Saturday, October 13, 2012

Cancelling trades on an exchange: When is it a good idea?

When inexplicable things happen on an exchange, many people argue that those trades should be cancelled. I think it is useful to be clear about the test to apply for this.

The key question should be: Did something foul up in the order matching software? If order matching went wrong, or if there was a systematic breakdown of connectivity to the exchange, then there is a case for cancelling trades. We'd say that persons placed certain orders, but the exchange mis-handled the orders, hence the observed series of matched trades and prices is unfair.

If the exchange and its rules worked as advertised, this reason peels away. In fact, I would argue that particularly when there is a fat finger trade or something like the US `flash crash', it is important to not cancel trades, to cement faith in the trading process.

The recent events surrounding the fat finger trade by Emkay are a good example of this line of thought. Owing to a human error,  a basket trade to sell Rs.17 lakh of Nifty was instead placed as an order  to sell 17 lakh nifties (where one `nifty' is a basket of 50 shares adding up to the present level of the Nifty index expressed in rupees). If Nifty is at 5000, then an order for "100 nifties" is an order for Rs.500,000.

Through this human error, a very large sell order appeared on the market. At that instant, everyone looking at the market would have been taken aback. What was going on? Has a huge event unfolded which some informed speculator knows about, but I do not know about? It takes nerve in that moment to be on the other side of the order. We must reward the people who did not lose their head when everyone around them was losing theirs.

When the big Emkay order came in, many of the orders which were matched were limit orders which had been patiently waiting there. This does not, in any way, change the analysis. Waiting with `deep out of the money' limit orders is a hazardous business. As an example, consider the persons waiting with deep out of the money limit orders, standing ready to buy at very cheap prices (e.g. 10% below the current market price) when the Satyam scandal unfolded. They lost money big time because the informed speculators, who understood the Satyam announcement and placed massive market sell orders, knew more than them. Waiting patiently with limit buy orders, 10% away from the touch, is not free money. ("The touch" is finance parlance for the bid and the offer price). It is a risky trading strategy.

Two trading strategies matter most in stabilising a market when crazy things have happened. Traders  have to be there ahead of time, with limit buy orders far away from the touch. The limit order book should be thick with orders; i.e. the impact cost associated with a giant market order should be low. And there have to be traders who see that the market has crashed, are able to work the phone and gain confidence that this is an idiosyncratic shock, and come into the market and buy. The more the capital and intelligence behind such trading strategies, the more stable the market will be.

If trades are now cancelled, these two trading strategies will have suffered the risk and got nothing in return. In the future, they will be more circumspect about stabilising the market. Similar considerations apply on the other side. When there are strange and large upward moves of the market, we want rational speculators who short sell and bring the price back to fundamentals. The market must be designed in a way that supports and enables this. At present, it is not [link, link].

Fat finger trades will happen. There will occasionally be strange rumours and other odd things that will make markets fluctuate away from fair price. In those situations, what we want most is for clear-headed rational speculators to put large scale capital into making money by stabilising the market. The rules of the market should reward the people who perform these roles. Trades from their orders should not be cancelled.

The Emkay story has gone well for the Indian securities markets. The market design worked as it should have. A human error was made, there was a brief market-wide suspension on the equity spot market (but the futures market continued to work). A call auction took place to discover the price, and within minutes everything came back to normal. Emkay took full responsibility for their trades and came through with the money. We shouldn't stumble in the policy analysis that follows this story.

2 comments:

  1. Media reported that the regulator was looking at mechanisms/ rules to cancel trades. I agree with you that in the circumstances resulting from Emkay this would not be a valid response, especially when all the systems worked as advertised.

    The key question in my mind is that this was a very large market order - something unusual for an institutional order. There are very simple ways to prevent recurrence before mistakes happen. We should not have to have complicated methods to deal with situations after the fact.

    Suggestion: Market orders to have a exchange validated upper limit (1) for retail orders and (2) for institutional orders. In no case should a single market order be greater than x% of the total depth of the book. ALso, large institutional orders are almost always broken up by execution algos into hundreds of smaller orders. They do not need to send in a super large order.

    In the current case the FF order swept thru the entire book leading to a huge immediate drop in LTP (the book as everyone knows contains hundreds of price points). Orders within the 20% price band are valid.

    This sweeping of the book has other un-intended consequences. It triggers stop loss orders which get executed. Will the proposal to cancel FF trades, also address unintended stop loss orders/trades - I think not.

    I do believe that a bunch of industry knowledgeable folks can come up with good working solutions, not knee jerk reactions, to solve these types of situations which have happened several times in the past few quarters.

    If we have a genuine desire to address the situation and the right experiential base, we can easily address these and other similar situations, while retaining the sanctity of the market and the matching process.

    Media of course is completely clueless, blaming the exchanges, the systems, making up "flash crashes" etc.


    ReplyDelete
  2. If there is a proposal on the table (SEBI/SMAC), it should alo be placed for public review & comment on the SEBI website before rules are formalized. Similar public reviews are invited by regulators around the world.

    ReplyDelete

Please note: Comments are moderated. Only civilised conversation is permitted on this blog. Criticism is perfectly okay; uncivilised language is not. We delete any comment which is spam, has personal attacks against anyone, or uses foul language. We delete any comment which does not contribute to the intellectual discussion about the blog article in question.

LaTeX mathematics works. This means that if you want to say $10 you have to say \$10.