In addition to the previous post, I hope it is not boring?
I talk about trailing, to communicate experiences and inspirations for engaged coders.
Most of the most successful senior traders (e.g. traders world-championship) did bother themselves very intensive with money- and order- management.
Traders can use stops (SL, BE, TP) for protect orders, running into loss or profit.
The common known methods are:
- pyramiding or scale in and out, stacking trades, when a trade run into profit
- take BE and partial profit contemporary
- trail trades as long as possible (with open end)
But considering, most of the money-management-theories did develop in the area of trading Shares. Please refer e.g.to the “Kelly-criterion” or the “fixed-ratio-method”. One other method was the famous “turtle-experiment”.
Basically FX is a 50:50 game, because in comparison to the stock market, there are only few fundamental data to predict the very fast market movements with an edge. On the other side, we should consider the high leverage, used by the small traders trading FX.
In comparison: The commercial hedgers and large traders are not focused to trade 4 or 5 digit-movements with high leverage.
The big guys buy and sell futures (options) for international companies to compensate currency fluctuations. With simple words, they do not have the high risks, concerning price oscillations on the 4th digit. We should that keep in mind.
In the field of FX, there are experiments with random entries and exclusive money management.
These methods were profitable, but did work with sometimes very big drawdowns.
Therefore in summary, it might be worth to bother with the a.m. issues.
I did already mentioned, that some kind of automatic trailing might be complex and needs dynamic processing. E.g. if price will move at the daily maximum price-movement, the SL-levels, the volatility and momentum will slow down or vice versa accelerate. In consequence the trailing stop-distances has to be adapt dynamically.
Have a glance to the functionality and description of the “TrailingRunningKnot”-EA in the b.m. link.
For additional information about trailing EA´s please refer to:
http://forex-strategies-revealed.com/mt ... g-stop-eas
Below you will find attached a collection of some trailing EA´s, but take care before use, due to build 6XXX malfunctions, I did not checked.
http://www.forexfactory.com/showthread.php?p=4618597
Anyway, in most of the trade-manager, we still miss some conditions-tasks or call it artificial intelligence.
For example…
Quoting Milanese:
“ UseHighLowForMovingSteps=True if this is set true the trademanager will use the distance form H/L of the last five candles to move the SL this for a better managing of open positions in win, so the SL jumps in profit will be analog to the price movement, here is too used BEPercentSLIncrement for the % calculation of the high/low distance.
Now the trademanager has a certain "intelligence" in moving the SL.”
Milanese (also “Steve” MPTM) did implement the a.m. trailing module (algo), considering the market move (HL-candle-control for a defined 5 candle period) and in addition supervising the main trend with the LWMA240.
For myself, for supervising trends during daytrading and for scalping, I`d prefer in almost every timeframe the SMA20 up to SMA60. These MA´s are reliable, depending on the appropriate volatility of the appropriate pair and timeframe, especially during the starting period after entering an order.
BTW: Followers of Heikin-Ashi-candles prefer two MA´s (SMA5 high and SMA5 low) in the faster timeframes (e.g. M15). As long as the HA-candles moves with the trend, resp.in accordance with the SMA-direction, not touching the opposite MA with the body, the trend will be still stable and continued. In conjunction with a volatility-channel indicator (for locating turning points), such as Bollinger or TMA and considering support/resistance-levels in the upper timeframes, it might be worth a try.
For order-management and especially trailing we basically have to consider further basic indications (conditions) for decision making:
- TrendFilter (e.g. Moving averages, CSS)
- TimeFilter (by trading-hour, duration- or expiration-time)
- Money-Management Control
- Volatility-Filter (such as ATR, Bollinger, TMA …)
- Volatility control by hourly maximum volatility or maximum volatility per weekday
- Momentum
- Volume-Filter*
- Oscillators like Stoch, RSI, Williams%, CCI, MACD or whatever choice
- Candle-pattern and/or formation-signals
- Fibonacci- and /or pivot-levels
- levels of support and resistance
*Volume: amount of transactions, not real traded volume.
For volume please refer to the future-market.
The a.m. list is only a short briefing of indications and used for example.
Some of the features are already realized in TM´s, but sometimes not easy to handle.
BTW: Please take into account, that there exists a general misunderstanding referring independent events in the field of probability calculations resp. future predictions.
To assume that more indications used, the success-rate in reference to the probability would increase, is erroneous.
With simple words, independent probabilities have to be multiplied, instead of adding.
An alternative might be a decision model maker, like a todo-list.
Interested traders please google for decision-making models and details, because it goes beyond the scope of the issue.
With other words, we can use any criteria, we can evaluate for e.g. a long- or short-move with “1”.
It is even not mandatory to ask exclusively for measurable values.
Examples:
E.g., we ask if the price is above the SMA20.
If it is, than count = 1 for long and vice versa for short = 1, using two global variables. Any oscillator indicates short, than count = 1 for short. If the price-movement did arrived long an expected maximum value (turning point), than Count = 1 for long. Analog procedures, whatever events we like to consider for long or short.
In the next step by considering the success-/hit- rate concerning different conditions, we can weight differently each criteria, instead of “1” use “2” or “3” by the amount of recorded hits, any event did perform in the past (like the fuzzy-logic method).
An advanced operation might be to weight the criteria by percent (after empirical data analysis).
We even can recognize certain relationships, which maybe occurs with a certain probability in combination.
Unfortunately Empty4 does not provide a data-recorder and data-output for certain statistical purposes and Analysis. To misuse the strategy tester for similar tasks will not supply valuable results.
In reference to the MPTM-, the hidingpending- and allpairtradingmanager-EA, one of the question remains, if it might be an advantage, to use TM-EA´s separately for special purposes (KISS).
A further question, I´d like to ask: Is there any Empty4-build-in-function to open resp. to access to a further EA or script automatically (to interact each other), if a defined condition will match?
Have a nice weekend
Over60 |
