OK Jem, I've got a basic backtest running. I couldn't do anything really fancy (like optimizing the parameters across all pairs) because of TS limitations. This process takes a whole lot of manual steps so I only tested a few cases.
I tested it on 5 years of data, trading 0.1 lot. Based on your rules and 30pip backtest, only 9 of 27 pairs were profitable over the 5 year period. 20 pairs were profitable in the last year. If you traded all 27 pairs, you would have gone over $50k into the red, but with a good run at the end you would have made over $13k in the last year:
SL30.gif
Two suggestions:
- 30 pips is too tight. Your win rate is low and your average trade size is small. Something along the lines of 150-175 pips works a lot better. Or, better, use ATR-based SL. 1*ATR works OK, 3-5 * ATR works better. Depending on the pair, your profits can double, win% can almost triple, average trade is 2-4x larger. If you size your position based on initial risk, the smaller SL might make more profit, but I think it's easier / more fun to trade something that wins 3x more often and has larger average trades.
- Whipsaws will kill you. You want to avoid false reversals that get whipsawed for a loss. I can think of two simple ways to do this: 1) set a reversal stop when MACD turns, rather than reversing immediately, or 2) reverse when the MACD stays steady in its new direction for N candles. The larger N is, the more you miss out on when the market turns, but the less often you get whipsawed. It seems to work best if you wait 6-8 candles!
Running with 3*ATR SL and reversing after 7 candles, in 5 years you lost $12k, and in the last year you made $9k:
SL3rev7.gif
Hm. That doesn't look much better, does it.
I think these parameters don't hold up well across all pairs, and in particular the reverse-after-7 might hurt more than it helps. Setting a reversal stop might work better but I didn't get that working yet.
I tried some simple "only trade the ones that have been working lately" filters and those didn't help at all. From what I can see, the results are kind of random, not in long streaks of wins or losses.
So: from this test at least, it looks like your approach has been working pretty well since the middle of 2011. Before that, it lost a ton. You could trade it as is, and just watch out for it to die again.
I found one thing that worked really well: trade only JPY.
Looking at the profits for all pairs, xxxJPY pairs were consistently the most profitable. (Except USDJPY, for some reason.) So here's the result of trading with 3*ATR SL, 10*ATR PT (optional but it helps), reversing as soon as MACD changes color, and trading all 27 pairs (blue) or just xxxJPY pairs (red):
JPYonly.gif
So that starts to get interesting! If you trade only xxxJPY, you have 1/4 as many charts to follow, make 1/4 as many trades, and the profits are a lot more consistent. Except the last half of 2011, you do about as well as, or a lot better than, trading all 27 pairs. You could trade 4x bigger size and make a LOT more profit with about the same risk.
You do not have the required permissions to view the files attached to this post.
) but as long as the trading history is consistent, the math doesn't care whether you're trading stocks, forex, or tulips. Look at your trade history, calculate an appropriate risk level (I use Kelly calculations), and size your trade so you risk the appropriate amount.
.] That never made sense to me & won't trade live on such accounts.
