Deep Value & LEAPS

Anyone here have experience supplementing a deep value strategy with buying long-dated call options? Naturally I’d like to test such an options strategy but am not aware of any website that has such functionality. Welcome any input, thanks.

You might consider buying call spreads as opposed to straight calls. This way you have more control over defined risk (by varying the width of the spread) and theta is no longer an issue.

When you do this, you may find shorter dated options are more manageable (they’re more liquid).

These things can be backtested at tm.cmlviz.com. Personally I’ve found selling put spreads (same thing basically) on highly ranked stocks to be very profitable.

Interesting idea. What have you found to be the optimal structure for such a spread, i.e., duration, in/out of the money, etc.?

I do this, but I just estimate the expected return of the option based on the expected return and volatility of the underlying. You actually do not need the volatility for expectation itself, but it’s helpful determining the distribution of option returns.

Let’s say you’ve got the following scenario for a long dated call option:

Current stock price = S(0) = 1
Strike Price = K = 1.1
Years to expiry = T = 1
Discount rate = r = 1%
Dividend yield = d = 2%
Option bid/ask midpoint = V(0) = .15

The risk neutral assumption which pervades quant finance would then say that the underlying and long option position have the same expected return. However, in the risky world, we are not prohbited from estimating the underlying’s drift.

Let’s now assume that, based on some ranking system, we know the stock has an expected annual return (m) of 32%, meaning that:

E[S(T)] = 1*exp((-d+m)*T) = 1.3

This provides enough information to impute the expected return on the option position since:

E[V(T)] = Max[0, S(T) - K]

So the undiscounted expected annual return on the long dated call is:

E[p(0)] = (E[V(T)] / V(0))^(1/T) - 1 = ((1.3 - 1.1)/.15)^(1) - 1 = 33.33%

or roughly the same as for the underlying. However, there is significant upside gearing: the same position bought for $0.1 yields an expected 100% ROI. You could iterate through process using the current options strip pricing to look for the desired risk versus reward – the risk/return optimal strategy will obviously depend on the market prices. Furthermore, we could take this a step further regarding the distribution of option returns if we did something with the volatility of the underlying. But, as we see, this information is not required for the taking the expectation.

So, yes, while you could backtest such a strategy using historical market data, this is not required to estimate the expected return on any given position due to the no-arbitrage pricing principle.

I sell put spreads usually 60/50 or 60/40 delta. Through backtesting I found 45-60 days to expiry and a profit target of around 40% of premium worked well, so that’s what I do. So on a $5 wide spread I’ll try to get $2.50 and cover for $1.

The ranking system I use for this is looking for quality and stability (lots of trimming of outliers). Nothing very fancy and no value factors.

Deep value or oversold stocks sound good for this kind of thing, but I was never able to make it work as well we as just picking “good” stocks in general.

Thanks everyone, that’s helpful.