Bi-Directional Algorithm

Levkovich's Buy-Low, Sell-High Model

The chart below is a 12-year chart of SPY. In 12 years, SPY had 0 return. On Nov 11, 2010 SPY was at the same level recorded on Dec 24, 1998. No one was able to buy at A, C, E and sell at B & D.
Citibank's analyst Levkovich discovered a buy-low, sell-high model that could beat the market by selling 10% of SPY when it was up 20%, and buying 10% when it dropped 20%. The return was 14% in this 12-year period.

Using Stock Market Capitalization vs GDP as filter, the return was 36% (When Stock Market Capitalization / GDP > 115%, sold 10%; if Stock Market Capitalization / GDP < 75%, bought 10%). However, Levkovich's model is not perfect (it is directional). If we entered the market at B or D, the model would generate a loss.

We built a bi-directional algorithm based on Levkovich's model at much higher frequency. The back-testing shows impressive results. Sharp ratio is 2.80! Maximum risk for long and short 10 contracts of YM is only $1,000 per day, $3,000 per day for +/- 10 contracts of GC, $1,000 per day for +/- 10 contracts of QM. We tested this model on 44 US stocks. The results are impressive. » more
From August 2011 we began trading this model live. » more

The Collar Strategy

Collar is a proven strategy that can transfer the downside risk of investing in stocks. Szado and Kazemi of the University of Massachussetts performed a study of QQQ.

Over the 9-year period from March 1999 to March 2008, the collar strategy returned more than 150% cumulatively, while QQQ lost 12% during this period. Annualized return was 23.01% with standard deviation of only 9%. While the standard deviation of QQQ was 27%!
Szado extended the study and discovered a way of boosting the return  by adjusting the strike prices of the collar. (When the stock price has moved up, adjust the strike price of the call up. Szado called this the
Active Collar.)

The Algorithmic Collar

We discovered a better way of boosting the performance of the collar strategy by actively trading the underlying stock in addition to adjusting the strike prices of the collar. » more

Most of the time, the stock does not move after adjusting the strike prices of the collar which has to take loss on the existing collar.   

 Risk-Free Dividend Arbitrage with a Collar 

With proper timing and the collar strategy, we can capture rich dividends risk-free.

The strategy can be implemented every month. In the monthly approach, we need three stocks. Each one has ex-dividend date in a different month. Assuming all three stocks earn 10% per year, this translates to a quarterly return of 2.5%, each and every month. Annual return is then 30%. Assuming that there is 50% chance that the stock will be called away before the ex-dividend date, the annual return will be 15%.  
Many stocks pay more than 15% annual dividend.  » more

Bi-Directional Model

We trade the same asset long and short at the same time automatically.

Algorithmic Collar

We actively trade the underlying stock with a collar.

0-Risk Dividend Arbitrage

  • We foud opportunities in 0-risk dividend arbitrage.


Risk Disclosure

Investment involves risk. Hypothetical performance results have many limitations. No representation is being made that any account will or is likely to achieve profits or losses similar to those shown. One of the limitations of hypothetical performance trading results is that they are generally prepared with the benefit of hindsight. In addition, hypothetical trading does not involve financial risk, and no hypothetical trading record can completely account for the impact of financial risk in actual trading.