I came upon a pdf that describes the original turtle method.
It is a breakout system, similar to ACD in some ways, and uses High and Low of past 10 bars for breakout levels.
It also uses a calculated value to determine when to add a position.
The original method limited that to 3 or 4 while I wrote the EA to use up to 12.
Some pairs get a margin call from the strategy tester while others are able to open 12 positions so I set default to 8.
I developed an EA using most of the rules and added additional filters I have found useful over the past 10 years.
- For overbought/oversold I use my stochastic-rsi filter using chart timeframe.
- I have used CCI in the past but this seems to work better.
- For trend I test selecting from 5 MA types over 1 of 3 time frames and look for some slope to the line.
- The MAs are the four basic plus lsma.
- The timeframes are H1, H4 and Daily.
- Session filter.
- From my experience over the past 10 years with many different EAs almost all pairs traded in London session while some pairs do not do well in other sessions.
- Day of week to avoid trading news days in forward testing.
I am considering using multiple time frames for trend and maybe overbought/oversold but still need to code.
The game changer is the add to position rules and the trading session rule.
Tested on EURUSD M15 for this year showed about $150 profit without add to position and over $1200 with add to position.
This is also why the ACD method(300% gain on backtested basket for 2016) and now maybe the Strong system can become more profitable.
The basic idea is keep the losses small and wait for the grand slam home runs.
Using ACD rules the EA was successful on over 20 pairs.
Turtle rules still need to be tested but shows profit on EURUSD this year so far.