Pipruit: And how far back in history should I apply to estimate extremes? Commander in Pips: Well, I suppose – not less than 130-150 trading periods (candles). If you will take too much, then it could lead to skewing of real level of overbought and oversold. Because volatility could change in recent times, and applying too far historical volatility data that was much lower/higher could lead to wrong levels for targeting overbought/oversold levels. Pipruit: Could you give me an example? Commander in Pips: Sure, let’s take a look at this EUR/USD weekly chart # 2. We have to use a historical chart, just to see then – how estimated levels have worked in future. According to our algorithm we have to estimate extreme levels for DOSC. I prefer to use not just single value but average of 2-3 extremes. So, I’ve pointed by red ellipses the extreme highs for that historical period and the extreme lows, that we will use for calculation of levels overbought and oversold: 1. Overbought = (532.86 (10.05.1998) + 553(02.01.2001) + 559(05.19.2003))/3 = 548.28 ~548 points 2. Oversold = (-386.71 (03.01.1999) - 442(05.01.2000) – 365(08.18.2003))/3 = -397.90 ~ -398 points 3. We will use 90% from both of them. Hence our overbought = 548*90% = 493 points and oversold = -398*90% = -358 points. So, let’s mark this level on the chart. When market will approach to these areas – we will suggest that it overbought or oversold. Chart #2 | EUR/USD weekly DOSC(7) And now we will check how it works – the green vertical line shows where our future has begun.