CrazyIvan FX – The Nuts and Bolts
Mole and I are in the final stages of beta testing, and the only things left to decide are which timeframes we offer (360, 480, 720) and whether we offer it unfiltered (faithful to the original Ivan) or filtered. Please post your ideas in the comments section.
Crazy.Ivan.FX is a set of proprietary filters applied to a subset of Ivan Krastin’s setups (Gap Open, Inside Period, Fakeout, Retest, Retest Variation, Trend Trade, Failed Hammer, Failed Shooting Star). The purpose of these filters is twofold. One is to stop trading in whipsaw tape and reduce drawdown. The second purpose is that in strongly trending high volatility tape we only take trades in the direction of the trend.
It should be noted that High Volatility/Trending tape behaves like a completely different system, with win rate 64% and expectancy before slippage of .67. This is excellent and it’s the one we are using on the 30-min charts, but unfortunately it only happens in exceptional tape, around 15% of the time and abnormally distributed. An example would be AUD/USD earlier this year during it’s outlier move.
Bottom line on average you are going to get approximately 130 trades/year on 360, 100 trades/year on 480, and 65 trades a year on 720, at an expectancy between .15 and .5 depending on market phase with an average of between.25 and .3 before slippage for most years. Unfiltered will generate 300 trades per year per instrument on 360min, 220 trades per year per instrument on 480 min charts, and 150 trades a year on 720, with a slightly lower expectancy of .2, with dramatically worse drawdown characteristics than filtered.
Average winning trade is 1.26R while average losing trade is -.75R. The win rate hovers either just above or just below 50% while the return is, in truth, accounted for the fact that the wins are bigger than the losses.
This strategy is breakeven on a 38% win rate due to the wins being bigger than the losses. It is not unusual for several months of above average win rate (in the 55-60% range) followed by several months of below average win rate in the 35-40% range. Long term win rate is approx 52% unfiltered, and filtered depends entirely on the composition in volstat/stretchstat terms.
This is going to work out at between 32.5R and 39R / year for 360 and 24.375 and 29R / year for 480. There is no appreciable difference between instruments, ie AUDJPY is not better than ZB, it all comes down to market phase. The more clean trending market we get in either direction in a year the better the expectancy is.
The primary difference between the filtered and non-filtered is that the drawdowns are more shallow and do not last as long. Peak/Valley drawdown is 10R Max, compared to 15R unfiltered. It is extremely rare to get more than 4 or 5 losses in a row filtered.
Unfiltered it works like this:
360 is between .15 and .35 expectancy with an average of .2, which on 300 trades / year gives 60R / year with a peak/valley drawdown of 15 R and max losers in a row of 9 (most rats could not handle large losing streaks in a row)
480 is the same on 220 trades a year which is 44R / year with 15R drawdown (drawdowns can last 3 months unfiltered)
A representative example of the 480 is 2011. We have seen years *much* better than this and also worse than this. Note the histogram below, showing wins >3R are rare.
Here is the google docs spreadsheet of another sequence of results which would be typical. Note that typically winners and losers are clustered together.
So now it’s up to you. When it comes to the 360 vs. the 480 – which timeframes would you prefer? And do you think the 720 will even be necessary given that the 480 will offer three roll overs per day (7:00am, 3:00pm, and 11:00pm EDT)? Myself I’m leaning towards the 480 min as it is the best tradeoff of expectancy/lifestyle in my opinion.