How Norwegian algo traders made their money

For anyone wondering how a Norwegian court found two men guilty of manipulation by figuring out how a trading algorithm worked, and then trading against it, a reading of the court documents reveals how it came to its judgment.

It all revolves around sending misleading signals to the broader market – which is a form of market manipulation, and illegal, in most mature markets.

The Norwegian Business Daily has scanned and posted the Oslo District Court’s judgment against Svend Egil Larsen and Peder Veiby, the two day traders who fooled an trading algorithm out of some £40,000 over a five-year period. The FT Trading Room brings a summary for those not proficient in Norwegian.

Thanks to the trial’s detailed explanation of the trades, the readers of the judgment can see exactly how Larsen and Veiby made their money.

In one excerpt, it says:

“As an example, the court refers to the trades that took place between Veiby and TMB [the brokerage house Timber Hill (Europe) AG, a unit of Interactive Brokers Group] in the WWI share [Wilh. Wilhelmsen ASA] on 28 November 2007…:

1. Veiby bought 1000 shares for NKr205.50 [£22.30 at today’s exchange rate]

2. Veiby bought 500 shares for NKr206

3. Veiby bought 300 shares for NKr206.50

4. Veiby bought 200 shares for NKr207

5. Veiby bought 200 shares for NKr207.50

6. Veiby bought 200 shares for NKr208 (only 50 shares were bought from TMB, as the buy order for the first 150 shares was matched with an earlier sell order from Deutsche Bank)

7. Veiby bought 200 shares for NKr208.50

8. Veiby sold 1000 shares for NKr207.50

9. Veiby sold 500 shares for NKr207.50

10. Veiby sold 500 shares for NKr207

11. Veiby sold 500 shares for NKr206.50

12. Veiby sold 100 shares for NKr206.50

Veiby bought 2,600 shares in the first seven transactions, most of them from TMB, and in the last five transactions, Veiby sold the shares back to TMB. All the transactions were carried out within 1 minute and 30 seconds. Veiby’s gross profit from these 90 seconds was NKr2,075, or NKr1,835 after deducting transaction fees.”

It certainly looks like easy profit. The algo was so transparent that Vieby and Larsen discovered this trick independently of one another – it seems they had no contact before the trial.

Veiby himself explained to the court that he started doubting the legality of what he was doing because it was too easy to make money. When the Oslo Stock Exchange suspended trading after one of this episodes, he contacted the exchange to find out if he was doing anything illegal.

And so he was, according to the judges, who found the two traders guilty of market manipulation. There are a number of interesting points in the court’s reasoning.

The court makes clear it is not concerned with any harm suffered by the counterpart, TMB, but only with the consequences of the traders’ actions for the broader market. So nobody is suggesting that the law aims to protect those who trade with vulnerable algorithms, even if this may be the effect.

Therefore the question of guilt depends the court’s interpretation of the European market abuse directive on which the relevant Norwegian legislation is based. The court cites this part of the English version of the directive:

“Market manipulation” shall mean:

(a) transactions or orders to trade:

– which give, or are likely to give, false or misleading signals as to the supply of, demand for or price of financial instruments, or

– which secure, by a person, or persons acting in collaboration, the price of one or several financial instruments at an abnormal or artificial level…

The day traders did not issue false transaction orders – their trades were actually carried out. But they were carried out with the goal of making TMB’s algo shift its bid and offer prices. This, the court found, falls foul of both definitions of market manipulation:

“The accused not only tried to profit from market price fluctuations, they sought to profit from price fluctuations they themselves created. By influencing another party, TMB, the market as such was influenced. The court therefore finds that the …transactions gave false and misleading signals as to the demand, supply, and price for the shares in question. These transactions were not an expression of a real market interest in buying or selling, the prices were an expression of the attempts by the accused to make TMB move its quotes up or down.

“The court further finds that these transactions ensured that the price of the shares in question was at an abnormal or artificial level, if only for a short while. The last price a share is traded for is an important signal in the price picture of a share, and is among other things displayed in the exchange’s ticker list. It is also used in indices etc. “

The court added that it it did not consider the purpose of the transactions (to move TMB’s quotes) “legitimate” – a relevant consideration in the market abuse directive. It also found that the volatility created by their trading could hurt share issuers as it would scare off investors.

There was all in all nothing to encourage the defence in this judgment. The day traders had argued that their behaviour could not be seen as manipulation because no human trader would interpret their transactions in the way the algorithm did. The court thought this was beside the point as the charge was misleading the market, not TMB.

Commentators in Norway have had much more sympathy with the defence’s argument. The traders have said they will appeal, so the verdict could well be overturned. The District Court itself acknowledged that there was little precedence to rely on.