|
PureBytes Links
Trading Reference Links
|
I don't think back testing with TS would be as difficult as you say. It
wouldn't be necessary to include the price of the options as their fair
value can be calculated, and a fudge factor included to compensate for
market prices. The fudge factor could be correlated to a stock's historical
volatility.
As I recall, in TS there is a function that returns the option expiration
dates, which are the third Saturdays of each month. And as the strategy uses
options that are one or two strikes away from the money, most of the data
required to test is either available or can be synthesized. I'd expect that
determining the better strikes to arrive at the best performance would vary
for different volatilities. Perhaps something like this...
Assuming a position was opened on a Monday after expiration, and the stock
was $63. The strike price of the call to sell might be $70, which is the
second strike away from the money. ($65 being the closet strike away from
the money). And depending on one's confidence of the stock rising, the
strike price of a put might be either one or two strikes away from the money
at $60, or $55 respectively. The number of strikes away from the price could
be inputs.
The logic of the system would be conceptually be:
Buy 100 shares of a stock on the Monday after an expiration date
Sell a call that is two strikes OTM (above) the price of the stock
Buy a put that is one strike OTM (below) the price of the stock.
Assume that netting the option premiums is nil or an input value that
represents some percentage of the price of a stock.
If the price of the stock on the third Friday of the following month is
greater than the strike price of the call plus .75, sell the stock (at
market) else reset the options given the above parameters. 75 cents is an
amount above which assignment of a sold call will certainly occur. Also an
ending date, as well as a start date, should be included, as a decision to
buy or sell a stock is overridden by another system that selects a subject
stock.
Colin West
I think you've got a great idea here but testing it is oh,so complex......
WHY ?....due to the combinations of strikes and expiration months......
To effectively optimize and backtest in TS2000i, you'll need to populate
option prices as DATA(2)->DATA(50) !!
Moreover, to do it effectively, you'll need some way to create an "adjusted"
options premium and price that works on a sliding time-frame.....say 3
months. Otherwise, you'll have to program complex "rollover" rules so that
option positions are closed-out and new ones re-opened as expiration
approaches.
TS2000i has all of the tools to do this......it's something I had always
THOUGHT about after reading about the terrific successes of some of the
market-neutral option traders......some claiming 90%+ win-rates.
> -----Original Message-----
> From: cwest@xxxxxxxxxxxx [mailto:cwest@xxxxxxxxxxxx]
> Sent: Friday, May 25, 2001 1:08 PM
> To: Omegalist
> Subject: RE: need some ideas to (dis)prove an assumption
>
>
> Mike,
>
> It isn't my intention to intentionally change the initial proposition! It
> simply evolves as respective interpretations eventually reach a common
> understanding. Winston Churchill got it right. We're divided by a common
> language.
>
> Here's an example of one of today's trades. I bought APC at $63.50, sold a
> Jul 70 call for $1.80 and bought a Jun 60 put for $1.25. (Not surprisingly
> I'm a little underwater at the moment excluding commissions). The decision
> to buy APC was based on its fundamentals, but that's another story.
>
> At each option expiration date, if the sold call isn't assigned,
> I'll reset
> the collar for as long as I hold the stock. If I'm assigned, then
> my return
> is satisfactory (to me at least). It's a little better than a 1:2
> risk:reward. If APC falls below the strike price of the put, then I'll
> reassess the position at the next expiration date. This strategy
> has worked
> well over the long haul, particularly as it quantifies risk and ensures
> acceptance of my demeanor in the local community. BTW, I have
> acquaintances
> who wished they'd put a collar around their stocks a year ago!
>
> Let's say it's my assumption, although I believe it's shared by many, that
> the price of a stock infrequently rises by more than the
> difference between
> the price of a sold call at one strike above the money and the price at
> which the stock was purchased, within 30 days of purchasing the stock or
> until the time of the nearest option expiration. In the above example, its
> two strikes above the money. What I'm endeavoring to discover is
> the degree
> of merit of the assumption, and interpret it in a (TS) system.
>
> Colin
>
> this "assumption" of yours is a strange beast for sure, because,
> chameleon-like, it seems to change its color with each new post. First it
> was about the time decay of options, then about the behavior of a collar,
> and now the topic is the volatility of stocks. Granted that there is a
> connection between all of them, yet these problems do differ from a
> theoretical viewpoint, as well as in practical application.
>
> Be this as it may, here is a rough-and-ready test of the assumption below
> which is relatively straightforward; a simple spreadsheet should suffice.
> Load a stock's daily OHLC historical data for as far back as you
> care to do
> your test. For every day during this period have the program check whether
> the MaxHigh of the next 30 days was greater than today's price plus strike
> distance, or not; and do the same for the downside. You have to take into
> account the different strike distances that vary with the price of the
> stock. Count the number of yeas (strike was exceeded above or
> below) vs. the
> number of nays for this stock, and it should give you a good first
> impression. You might do this for a universe of stocks such as the SP500,
> and calculate an average.
>
> Of course, this primitive method can be much refined with all the
> weapons of
> descriptive statistics at your disposal, but I won't go into that here.
>
> Hope this helps, and I have correctly understood your assumption
> this time.
>
> Best,
>
> Michael Suesserott
>
>
> -----Ursprüngliche Nachricht-----
> Von: cwest@xxxxxxxxxxxx [mailto:cwest@xxxxxxxxxxxx]
> Gesendet: Thursday, May 24, 2001 22:47
> An: Omegalist
> Betreff: RE: need some ideas to (dis)prove an assumption
>
>
> Mike,
>
> I appreciate your input, and believe I clearly understand it, but the
> essence of what I'm attempting to (dis)prove is an assertion that stocks
> infrequently rise by more than the equivalent of one strike above their
> price within 30 days. That is, on 4/23 the price of a stock was
> say $50, and
> by 5/18 it was less than $55.
>
|