1. The Quest for Efficiency November 7, 2006 Hide-and-Seek in the Market: Placing and Detecting Hidden Orders Rudy De Winne (FUCaM University - Mons - Belgium) & Catherine D’Hondt (EDHEC Business School – Lille - France)

2. The Quest for Efficiency November 7, 2006 OUTLINE  Motivation and context  Research questions (2)  Data  Methodology and results  Conclusion

3. The Quest for Efficiency November 7, 2006 MOTIVATION Most trading systems moved towards greater transparency… –Automation enables the dispatching of information in real time –Regulation? But the use of hidden orders becomes a widespread practice Euronext, Toronto, Sydney, Nasdaq, …

4. The Quest for Efficiency November 7, 2006 TRANSPARENCY LEVEL on Euronext What market participants see on their screens The displayed quantities available at the different bid and ask prices The number of orders contributing to the depth at a bid or ask price What they do not see on their screens ID codes Hidden quantities (from iceberg orders)

5. The Quest for Efficiency November 7, 2006 RESEARCH QUESTIONS  Why do traders hide their orders ?  Pardo & Pascual (2006), Anand & Weaver (2004)  How do other traders respond to the detection of hidden depth?  Griffiths, Smith, Turnbull & White (2000), Ranaldo (2004), …  Do traders also use information they can infer by monitoring LOB movements (presence of hidden orders) ?

6. The Quest for Efficiency November 7, 2006 How to guess the presence of hidden depth? Qty BidBidAskQty Ask 40050.00 50049.6052.20 25049.5052.30300 49.3052.60100 30049.0053.00700 52.00 300 500 H_Ask 2500 1250 0 0 1000 H_Bid 0 1100 0 1000 550 Market buy order for a quantity of 300 shares 52.00 400 ?

7. The Quest for Efficiency November 7, 2006 DATA Public data: orders & trades + some additional private data Limit order book data: both displayed and hidden quantities at each limit price in the whole order book Sample: CAC40 stocks October to December 2002 (64 trading days)  More than 12 million of LOB states (continuous session)

8. The Quest for Efficiency November 7, 2006 Why Do Traders Use Hidden Orders? Traders are more likely to hide their limit orders when the order size is large relative to the prevailing displayed depth

9. The Quest for Efficiency November 7, 2006 Why Do Traders Use Hidden Orders? Market members are less likely to hide part of their own limit orders than clients because they monitor the market more actively [Anand & Weaver (2004)]

10. The Quest for Efficiency November 7, 2006 Why Do Traders Use Hidden Orders? Traders tend to hide when they submit non marketable but competitive limit orders

11. The Quest for Efficiency November 7, 2006 Presence of HO at the best quote When some hidden orders are detected at the best Bid/Ask, they are thought to be present during 9 successive states !

12. The Quest for Efficiency November 7, 2006 Our order aggressiveness classification Category 1: orders that consume the whole displayed depth at the best opposite quote Category 2: orders that consume a part of displayed depth at the best opposite quote Category 3: orders improving the best price on the same side Category 4: orders increasing the depth at the best quote Category 5: other orders How do traders respond when they discover hidden depth?

13. The Quest for Efficiency November 7, 2006 Methodology Ordered-probit model Latent equation : Ordered response :

14. The Quest for Efficiency November 7, 2006 Aggressive orders are more frequent when bid-ask spreads are narrow [Ranaldo (2004)] RESULTS: Order aggressiveness

15. The Quest for Efficiency November 7, 2006 The deeper the market on the same side, the more aggressive the subsequent orders. The deeper the market on the opposite side, the less aggressive the subsequent orders. RESULTS: Order aggressiveness

16. The Quest for Efficiency November 7, 2006 Orders are more aggressive when hidden depth is discovered at the best opposite quote See Pardo & Pascual (2006) RESULTS: Order aggressiveness

17. The Quest for Efficiency November 7, 2006 RESULTS: Order aggressiveness When discovering hidden orders at the best quote on the same side, members respond differently when they trade for their own account or not.

18. The Quest for Efficiency November 7, 2006 CONCLUSION Euronext members hide their limit orders when the order size is large, the price limit is competitive and they trade for their clients. Traders are able to infer the presence of hidden orders by monitoring LOB movements and they adapt their behavior  Traders do not associate hidden orders with informed trading (depth improvement)  Development of algorithmic trading plays a key role in detecting hidden orders  Hidden depth is important to understand traders’ behavior !