David Varadi recently posted an idea for a Parabolic Stop and Reverse (PSAR) oscillator. Full details here:
Here’s a summary:
1) Record the days in a Parabolic SAR long or short position until the trade reversed
2) Take the 100-trade percentile ranking of the days since the current position was initiated versus past trades. This is the PTO.
Alternative: you can do this separately for long trades versus short trades to create two oscillators PTOup and PTOdown. Use only 50 trades to calculated the percentile ranking for each.
This was relatively easy to try out thanks to a data export and good old excel. I looked at the second variation – Separate up and down PTO oscillators.
Results:
I wanted to cut to the chase so looked at more extreme measures. For Up PTO, these are readings below 0.25, which means the number of days that the price has been above the parabolic stop (up trend) is in the bottom 25% of readings for the last 50 days. Likewise, for the Down PTO, I looked at readings over 0.75, that is days when the market has been below the parabolic stop (downtrend) for some time, with the down PTO in the top 75% of readings for the last 50 days.
Trade entered at the close. No frictions/ costs etc. Just top level analysis for now. From 2000 on the S&P 500 cash. Long only.
Here are the scores on the doors:
Up PTO 1267 trading days. Total 42.37% return. Average 0.03% per day. Down PTO > 0.75: 817 trading days. Total 49.17% return. Average 0.06% per day. There is some overlap between these readings, but they seem to say that the best returns come when at the start of up trends before they are long in the tooth and also when down trends have been dragging on a bit. More specifically, the best returns come when down trends have been dragging on a bit. Common sense you might think, but its interesting to see it quantified. How about the opposite? Up PTO > 0.25: 1553 trading days. Total -29.77% return. Average -0.02% per day. Down PTO 2003 trading days. Total -37% return. Average -0.02% per day. Here’s how the returns look: Conclusion: The idea certainly seems to have merit has measured in this way. I might have a look at how it interacts with the DVB next.
Long if Up PTO
Long if Down PTO > 0.75 out of market otherwise.
Worth noting that a good chunk of the PTO out performance happened at the beginning, near 2000.
Related posts:
