How have EOD markets evolved over the past 20 years?

[fatowl]

My end-of-day (EOD) stock trading system simulates very well over the past 50 years or so. I have listed and delisted equities data going back to 1950, and my system's performance evolves over the years in simulation:

1. 1960 - 1974: High trade expectancy and minimal drawdown.
2. 1974 - 1984: Lower trade expectancy and more frequent drawdowns.
3. 1984 - 1992: Similar to 1960 - 1974.
4. 1992 - 2001: Extreme outperformance with huge expectancy and smallest drawdowns of simulation.
5. 2001 - 2010: Underperformance with low expectancy and big drawdowns around the bottoms of the US markets (2003 and 2009). Similar to 1974 - 1984 but a little worse.
6. 2010 - today: lowest expectancy of simulation. Gains are flat and drawdowns are more frequent and longer lasting.

My strategy is contrarian: Bet in the opposite direction of large, sudden price swings. My trades typically last 4 days. This type of trading does well in choppy markets and does poorly in prolonged, straight-up or straight-down markets without breaks. My stock universe covers all stocks on the US exchanges except NYSE Arca.

Why has my performance been so poor lately? I know that technology has been a major change over the past 20 years with online trading, lower commissions, and more access to algorithmic platforms. Also, 2012 and 2013 have been largely non-volatile, straight-up years for many stocks compared with prior years.

Is this all there is to it? I'm hoping you guys and gals have some insights that I don't have.

Also, I've been trading my system live for the past 15 months and have a +10% cumulative gain. I reached a high of +27% back in November 2013 and have been in a drawdown since then.

 

[OrderFlowDashPro]

There are "higher frequency" traders who are likely impacting your models. These traders are basically arbitraging the opportunities that used to take a few days to resolve. Reading the tape, I have seen the models work across a variety of time time frames and sub frames.

As for why these type of opportunities exist in the first place, I suspect it has to do with the impact of leverage that the types of traders have access too and how long they can "hold" that leverage. A trader who only has day leverage may be able to push the market out-of-balance on a smaller time scale creating an opportunity for a larger time scale correction. On the other hand, that same longer term opportunity might be arbitraged by higher frequency traders.

But, yes the factors impacting your models are likely -- higher frequency traders, the types of traders in the market and the leverage they have access too, volatility, and also more fundamentally regime may be at work too. Your models are also likely to be affected negatively by macro-market fundamental moving news.

 

[fatowl]

There are "higher frequency" traders who are likely impacting your models. These traders are basically arbitraging the opportunities that used to take a few days to resolve. Reading the tape, I have seen the models work across a variety of time time frames and sub frames.

As for why these type of opportunities exist in the first place, I suspect it has to do with the impact of leverage that the types of traders have access too and how long they can "hold" that leverage. A trader who only has day leverage may be able to push the market out-of-balance on a smaller time scale creating an opportunity for a larger time scale correction. On the other hand, that same longer term opportunity might be arbitraged by higher frequency traders.

But, yes the factors impacting your models are likely -- higher frequency traders, the types of traders in the market and the leverage they have access too, volatility, and also more fundamentally regime may be at work too. Your models are also likely to be affected negatively by macro-market fundamental moving news.

Thanks! I'm confused what you mean by "arbitraging the opportunities that used to take a few days to resolve." Do you mean that when a stock is oversold, there were not many high-frequency traders to bring the price back down a little bit? How are you ascertaining this by reading the tape? You can't see who is buying or selling. You can only see the amount bought or sold.

 

[OrderFlowDashPro]

I can't see who is buying or selling, correct. But, as an experienced tape reader I can see how the programs operate. I also know when my own quantitative programs trigger and can see how these bots operate. The HFT traders buy the stock intraday when it is oversold and return it to value before your EOD systems have any opportunities.