Norwegian Day Traders Convicted For Manipulating Computer Trading System

An anonymous reader submits news of the conviction of two Norwegian day traders, Svend Egil Larsen and Peder Veiby, who were on Wednesday fined and given suspended sentences (Norwegian court, Norwegian document) for cleverly working out — and cashing in on — the way the computerized trading system of Interactive Brokers subsidiary Timber Hill would respond to certain trades. They used the system's predictable responses to manipulate the value of low-priced stocks. The pair have gotten some sympathetic reactions from around the world, and promise to appeal.

Comments

Intriguing

[The Dodger]

I'm not a lawyer and I don't speak Norwegian so I can't read the court document to find out exactly what happened. I am, however, an electronic trading specialist and I've also been a trader at a big American investment bank (one that didn't go bust, by the way, despite my best efforts).

Rumour has it that these guys realised that there was a flawed algorithm (which turns out to have been operated by Timber Hill) making a market in illiquid shares, which set its quotes based either on the prices at which recent trades in those shares had been done, or on the algorithm's own position in the stock.

To give some background: if you are making a market in a stock, that means you are prepared to buy from people who want to sell and sell to people who want to buy. Unless you're feeling particularly generous, you want to buy at a "low" price and sell at a "high" price. In liquid markets (i.e. where there are lots of people buying and selling), you can typically rely on the market mid price (i.e. the best bid plus the best offer, divided by two) and "spread" off that (e.g. add a cent to it to get your ask, subtract a cent from it to get your bid). As the market (i.e. the mid price) moves up and down, you can adjust your bid/ask to follow it and, if you end up buying or selling stock, you can adjust your bid/ask to make it more likely that your quotes get hit/lifted to flatten out your position (e.g. if someone hit your bid and sold you shares, you would probably lower both your bid and your offer, in relation to the market, to make it more likely that someone will buy the shares off you and less likely that you'll buy more shares).

However, in illiquid markets and, in particular, in markets where you are the only market-maker, you may not be able to rely on a market mid, because you are the market, so it's up to you to set the price.

So, let's say you start off with a quote of 99.99/100.01 and a quantity of 10,000 on each side. I come in and lift your ask (i.e. I submit an order to buy at 100.01, which matches against your ask) to the tune of 1,000 shares (i.e. I buy 1,000 shares from you). You are now "short" 1,000 shares, so you might adjust your price to make your bid more attractive to potential sellers - i.e. you change your quote to 100.00/100.02 - and you keep quoting with a 10,000 quantity on either side.

I buy another 1000 shares from you. You shift your quote to 100.01/100.03

I buy another 1000 shares from you. You shift your quote to 100.02/100.04

I buy another 1000 shares from you. You shift your quote to 100.03/100.05

I now own a total of 4000 shares, for which I paid a total of [(1000*100.01)+(1000*100.02)+(1000*100.03)+(1000*100.04)=] 400,100

I now hit your bid and sell you back all 4000 shares at 100.03 for a total of 400,120

I just made myself $20. Thanks very much. Rinse, lather, repeat.

Now, you can see how some people might claim that I'm manipulating the market because I'm issuing orders into the market with the intent/expecation that the price will move as a result. But it's all a bit of a grey area.

However, I might argue that I'm merely taking advantage of bids and offers that are already in the market. If the market-maker on the other side wants to quote prices that allow me to make a profit (or, more accurately, if he's been stupid enough to roll out a market-making algorithm that does that), then why shouldn't I take advantage of it?

If this is what happened, then I'm surprised that Timber Hill decided to make an issue of it. If I'd been that stupid, I probably wouldn't want to draw everyone's attention to it. I would put the loss (which is this case appears to have been kless than $70k) down to experience, fix my algorithm and move on.

People/banks/brokerages/traders/hedge funds do make mistakes like this. A long, long time ago, when I was younger and far more stupid than I am now, I once gave a trader a market-making algorithm that used the market