, and the back price of the horse is say 3's and the lay price is 7's (it happens). Which odds would you use to determine the true chance of that horse as an input to the KC equation? (or would you simply choose the midpoint?)

The short answer is that I would never use the KC in an in-running situation - especially one as volatile as a horse race. Apart from the near impossibility of knowing one's edge in a rapidly changing environment, to gain the most from the KC you must match all your bet - which wouldn't necessarily happen in-play. Neither would I consider alternatives to KC without some mathematical backup. The whole point of the KC is that it is mathematically principled and WILL work provided (a big proviso) that you have a reasonable estimate of your edge in each situation. I'm acting as an ordinary punter in this respect in that I'm betting pre-race with an edge calculated from several year's historical data. My in-play automation uses level stakes.

I don't use any complicated formula, but i do consider maths and physics made me open to experimenting and testing, adjusting and tweaking. And thinking, thinking, thinking. And looking for patterns. Then exploiting them. Which is what my automation does. It's not fancy and not complicated, it took time and observation, I started with a theory and refined it.

Snap. Also I must say that it is refreshing to be part of a forum where the members are not polarised between form junkies and mathematical wonks. There must be a place for both. Some years back I was the member of an email group where anyone putting forward mathematical ideas was derisively classified as a 'tosser' by the form experts - a reference to the technique of using coin tossing to demonstrate probability theory.

[NileVentures]

I was the member of an email group where anyone putting forward mathematical ideas was derisively classified as a 'tosser' by the form experts - a reference to the technique of using coin tossing to demonstrate probability theory.

That reminds me of my Univercity days, where the Maths Dept was split between Pure Maths, Applied Maths and Maths & Statistics. With tongue in cheek, we calls the stats guys "tossers".