So your filter condition would be Close > 20 EMA > 50 EMA for long?
I’ve tested similar strategies in the past, so I was equally surprised by your positive results 
It’s because usually nearest close in profit is much smaller than potential stop loss and this strategy needs significant win rate to breakeven. Volatility helps a bit with profit side, but still it also widens the stop loss side, so it’s a lottery. Do you have any not major pairs to run with this? Cheers!
Adding EMAs only flattens the curve, but overall results are not improved.
I’ve tried to add criterion for size of larger bar to be larger than n ATRs to avoid catching IB from small doji candles - also without any luck for long term results 
Yes, that is correct. The price needs to be above 20 EMA for a valid long entry in an up trend and below 20 EMA for a valid short entry in a down trend…
Yes, I was going to plot a backtest on AUD/JPY asap, as it is the most volatile pair, and volatility would notionally appear to be a key factor. I’ll run this and post it soon as I can.
This seems a viable tactic. In fact some people do use IB’s simply for trend-following entries.
I’ve ran for AUDJPY since 2012
Winrate ~70%
Risk to Reward ~ 2.46:1
So basically it gets barely in plus after 10 years
It seems that this is the main issue.
What if you kept a trade going for a min of two trading days? We can also try 3,4 and 5.
Or we can try to bring the SL up to the middle of the IB.
A look-back over the last 3mths on AUD/JPY is either a small win or a small loss but not the clear winner I was hoping to uncover.
The problem is indeed the average loss:win sizes - av. losers are bigger than av. winners. The difference between the two is one trade - it depends on whether the trader set a new sell order when a second IB printed while a long position was already running. In this particular case, raising the sell order to its new level would have netted a small aggregate profit. Keeping it where it was set on the initial IB nets the small aggregate loss.
Have to say I don’t use pip counts to determine the size of winners and losers. I recommend that, having determined the IB range, the buy and sell orders are then size-adjusted so that the loss in each case is a standard percentage of account capital. The impact this has on back-testing is that a loss on a 100-pip range on one IB is the same as the loss on a 50-pip range on another IB: its not twice as bad.
In any case, the raw system isn’t showing anything like the promise I initially thought it would.
But as the basic TA principle remains valid I’m sure its worth a bot if time to optimise based on this. Some of the tactics have already been posted up, see above. I need to do more work on these.
I’ve made a quick rewrite to include this and it helped a little bit on AUDJPY, but totally destroyed EURUSD 
USDJPY obviously gives much better results
Below for 2% at risk for EURUSD / AUDJPY / USDJPY
It may be tradeable on USDJPY, but I would feel uncomfortable not knowing why it worked so far only on this pair 
Absolutely. What the heck should make this work so well on one pair which isn’t on the face of it uniquely best suited in itself to this strategy?
Regarding inside bar trading, JohnScott31 forum member has a simple interesting way of trading IBs on his website which he probably shared here also. He trades them also on 8 hours bars mostly I think.
Worth checking out his strategy against this one
Ah, I thought he did but I couldn’t locate it when using Search on BP. I’ll check out the website, many thanks.
Has anyone automated this strategy for CTrader yet?
