Archive for December, 2013

I have been working on modeling and predicting stock movements and have seen that the LM strategy performs quite well for random baskets of stocks even with a 2% transaction cost.  I have the problem that my initial capital is quite small so I cannot afford to trade 100 share baskets corresponding to the option contract sizes.  As I am willing to lever the strategies, a modification is to use only options and use the same contract size options for the stop-loss.  So instead of buying 100 shares of a stock I can buy an at-the-money call instead.  This has the benefit of getting rid of the trading costs of the stocks altoghether.

When my directional estimate for price movement is positive, I buy a call @P and a put @(P-stoploss) and sell both the next day.  The transaction costs are then $10 in OptionsHouse which is 2% for $5 stock and less for more expensive stocks. The performance seems at least as good as the strategy including buying actual stocks although I am using Black-Scholes to estimate prices which have defects which lead to the volatility smile etc.  But here is an indicative result:



Read Full Post »


With 2% transaction costs, the strategy seems quite good with another mix of stocks.  The Sharpe ratio for the backtest below is 1.99


Another random basket of optional-stocks with LM strategy backtest of Sharpe 1.90 and 2% transaction costs.


Read Full Post »

A cursory survey of stock and option trading platforms shows that Interactive Brokers is the cheapest with $1 stock and option trades with minimal per share additions.  The problem is the high minimum balance of $10,000 required to use their services.  Then one can look at low minimum balance platforms and finds $3-5 per trade with eOptions, OptionsHouse and so on but they have an obscenely high options exercise price of $9-$10.  I use options in my strategies as a sophisticated version of a stop-loss.  One idea that I am exploring is that supposing my strategy signals taking a long position on a stock at P and long position also at P(1-0.015) strike option and the daily move loses delta% of the price, then I sell the stock and the option (with transaction cost closer to $6) versus exercising the option for $9-$10.  The P&L for selling both is -P*delta + P&L of options position.  Now given that delta > 0.015, the option is in the money and sells for more than (delta-0.015)*P and we should still achieve the desired stop-loss of -0.015*P and save $3-4 that would be required for exercising the option.

Read Full Post »

It felt that I was in a school hallway of some sort.  My aunt, my father’s older sister came and gave ‘us’ some warning about our activity.  There was an arrangement of leather-covered couches in a hallway.  I don’t know who else was with me.  I took my time and then as I exited the hallway by stairs, I found a cigarette on the way.  On top of the stairs I found my father waiting for me.  He turned to leave the building and I followed him out.  The dream, which had more adventures before this point that I cannot remember, ended.


Read Full Post »

Reading some old papers of William Sharpe, I am coming to a better understanding of the conceptual tools that have been developed over time to understand ‘risk’.  We are always really interested in future risk and future returns although we turn backward to look at historical returns and other possible factors in order to estimate risk.  At least on backtest results my LM strategies seem to be quite successful and this points to a feature of the return series that is not studied and applied to risk and return prediction methods, which is the characteristics of sequence of movements in the return series.  These affect risk and return predictions and do not appear without reason.  The financial markets have regularities because of collective trading behaviours of large numbers of individual and institutional traders, and these regularities are not always removed by arbitrageurs.  There are thus opportunities for novel estimations for future risk and returns based on different principles than the straightforward historical mean and variances of returns that have been prevalent in the portfolio literature.  I will come back to new estimates in a later post.


Read Full Post »

I don’t remember much of the interesting parts of the dream in which I was in some adventure with an Argentine man as companion.  After a number of complex adventures, I remember the topic of conversation had turned to the Anglo-American attitude towards the Japanese.  T. Roosevelt had been quite fearful of the Japanese.  American entry in the Second World War had put Japan under occupation after bombing it with nuclear weapons.  The dream was rich but my memory of the details remains beyond reconstruction.

Read Full Post »

The strategy takes long/short position by a linguistic analysis of return series in a 2000 day window and then takes positions along with an option position for a stop-loss of 1.5%.  The option P&L is included in the backtests.  Transaction costs are assumed to be 0.2% which is reasonable for sufficient size.  I am putting down graphs of some previous performances to give the sense of the performances.


This is just a sampling of the strategy performance in backtests.  There are a few outstanding issues before I can begin trading these strategies, such as whether using Black-Scholes prices for options is unrealistic and whether 0.2% transaction costs can be achievable.


Read Full Post »

Older Posts »