A 10-level deep limit order book contains 40 pieces: 10 bid prices, 10 bid sizes, 10 ask prices, and 10 ask sizes. In liquid markets (such as the ES), the book changes hundreds of times every second second. While it is indeed possible to preserve all of this information on the historical basis, the resulting data file would be gigantic. That would make it very difficult to analyse, to backtest, and to optimize. What JBT does is that it compresses the book information in two ways:
1. The book is represented by 2 pieces (instead of 40), which are book balance and the level 1 mid price. 2. The intra-second balances are averaged, and only the resulting average 1-second balance is recorded (instead of all the intra-second changes). This technique reduces the amount of recorded information by a factor of 2000 or so. The downside is that certain details (such as the bid/ask sizes at any particular level) are lost, so it is not possible to do things like those mentioned in the paper that you referenced. On Sun, Dec 22, 2013 at 2:55 AM, toughluck <[email protected]> wrote: > The paper posted above: "Forecasting prices from level-I quotes in the > presence of hidden liquidity" is extremely interesting. > > However, in order to back calculate the 'hidden liquidity' one would need > the bid and the bid size, and the ask and the ask size. I assume one could > reconstruct the bid and ask from price based on the assumed spread, but > unless I am mistaken one can not reconstruct the bid size and ask size from > the balance. > Was there a reason to save the calculated balance quantity rather than the > more raw bid and ask sizes in the history file? > If I wanted to implement some H calculations, would I need a branch of JBT > that saved bid size and ask size in the data file? > > The other thing I found interesting, and probably controversial, is their > statement that in markets with a lot of hidden liquidity, the bid and ask > sizes (I think that implies the balance) lose their value as a predictor of > price movement. They are almost saying that less liquid markets are better > to trade than ES :P. Or put in another way, one would calculate H for > several instruments, and then trade one with low observable hidden > liquidity as it could then be predicted better (strategies would work > better). > > -- > You received this message because you are subscribed to the Google Groups > "JBookTrader" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to [email protected]. > To post to this group, send email to [email protected]. > Visit this group at http://groups.google.com/group/jbooktrader. > For more options, visit https://groups.google.com/groups/opt_out. > -- You received this message because you are subscribed to the Google Groups "JBookTrader" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To post to this group, send email to [email protected]. Visit this group at http://groups.google.com/group/jbooktrader. For more options, visit https://groups.google.com/groups/opt_out.
