Trading vs Poker

I've heard many people say that if you are good at poker, you will be a good trader. I never quite related to that fact well enough in my college days. Not any more!

Let me first make it clear that I am talking about trading being similar to poker, and not investing. Investing, obviously based on fundamentals, is very similar to bridge. But that's a topic for some other day, which will come very soon.


So yeah, back to poker. The analogies I am making is for the secondary market trader functioning in fairly liquid markets. Let me begin with the similarities which exist in the setup of poker and trading.

Poker is a zero sum game, and trading is nearly zero sum, with its trade execution costs (transaction costs, bid-ask spreads, etc)

The cards held by the player represent the fundamentals of the company's stock. Hardly useful for trading, especially during a constrained time horizon. Your trading strategy will win depending on a lot of factors, and fundamentals play a very small role. Rockets are busted as frequently as they end up winning.

Each step while opening the community cards represent the unveiling of key macroeconomic market parameters, e.g. change in interest rates. They are macroeconomic because the affect everyone's holding. They will favour some companies/positions and work against some other. The trader exits the position if the data unveiled hampers the position. The trader might also have to exit the position if the cost of continuing the position (e.g. margin calls) is out of his budget, even though the odds are in favour.

The bets by a player represent the information the company is trying to communicate to market players, in the form of press releases and Financial Statements publications. Frequently enough, they hide/misrepresent information, or they hold up their biggest piece of information till the fourth quarter (which represents the biggest bets being made on the river).

Reading people's faces represents the Behavioral Finance element. The trader will win the pot if he reads the mind of the junta correctly. Understanding the sentiments of the junta is the (I can't stress this enough!) most important art to be a good player/trader.

Apart from these, various participation strategies exist which are similar in trading and poker

There's also the reputation factor. Consider a trader who is widely perceived as a god by the market. If he decides to take a certain non-anonymous position in the market, that position will be regarded a sure shot success strategy. Therefore, it will be hard for him to find a dealer who takes the counter-position of this trade.Similarly, a player with a reputation of being tight, will always find a lot of folds when he decides to participate in the pot.



There's a poker strategy which many people use, I call it a test bet. Its similar to a c-bet, but the motivation is different. Immediately after the flop (and sometimes the turn), I'll put in a small bet to judge what others are upto. If this bet results in some action, I am in a much better position to judge whether to raise/fold/call. This is similar to the trading strategy where traders place small limit orders above and below the current market price, to gauge what the market is upto. In case these small lots get executed, they might pull in (similar to a raise) or pull out (similar to a fold) the big lots.

I am sure there are many more strategies which are similar in poker and trading. But I haven't played poker in a while, and I haven't done trading at all. Any more analogies are welcome!

The Realty: contd...

This is in continuation to my previous blog, "The Realty".

In my previous article, I barely touched upon the "other" factors. I included all of them in a single subjective variable, which, as many pointed out too, is wrong. Here, I will try to include those other factors in an as objective way as possible. This way, I intend to end up with a multi-variable regression with 5 to 10 independent variables.

This analysis is certainly fraught with dangers, the biggest one being that the independent variables can themselves be correlated among themselves. I do not have the tools and knowledge (more importantly the enthusiasm) to correct for that.

Besides, I am using the data of only 10-odd cities which is clearly a very small sample size.

Including these other variables independently should also probably reduce the R-square value of population density. But lets keep the conclusions for later.

Lets list those other factors, their construction and the data source that I'll use for those factors:

1) Infrastructure and Transportation: This is a tough one. Here, I intend to measure the existing level of infrastructure in the city, and in particular, the transportation facilities, on a per capita basis. Thus, it should include factors like road length (road area might be better), rail length, number of buses, number of train bogeys, niche transport facilities (metros, availability of air transport) etc. So the proxy that can be chosen for this variable is

Total Road Length/Population]

2) Education opportunities per capita: This can be a bit tricky. I can't simply use the Total No. of Educational institutions per capita, because it grossly undermines the value of higher education opportunities and also doesn't take into account the fact that institutions can have variable intake capacities. The best way will be to use a number relating to the total intake in various colleges, needless to say, on a per capita basis of existing population. The value of this variable is

[Total Intake in all the city's colleges/Population]

3) Income per capita and job opportunities: This one is straight forward. I'll just use the per capita income of the city. This variable is also a great indicator for the existing job opportunities for two reasons, one, because a high per capita income means there are opportunities waiting to be ceased in the city and second, a high per capita income in turn creates a lot of trickle-down job opportunities (by the way, I am a firm believer in trickle-down economics. There's too much regulation and taxes in this world at the moment). Reiterating, this variable is Per Capita Income

4) Climate and Environment: There are two dimensions to this factor. First is, how the mean temperature in the city compare to the country average. The second is how much deviation is there in the temperature (month-on-month standard deviation) over the whole year. I am tempted to include precipitation here, but the effect of the amount of rainfall is uncertain, as there's no preferred range of precipitation, at least its not a narrow one. Things like proximity to beaches or hill stations will add to this factor, but this fact is best incorporated by adjusting the variable value in a subjective manner. Thus, this variable is (before adjustments) (|xyz| means absolute value of xyz):

|Mean City temp. - Mean India temp.| x Std. Dev. in the city's temp

5) Safety of Living: For this, I have to pick a type of incident which best represents the unsafety of the place and for which the stats are also highly published. This variable can best be proxied by the number of rape cases in the city, per capita.

6) Taxes, Regulatory and Legal issues: This variable should measure how convenient is the legal framework of the city to settle in. This is, by definition, a subjective variable. Still, I think it can be proxied by the Road Tax percentage in the city. Thus, a proxy for this variable is
Average Road tax rate %

6) Herding behavior, Metro Premium: A premium can be assumed for cities that are already developed, esp. if it's a metro. Thus this is a binary variable indicating if it's a metro.

As expected, data collection of this level is a daunting task. I tried collecting this data, but I am having many difficulties in completing this task. This exercise has to be left incomplete.

Regret ratio

Consider two situations:
A trader starts with a million in his hands. He enters the market at a good time and, in the next three years, reaches a position valued at 50 million. After which, a recession hits, and his position is reduced to 5 million in another 2 years' time. He now exits his position. In 5 years, he made 4 million, a cool 400% return.
Now consider another (counter-cyclical) trader, who starts with the same 1 million. After 3 years, his position reaches a valuation of just 10 thousand. He recovers in the next 2 years, and ends up with a position valued at the same 5 million.

I recently finished reading 'Fooled by Randomness', and Taleb describes a situation very similar to the above in that book. I have two adjectives for the guy, arrogant and genius, in that order.

But the question he raises is whether these two 400% returns are equal? Clearly not! Even if these paths are equally volatile, these returns are not the same. Conventional risk-adjusted return measures do not work here. Even ratios which aim to distinguish between upside and downside deviation (like sortino) are not much of a use. The problem here is more behavioural, and the way they perceive their returns.

The first trader will always think of his experience with regret and sorrow. He will keep thinking how great it could have been, had he exited at the peak. Because he perceives it as a missed opportunity, he will underrate his return. The opposite is true for the other trader. He will be a lot more satisfied with his return, and he will perceive this return to be more than the actual value.
It is this difference in perception that I want to capture with a number. I suggest the following:

Regret ratio = (Maximum Valuation - Ending Valuation) / (Ending Valuation - Minimum Valuation).

This ratio should work very well in all those cases where the return is positive during the complete 5-year period. I do not intend to measure regrets when the total return during that period was anyway negative.Hence, it is a non-negative number.

The numerator here is essentially a measure of the feeling of regret with the experience. The denominator measures the opposite feeling of satisfaction or good luck with the experience. Their ratio should denote the net effect of regret from the experience. The higher this number, the larger is the feeling of regret. The smaller this number, the more the person gets a feeling of satisfaction.

Algorithmic Recession

I hereby make a bet that the ongoing revolution of algorithmic trading is going cause a global recession once in the next 100 years
My reasoning is fairly simple. Due to the stochastic nature of market movements, we will continue to see minor upturns and downturns (essentially caused by behavioral reasons, rather than any fundamental one). The algorithms in place that time will somehow pick on one of these upturns, and inflate it. All the algorithms in place will try to benefit from this anomaly, participate in this bubble and inflate it. Until some human intervenes, and exits the bubble. This will be followed by many humans and algorithms doing the same, causing the bubble to crash.
This story seems to have the two necessary and sufficient ingredients of a boom-bust cycle. It contains a financial or technical innovation (algo trading) being used excessively, which is necessary to fuel a bubble. The restoration of sanity, due to human intervention, triggers the crash-correction.