1
$\begingroup$

I am looking at creating an ML model to price an exotic equity option which has a barrier where the buyer is paid out if the barrier is crossed, and multiple observation dates where the price is checked if it has crossed the barrier.

I am struggling to figure out the best way to represent a dividend schedule. It is important that this is accurate as ex-dividend dates close to observation dates can significantly affect the price as the underlying generally falls proportionally to the dividend paid out.

If you have any ideas or have done this before, any help is appreciated!

Improve this question
$\endgroup$
4
  • $\begingroup$ Why would you use an ML model? Surely this could be done with Monte Carlo or a Binomial Model, then adjusting the stock price at the ex-dividend date in either case $\endgroup$ Commented Oct 13, 2024 at 13:58
  • $\begingroup$ Monte Carlo takes way too long for our use case. We already have a Monte Carlo model but quoting complex products takes too long. We are fine sacrificing some accuracy if we can get a fast quote. $\endgroup$ Commented Oct 13, 2024 at 16:25
  • $\begingroup$ A ML model will not do you any good as you'll likely have to normalise prices, which raises the question of how you should normalise the prices, which in turn will cause you more headaches. If a Monte Carlo method is taking too long, try a Binomial Model and adjust the price at the ex dividend date. You can choose a smaller step size to make it more accurate and use a Boolean array alongside the Binomial Model of stock prices to determine whether the barrier has been breached. Check the Wikipedia page on barrier options. There are plenty of methods better than this $\endgroup$ Commented Oct 14, 2024 at 17:37
  • $\begingroup$ These exotic options are already priced in percentages so they are price normalised. ML model is the only way we are going to get the speed we need $\endgroup$ Commented Oct 16, 2024 at 0:42

0

Reset to default

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.