1
$\begingroup$

The square root law is shown here: Market impact, why square root?

Let's say I want to execute 100 lots. But I have never executed before so I have no idea what n is historically. How would I determine what C is? Or is there a way to get a historical n?

$\endgroup$
3
  • 3
    $\begingroup$ Different Investment Banks or consulting firms claim that they have estimates of the parameters of Market impact equations for a wide range of stocks and will let you use them if you do business with them. $\endgroup$
    – nbbo2
    Commented Sep 24, 2020 at 11:55
  • 2
    $\begingroup$ I suppose your question is how the parameters are estimated? $\endgroup$
    – nbbo2
    Commented Sep 24, 2020 at 13:33
  • $\begingroup$ Yes. Well I am hoping to gain insight on the process of estimating the parameters. Would it make sense to input "n" as every trade that occurred and delta(p) as the price change immediately after the trade? I don't have access to bank/consulting estimates nor the money to pay for it. $\endgroup$
    – confused
    Commented Sep 25, 2020 at 5:30

1 Answer 1

1
$\begingroup$

A lot of the literature relies on estimating impacts of large orders (n), typically from major funds, that are split into child orders and executed over some period. Usually this data is proprietary and difficult to replicate. The metaorders used in this paper https://arxiv.org/pdf/1412.2152.pdf are one example (see footnote 3). This paper explains their methodology and cites many older papers [5,6,7,8] which explain the concave function and how they captured it. Unfortunately, this makes it very difficult to get a historical C without having more information about large orders and how they are split.

$\endgroup$
1
  • $\begingroup$ I'll take a look thanks. I was wondering if one could just calculate it with available trades data and the price movement immediately afterwards. And instead of doing a daily volatility, do like a smaller interval. $\endgroup$
    – confused
    Commented Sep 25, 2020 at 5:32

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.