Price Action for a million

Tuesday, March 1, 2011

Checkpoint

01st Mar 2011 - I have been losing money being greedy. Also, when I don't follow the rules. When the trade goes out of the channel, I still hold it dearly hoping for a miracle to happen. Multiple timeframe needs a lot of backtesting. First step is to manually create a log of all the trades of 1 min timeframe.

Saturday, February 19, 2011

Backtesting Tutorial

http://www.amibroker.org/userkb/category/amibroker-features/custom-backtester/

If this line is present
SetCustomBacktestProc(""); at the beginning you need to have the
if( Status("action") == actionPortfolio ) later to perform actual backtest. Just pressing the backtest button won't do.

for (i = 0; i < BarCount; i++)
{
. . . .
sig = bo.GetFirstSignal(i);
. . . .
}

Signal is for the current symbol only because backtest results are per symbol. Open position is a loop for individual bars for the time position is open. Signal is only one record for Entry and one for Exit.

Friday, July 2, 2010

Points to ponder for testing

1) Backtest
2) Walk-forward test
3) Optimization
4) Pyramiding/Scaling (sigScalein/sigScaleout)
5) Commissions
6)

Pre-requisites for testing

First and foremost, I need loads of data to test. This data will be divided for backtesting and walk-forward testing.
1) Download BSE EOD data from BSE website. I already have data for the last 3 months. BSE provides data from Jan 2007 till date. So that is approx 3.5years worth of data for 3000 bse stocks. Probably that should be enough.
2) Trading system in amibroker needs to be tweaked for backtesting, introducing SLs and Targets.
I will add more points as needed.

Failure of the Coiled spring strategy or lack of testing

It has been a losing streak for quite a long number of days now. After trying out Price action sanuk and Buff trading, I was trying out Coiled spring trading. Needless to say, it performed worse than the other too. Well on a macro retrospection, I came to the conclusion that out of the three strategies, coiled spring gave the maximum returns. Maybe I could not milk the strategy due to lack of Optimisation and Testing of the trading system. So this is the plan. I am going to run all possible tests, optimizations, walk-forward tests to see how the trading system performs. It might take a week to do the same. I would be refraining from realtime trading during this testing.