Should you back test option strategies? Sure you should. Just don't put too much stock--or should we say "options"--in your results. Before you devote too much of your trading portfolio or emotional capital into a back-tested options trade, trade the new signal live in small lots only.
The case of the breakout trade comes to mind, with lots of ways to program a breakout signal. One of the simplest signals is a close above or below a Donchian channel. Years ago, I back tested a breakout swing trade signal based on a break and 30-minute close above or below an offset 20-period Donchian channel on a 30-minute chart. Such breakouts were a natural signal to test, since Donchian had invented the channels for the purpose of identifying breakouts.
My testing revealed that the breakout captured 54 OEX points of movement in the direction of the trade over a six-week period. If one had participated in those trades with options that had deltas with absolute values of 0.70 (70 after the 100 multiplier was applied), that test would have resulted in gains of about $3,780 before slippage, commissions and fees were deducted. That result is a bit simplistic, of course, since it didn't take into account the effect that volatility and other changes would have on an option's value. However, that premise constituted one to measure anticipated results on the system, however imperfect it was. The gist of the results was that the breakout system turned out to be wonderfully successful.
With a couple of really, really important caveats.
Enormous gains occurred when runaway trades were caught at the beginning of the breakout, which is what Donchian channels helped the system do. Those gains more than made up for the numerous losses from whipsawed trades when a breakout attempt was made but proved unsuccessful. Still, by that time, I'd been trading long enough that one look at the data convinced me that this particular Donchian-channel-based, swing-trading system wasn't the right one for me.
Those "numerous losses" were more numerous than the total number of profitable trades. Almost 52 percent of the trades resulted in losses. The losses didn't tend to be interspersed evenly, either, but sometimes occurred one after another in a series of three or four losses as markets attempted to breakout. Not all the losses were small, either.
I knew that neither my confidence nor the small trading account I had at the time--2002, when I was cutting back the size of my account due to the market instability--could withstand those series of losing trades. If I happened to start out employing the system with a trade that caught a big breakout move, my account would have been okay. That first hugely profitable trade would have cushioned the numerous small losses that were likely to occur next. If, however, my employment of the system with live trades had begun with a series of four losing trades in a row, one moderately successful one and then another series of four losing trades, I might not have had enough of my then-small trading account left to continue.
Moreover, when the system suffered loss after loss, I would likely have tinkered with it or abandoned it, thereby perhaps not benefiting from the next hugely profitable trade. That would have been true even though my own back tests would have alerted me to expect those numerous small losses in a row. I was fortunate enough to have been trading long enough to understand my trading style and foresee the difficulties when back test results collide with trading style and confidence levels. Add in such mundane but often overlooked factors as slippage, commission costs, and leaking volatility eroding the price of long options, and the negative impact on the account and my confidence would have been amplified.
Don't underestimate the importance of your trading style and confidence level when you're evaluating a back test, even when the purpose of the back test is to mechanize a trading system as much a possible. Underestimating those factors proves the down fall of many traders who adopt new systems, sometimes decimating an account balance.
The case for adjusting or not adjusting an iron condor in trouble also comes to mind. As many subscribers already know, the premise behind high-probability iron condors is that the sold call and sold put are unlikely to be violated at expiration. Further out calls and puts are bought to hedge the risk from the sold or short options and minimize margin requirements, too. The idea, then, is that the credit taken in for the iron condor will most likely stay in the account, due to the high probability that the sold options will expire out of the money.
I tend to sell my sold strikes as close to 1.5 standard deviations away from the current price as possible, but certainly outside a one-standard deviation move for the length of the trade. Assuming a normalized distribution, that means that my iron condors have somewhere over 68.2 percent and a bit under 90 percent chance of being fully profitable at expiration, usually near 85 percent. It's this high probability of profitability that makes it possible for traders to tackle iron condors. It's certainly not their risk/reward profile. Someone who sells a 10-contract iron condor with about 85 percent chance of profitability may be earning about $1,300 but accepting about $8,700 risk ($10,000 risk - $1,300 credit to establish the trade).
That high probability of success has led Steven Lentz, a frequent CBOE webinar presenter, and others to back test systems in which iron condors are set up a number of days before expiration and not adjusted. Either profits are collected at a certain time, perhaps upon expiration, or the trade is closed at a preordained maximum loss. Upon the theory that the sold strikes that compose part of the iron condor are touched more often than they're violated at expiration, some systems don't even set a preordained maximum loss. The price either backs away from the sold strike or it doesn't. The iron condors either work or don't work. There's either a gain or a loss, and, whichever it is, it's locked in at expiration. The theory is that the high probability of success means that a huge number of profitable trades will more than make up for the occasional losing trade.
Once again, however, a back-tested result doesn't always play out as expected in real life, not when trading style, confidence, slippage and commissions come into play, too. When we're talking about probabilities, we must be aware that tested or live probabilities more closely resemble theoretical ones when large numbers of trades are included. A trader who trades three iron condors a month over a period of seven years may have live trade results that are fairly close to that theoretical 85 percent profitable calculation. That trader has traded 252 iron condors during that period of time. A new trader who begins trading iron condors with one a month may find at the end of the first year of trading that seven months have produced profits while five have shown losses. If that new trader had adopted the "let it go until expiration with no adjustments" tactic, the account may be decimated by losses. Or, perhaps, unwilling to suffer another big loss, the new trader starts tinkering with the setup, deciding to use technical analysis to determine those times when the underlying "should" reverse, so it is safe to let the trade run for a while, and times when the underlying might not reverse and so the loss should be taken and locked in. Those probabilities are no longer working as well when that sort of decision making becomes necessary.
My trading style and confidence levels require that I not let a loss get too big. My husband is retired, and our ages make capital preservation my first concern when trading. Although I understand the theory behind calculations about "touch" probabilities (the probability that the sold strike will be touched before expiration but not violated at expiration) and the reasoning behind the "don't adjust or set a maximum loss" trading system for iron condors, time and experience have taught me that's not a workable system for me. It's not one that I can leave alone and experience devastatingly large, even if only occasional, losses. So, I don't mess with the probabilities, but instead always adjust at a certain point when the probabilities of success have lowered, as demonstrated by delta levels.
So, test. Test a lot. Testing can help identify pitfalls, but be aware when you're ready to go live, that testing acts somewhat like sunlight on rain-slicked roads. All that shiny sunlight can bounce off the water covering a pothole as easily as it can off a thin film covering a solid surface. Don't speed into the live trades too quickly. Drive gingerly until you're sure of the surface beneath you.