how do you define the price?

L

limbonic

This might be a FAQ, but given a historical data feed like opentick (specifically, quote and trade tick streams), how do you define "price"?:

- bid/ask midpoint
- last trade

There is the additional complication of multiple trading venues: the primary listing exchange, regional exchanges, ECNs, etc. What do you find most meaningful for your models:

- primary listing venue
- NBBO, or some reasonable approximation of the US "composite market"

My interest angle here is to try to model historical price relationships between multiple instruments (factor models, that sort of thing) at intra-day frequencies and so what "price" is requires some thought before blindly storing a few hundred gigs of data.

(I am not interested in charting/OHLC data for multiple reasons, one of them being that OHLC data is derived from the original raw data which is fundamentally quotes and trades.)

Thanks.
 
If this question is too basic, please point me to where I can find some old discussions, etc.

I would imagine everyone has to answer what "price" is for themselves when starting to build a trading platform...
 
Quote from OTCkrak:

last price / close
More...

This leaves a few questions (relevant at very high frequency):

1. is the last price more meaningful than bid/ask midpoint?

Think about this: the last price may have been a market order and whoever took liquidity paid the spread cost. Depending on whether the market order was a buy or a sell, the price will bounce between bid and ask somewhat. Just looking at the last price only seems sub-optimal, since it has this spread error in it.

2. if you are reducing prices of several trades that occurred within a certain interval (say, 1 min) to just one representative price, why do you think the *last* price is the most representative? Why not, say, the average over 1 min?
 
Quote from limbonic:

This leaves a few questions (relevant at very high frequency):

1. is the last price more meaningful than bid/ask midpoint?

Think about this: the last price may have been a market order and whoever took liquidity paid the spread cost. Depending on whether the market order was a buy or a sell, the price will bounce between bid and ask somewhat. Just looking at the last price only seems sub-optimal, since it has this spread error in it.

2. if you are reducing prices of several trades that occurred within a certain interval (say, 1 min) to just one representative price, why do you think the *last* price is the most representative? Why not, say, the average over 1 min?
More...

You ask a very basic question that is hugely important and difficult to answer. Most systems traders (from what I can gather) use last_price as the "current price." For what I'm building, the last_price is used in the analytics, but the current bid (for a sale/short) or offer (for a buy/cover) must be in line with the system parameters for a trade to take place. So, to answer your question, I'd have to say a mix of last and bid/offer is the real price.
 
Quote from polr_trader:

You ask a very basic question that is hugely important and difficult to answer. Most systems traders (from what I can gather) use last_price as the "current price." For what I'm building, the last_price is used in the analytics, but the current bid (for a sale/short) or offer (for a buy/cover) must be in line with the system parameters for a trade to take place. So, to answer your question, I'd have to say a mix of last and bid/offer is the real price.
More...

Thanks. I've discussed this question in several places and have come to a conclusion that is similar to what you are saying. The "price" is what your expectation would be in an upcoming trade -- because this is how you are going to "mark to market". As such, you ought to be building a model for the expected price in such a transaction, taking into account such inputs as the last trade price, the bid/ask quotes, volatility, whether you will be taking or supplying liquidity, etc etc etc.

But I've found it odd that nobody would answer the question for a while -- after all, it is "a very basic question" :P
 
Quote from limbonic:

Thanks. I've discussed this question in several places and have come to a conclusion that is similar to what you are saying. The "price" is what your expectation would be in an upcoming trade -- because this is how you are going to "mark to market". As such, you ought to be building a model for the expected price in such a transaction, taking into account such inputs as the last trade price, the bid/ask quotes, volatility, whether you will be taking or supplying liquidity, etc etc etc.

But I've found it odd that nobody would answer the question for a while -- after all, it is "a very basic question" :P
More...

I think nobody answered it for a while because it *seemed* too simplistic. It's one of those questions that would elicit a response such as: "Well, that's obvious! Price is...uh...er...well...I'm not going to answer that stupid question!"

The problem with using last price as the "price" is that it's in the past. It doesn't exist. It's similar to a discussion I find myself having with friends all the time: Something is worth whatever someone else will actually pay for it (i.e., in the future). I don't care if a house sold for $2mil yesterday. Maybe the buyer was willing to drastically overpay, and the next buyer is willing to pay only $1.6mil.
 
Quote from polr_trader:

I think nobody answered it for a while because it *seemed* too simplistic. It's one of those questions that would elicit a response such as: "Well, that's obvious! Price is...uh...er...well...I'm not going to answer that stupid question!"

The problem with using last price as the "price" is that it's in the past. It doesn't exist. It's similar to a discussion I find myself having with friends all the time: Something is worth whatever someone else will actually pay for it (i.e., in the future). I don't care if a house sold for $2mil yesterday. Maybe the buyer was willing to drastically overpay, and the next buyer is willing to pay only $1.6mil.
More...

*yawn*

Don't try to act smart by making this over-complicated.

OP is answering his own question with his post. He can only answer for himself.
 
Quote from TSGannGalt:

*yawn*

Don't try to act smart by making this over-complicated.

OP is answering his own question with his post. He can only answer for himself.
More...

Of course he can only answer for himself. But he invited discussion and other points of view. Isn't that why we're all here?

And believe me, I know far too much about how much I don't know to try to act smart. :D
 
You must log in or register to reply here.
Back
Top