Zool » Fri Jun 23, 2017 8:47 pm wrote:Spent a couple of hours with the indi I posted earlier and have some thoughts after it Please note I have these ideas based on how the current version of Peaky works.

Peaky only place orders when there were a recent high or low. So if we're heading up from the bottom with a recent low and long trading direction, and arrive into the top sixths that will not trigger a sell stop trade until the price makes a new high and breaks the previous sixths' high.

This is a screenshot about what I trying to say, notice the trading direction and sixths status in the EA's comments.


SInce with the upcoming PoS we will be able to trade on the bigger TF trend's retracements, it is possible that the price will not make a recent high/low, so Peaky will not send a stop order when the price arrives into the trading zone. See this 5M TF picture about what I'm talking about:


Maybe an option to let Peaky place trades even if there weren't any recent high lows, only the price moved into the trading zone, would be useful.



The second thing I noticed that the Take Profit pips has a very big impact on how long will it take to reach it, and how big would be the drawdown meantime. Sometimes only a 3-4 pip change in TP (lowering it) cut down the time from 10-12 days into 1-2 days (on the 5M timeframe). But how could we figure out what would be the optimal TP amount? We can run backtest and optimize this parameter, but that would be curve.fitting and could cause several other problems. I put a MTF ATR indicator on the chart, and found that H1 ATR(20) is usable on the M5 timeframe as take profit. We would leave some pips on the table, but sometimes there is multiple trades taken.
Pictures for illustration:



The third thing is about "those" trades. There will be opposite direction trades taken in a long trend that will "never" (well we know they will reach it eventually) reach their TP. I don't know what can be done with these. Maybe the currently available "FollowAdverseMarketWithStopOrders" option will handle some of these, but there will be ones that cannot be saved with this trick.

Here is a serious one (with somewhat recent dates):
This is a H1 screenshot. The red rectangles marks the places where trades would been taken longs on the M5 timeframe. The lines are showing the trade open times and open prices (NOT the TP levels). This is a 4400 pip drawdown. I don't know what can be done against situations like this, but I think it is worth thinking about this, because it WILL happen.

Zool » Fri Jun 23, 2017 5:55 am wrote:Steve this will be huge! Thank you for your effort to put this together, and your kindness to share it with us!

As my small contibution to this:
I've modified the Sixths_As_Bands indicator, so it can be used to visually backtest those retracement patterns (to find optimal values etc.). You wrote: "The two previous rectangles mark trades from the previous PH's; I have not measured precisely, but that looks like another +90 using a TP of 30". With this modified indicator you can see where were the sixths lines at that time (the "30 pips here probably" was good, the "And here?" wouldn't been a trade, but there was a trade just before that completed with the big drop later).

There is an input to choose between Close and High/Low price for the sixths calculation. I added it there because the original sixths as bands indicator used high/low values to calculate the sixths, but in the peak hi/lo and peaky, the close price is used to calculate the sixths (usually not a big difference but big spikes can mess things up between the 2 method).

double high = High[iHighest(NULL,0,modehi,BarCount,i)];
double low  = Low[iLowest(NULL,0,modelo,BarCount,i)];