This is a post about testing random entries.
Many an hour is spent by technical analysts and systems traders agonising over the best entry techniques, the best systems. But frankly, is all the effort worth it?
Ed Sekota's Whipsaw Song contains in its lyrics his Rules of Trading:
- Ride Your Winners
- Cut Your Losses
- Manage Your Risk
- Use Stops
- Stick to the System
- File the News
What if the system follows all 6 of these rules but uses random entries to enter a trade? Can such a method possibly be profitable?
If random entries can produce a profit then it ought to give practitioners a great deal of confidence in the basic tenets of good trading set out above. And it is surely a great deal more difficult to over-fit such a system to the data.
Is this topic just about trend following? No, I don't think it is. "Stick to the system" sounds like trend following but it does not have to refer to a trend following system per se. Perhaps the first two rules are the most essential: ride your winners and cut your losses. Surely any successful type of trading has to abide by these first two rules? So this thread is applicable to trading and investing as a whole and not merely to trend following.
So what does the basic code provide?
- Entry. If there is no position in a particular instrument, take one: either long or short on a random basis. You always have a position in each instrument in the portfolio, either long or short.
- Exit. A trailing 5 ATR stop based on a 20 day simple average ATR. When a position is stopped out it will immediately be re-entered as per 1 above.
- Risk Management. Initially this will simply consist of limiting the initial position size on entry. On entry 1% of the portfolio will be risked (volatility adjusted fixed fractional position sizing based on the distance to the stop). No other attempt will be made to limit risk.
- The portfolio consists of a mixed assortment of over 100 futures. No attempt made to balance the portfolio in terms of sector or correlations analysis.
I ran 50 tests. I could have run 500 or 500,000 and perhaps should do so when I have the time. Or leave it to others to do so. Perhaps the statistics would be very different.
- No interest earned on unused cash balances.
- Slippage: 7% further explanation provided if required).
- Commission per contract: US $7
- Start Date: 1st January 1990
- End Date: 1st February 2013-02-01
- Starting Capital: US $ 1,000,000
Statistics on the 50 tests run came out as follows:
Percentage of tests profitable: 94%
I exclude the unprofitable test for the rest of this summary: each lost 100% of the capital.
Max Total Equity Draw Down:
Longest Drawdown (months)
Number of Trades:
Std Dev (annualised monthly)
Winning Trade Duration (days)
Losing Trade Duration (days)
Is there work remaining to be done before any conclusions can or should be drawn? Yes, a great deal. Extra statistics need to be included such as the win/loss ratio. The parameters need varying and stepping. Random portfolios need to be tested. Further risk control need to be experimented with. Long term trend filters need to be added. The definition of random needs to be varied so that the "system" need not necessarily always be in the market. Time period analysis needs to be conducted: how did this "system" fare in the past two ghastly years (not very well on the current parameters). And so forth.
Discussion, comments, derision, disbelief, conversation welcomed. Above all: "participation"!