Using the Mid-point calculation for technical indicators

http://www.forexstrategyblog.com/midpoint-calculation/

I recently posted an article on my blog testing the difference between using the closing price and the median price for an EMA cross over strategy. The results were actually pretty different, the median price moving seemed, for this test, to predict large market movements better than closing price alone. I am curious if any other traders have any experience with using the median price or if everyone just uses the closing price.

It would be nice to see the same backtest with EMA 10/19 close, or better EMA 9.5/18.5 if your backtesting plateform accept it. Why? Because the mid price is a filter in itself and it adds half a bar of lag. This could show if the additional filtering is responsible of the difference (less whipsaw) or if the mid price is more informative than the close price. It would also be nice to see a series of backtests with various pairs of EMA to check if the improvement is due to the mid price or just a lucky realization with these two periods for E/U over this dataset.

I can't do 9.5 and 18.5. I did the 10 19 close price test and it came out closer to the 9 18 close price. If you do that math the median price is much different than just a slower EMA. If you get a big candle the closing price average will be moved much farther than the median price. The two will also be much different if you get a large hammer, hanging man, tombstone doji, or dragonfly doji because all of these candles have a closing price that is far away from the median price.