[New series!] [Consistent Losing Strategies] 34 EMA Scalping//---------------------------INTRO------------------------------
Hi All!
Let me introduce myself as a semi-successful forex trader & lover of automation.
I've taken to algo trading and have been hunting down strategies (that usually use indicators) to automate, backtest, and hopefully implement in MT4.
Unfortunately, most strategies are complete bulls*** and the select cases that are shown to "prove" success are limited.
These strategy sources often do not provide useful analytics either.
I want to change that approach to trading! We can really benefit each other and the community by being methodical about backtesting
as well as evaluating our results with some kind of scoring heuristic.
As for what that standardized process looks like..well I'm still working on it.
I'm pretty much on Tv for multiple hours of the day, screening strategies via Pinescript and I'd like to start sharing my progress!
This is a new series I'd like to start on consistently losing strategies. I'll make all the code public, so if you think I've made a blunder
or approached a problem the wrong way, then drop me a DM or paste your fix into the comments.
//---------------------------STRAT------------------------------
34 EMA Scalping strategy (ref. forextradingstrategies4u )
How you're supposed to trade it:
BUY:
1. Market is in an down trend as shown by the 34 EMA
2. Price breaks above a downwards trend line
3. Price breaks above the 34 EMA
4. Look for a very bullish candlestick or chart pattern
SELL:
1. Look for the 34 EMA to show we are in an uptrend
2. Price breaks below an upwards sloping trend line
3. Price breaks below 34 EMA
4. Look for a bearish candlestick or a chart pattern
//---------------------------CONC------------------------------
Q: Why does it fail?
A: I believe this strategy relies too much on subjective input (aka, trendlines).
Q: Why does it fail as an algo?
A: The 34 EMA is no more predictive than any other EMA, although it does a good job at filtering out noise.
Q: Should I try it out?
A: No, it's trash. This is the proof that it is trash.
Algo
MACD, backtest 2015+ only, cut in half and doubledThis is only a slight modification to the existing "MACD Strategy" strategy plugin!
found the default MACD strategy to be lacking, although impressive for its simplicity. I added "year>2014" to the IF buy/sell conditions so it will only backtest from 2015 and beyond ** .
I also had a problem with the standard MACD trading late, per se. To that end I modified the inputs for fast/slow/signal to double. Example: my defaults are 10, 21, 10 so I put 20, 42, 20 in. This has the effect of making a 30min interval the same as 1 hour at 10,21,10. So if you want to backtest at 4hr, you would set your time interval to 2hr on the main chart. This is a handy way to make shorter time periods more useful even regardless of strategy/testing, since you can view 15min with alot less noise but a better response.
Used on BTCCNY OKcoin, with the chart set at 45 min (so really 90min in the strategy) this gave me a percent profitable of 42% and a profit factor of 1.998 on 189 trades.
Personally, I like to set the length/signals to 30,63,30. Meaning you need to triple the time, it allows for much better use of shorter time periods and the backtests are remarkably profitable. (i.e. 15min chart view = 45min on script, 30min= 1.5hr on script)
** If you want more specific time periods you need to try plugging in different bar values: replace "year" with "n" and "2014" with "5500". The bars are based on unix time I believe so you will need to play around with the number for n, with n being the numbers of bars.
