Exactly how efficient are the markets+Do trading systems works

I coded an efficient market based trading system.
Now this system works brilliantly on the Indian markets but fails to work on the u.s market data that i have.
The foundation of the system is that if the main index is bullish then there will be bullish influences on stocks at certain levels and the other way around for the short side. ("a rising tide lifts all boats") The system gets in and out quickly: holding positions for barely 2-3 days.
The risk adjusted returns are almost 3-4 times that of buy+hold nifty.
Perhaps every system is designed to fail in the longer run?
Technical analysts are always searching for techniques which will outperform the risk adjusted returns of the indices. The only issue is.
A trend following system loses money in trading range mode and a trading range system loses money in a market that is in a breakaway move.
The market never announces what mode it is in.
Trading range or trend?

For those who are wondering adx is too laggy to tell anything. The indicator itself loses money in mechanical tests.

Other interesting discoveries. the bill williams fractal breakout trading system works better on u.s markets than it does on indian markets.
Alexander elder tripple screen doesn't work with eod breakouts but can be used with intraday setups to trigger entry.
Donchian channel breakout don't work too well anywhere.

The medieval alchemists failed to turn lead into gold but laid the foundation for modern chemistry. Perhaps the modern trading system developer is doomed to fail in creating something that works all the time but will lay the foundation for a true AI.
The problem that is being attacked is not just entry/exit/sl signals. What is needed is a way to figure out the dna of the market. The underlying structure at any given time based on technical data/fundamental data and perhaps even news.
A purely mechanical system will not work. Unless you run a backtest for a trend following system on a trending market and a range trading system on a ranged market you are bound to get poor results over extremely long timeframes.

The bottom fishing picks of 3556 nifty all gave decent returns 10-20%
on pure fundas (aftekinfosys, aloktextiles). While the breakouttrades also did their targets: tatatea, hcc, rcom. Infy, bottom fishing at 200 sma failed to return much while tcs bottom fishing near 1200 region gave mediocre returns as well.
So net net, I don't really think my calls are outperforming the market by much.

Now what I need is
access to ridiculous amounts of funda data in a format that i can use
+ access to daily news
then I'll be able to move on to the next step in my journey of trading system development.
Mostly done making systems for now.
The only conclusion I can give right now is
a) trend following systems work in trending markets
b) contrarian systems work in range bound markets
c) "the trend is your friend until it bends" i.e false breakouts+range
and the problem remains , you never know which to use. Perhaps the right strategy is to check the fundamentals of a stock and figure out whether profit/sales are flat/growing and use that as a filter. You might miss the boat on the initial breakout when the first breakout quarter results come but you should be able to get on board after one quarter.

Look at dow jones charts 2000-2006 the trading range would have killed 90% of mechanical trend following systems on the eod charts.