Tricky Twins Spoofed Trading Computers | Hacker News Reader

Tricky Twins Spoofed Trading Computers

bloombergview.com

53 pointsjonbaer24 comments

Ex options trader here. Something about the ingenuity of this scheme is familiar.

So basically they make two firms, and each quarter swap which firm is doing the business in order to get privileges that only firms that haven't traded much in the previous period would earn. I wonder how they thought the SEC would not find this out, that there's two companies outside a financial centre with owners of the same name and birthdate, alternating their trades.

I wonder if it's going to be considered a loophole that will be closed or an abuse. I suppose it could be either.

The other thing, gaming the algos, I have a good idea of how it might have happened.

You see, options people (the ones I know) are not as technologically sophisticated as stock people or futures people. Many of them rely on quite simple systems to quote their orders. Several of these are off-the-shelf systems with functions you'd expect: an autotrader than joins the BBO in some size, on every single option, not particularly fast either. This works fine most of the time because options traders are not overly concerned with the action on a specific option; they need to know their delta/gamma/vega, and those can all be thrown in a big bucket to be hedged as appropriate. The spread is normally quite wide, so as long as you're not being gamed on everything, you have ample room to lock in some profits.

People writing sophisticated algos (whom I also know) know how to do an ignore size, ie don't just blindly follow a tiny order.

Regarding the "options people not being as technically sophisticated" One of the stories from the former PHLX Options Exchange, now part of the Nasdaq umbrella is that their rapidfire price protection feature was one of the killer features that was a competitive advantage over the other options exchanges back in the day. Now a days, all the major options exchanges have this feature.

A market maker required to quote in say 80% (by self-governance rules) of options in a given underlying has a lot of quotes to update when the market moves. AAPL has 1000+ options, so a market maker may be obligated to quote at least 800 options books.

When the market moved, there would be 800 options to update. With order/ack latency around ~30-50u today, but around 400u-1ms 10 years ago, it could take almost half a second to 1 second to get those quotes updated. In that eternity, a fast trader could swoop in and execute against the old quotes to their advantage.

Rapidfire implemnted an atomic cancellation of orders if too many options executed on a single side only such as only the bids getting executed. This significantly minimized the risk of getting eaten alive on say a couple hundred contracts every single time the market moved.

I know that ISE had some sort of automatic cancellation feature from the beginning or very close to the beginning, so not sure if PHLX was the originator of this idea. Also, nearly all electronic options exchanges have a "mass quote" feature that allows changes in prices over a large number instruments at the same time. I would expect that this feature is available on every US exchange and only a few overseas exchanges lack it, and managing the generation of mass quotes vs exchange message rate limits is one of the more complicated parts of designing option quoting systems. Finally, there is also a "cancel by underlying" feature that cancels all options on a specific underlying; if it is the case that the mass quote doesn't allow quotes to be cancelled (i.e. modified to 0 size), then the usual approach is to cancel by underlying and then immediately requote at the new price level.

> "I wonder if it's going to be considered a loophole that will be closed or an abuse. I suppose it could be either."

This is addressed in the notes at the bottom of the article:

> "Specifically, as the SEC says:

All orders for multiple accounts beneficially owned or controlled by the same person or entity, and all sub-accounts of a person or entity’s master account, must be aggregated when determining whether the 390-order threshold has been exceeded by that person or entity.

If in fact Behruz Afshar controlled both Fineline and Makino, then he should have aggregated them, and by pretending that Shahryar was the sole owner of Makino the twins evaded that requirement. "

> I wonder how they thought the SEC would not find this out, that there's two companies outside a financial centre with owners of the same name and birthdate, alternating their trades.

I'm wondering how harder is to find this out if it's not so blatant, say two owner who are friends and based on different locations. Or for added stealthiness, if there's a third party between them so they never contact each other directly (at least online).

Or perhaps 10 companies that all keep their trading level below 390 per day, but happen to be owned through some other company by the same person or a couple of people?

I suppose it comes down to the definition of an entity.

I've worked in automated option market making in the past; if I were on a jury hearing this case, there's no way I would convict these guys of spoofing. They are taking advantage of a feature built in to the market for purposes similar to this: hiding your intentions. The only people who could realistically be damaged by this are professionals, retail customers will either benefit from the tiny bit of extra liquidity added by the 1 or won't be adversely affected by the hidden AON orders on the other side. The algo traders who complained to the SEC should be embarrassed at doing so; they should have been burned on a couple of trades, figured out what they were doing wrong and fixed their code to compensate. Blindly stepping up your quote or order for size on a 1 lot in order to capture rebate is just foolish and anyone who can make money off you for doing so should be entitled to it.

I'm having trouble being outraged. Some of their behaviour was clearly fraudulent, and yet it seems to be a case of obeying the letter of the rules, but not the spirit.

As for the alleged "spoofing," again, I'm not sure what to say. It is normal for market makers to place orders on both sides of the market (though not at the same price). While either order could execute, they certainly don't expect both to do so. And further, market makers are regularly making and cancelling orders that they don't expect to execute. I think the court needs to find a clear distinction in what they did in order to convict them of a violation, and it is not obvious to me that the distinction exists.

The distinction between spoofing and market making is very clear. When a market maker puts an order on both sides of the market the would absolutely love it if both orders traded. That is, in fact, how they make money. In general they would love every single order they put in to trade, so long as it happens in the right order.

When a spoofer puts in a spoofed order, that order is all cost. Under no scenario do they want that order to trade. The hard distinction to make in this case is not between normal market making behavior and spoofing, but between normal hedging behavior and spoofing. Hedges look like pure costs, but they are costs you are happy to pay to give up some risk. So spoofing all comes down to intent, which seems pretty easy to prove when you have chat transcripts.

Whether spoofing should be illegal is another discussion, and probably one the legislature should have, but for now while it is illegal, it should be aggressively enforced as if it's not, it incentivizes the bad actors.

Footnote 3:

>All orders for multiple accounts beneficially owned or controlled by the same person or entity, and all sub-accounts of a person or entity’s master account, must be aggregated when determining whether the 390-order threshold has been exceeded by that person or entity.

The amount of human ingenuity that goes into these stupid, pointless schemes is a tragedy.

If your system has poorly thought out rules, it can be gamed. If these rules also go against market pressures, it will be gamed. I'd argue that in the case of these absurdly complicated SEC rules, it should be gamed.

If your stock exchange system (including the technical and legal aspects) needs a bunch of weird hacky proscriptions on user activity, then it deserves to fail.

Anti spoofing laws can be described in a single sentence. There is nothing complicated about them.

I'm having trouble coming up with a single rule you could make about trading that wouldn't count as a weird hacky proscription, if you view spoofing as weird and hacky.

They can be described in a single sentence, but detecting the difference between spoofing and legitimate activity can be very difficult, especially when some participants are trading cross-border and cross-asset; their orders are reacting to events which are not local to the exchange where they occur and legitimately change at a high rate because they are being affected by many different factors.

It sounds like the real issue here is that the exchanges withhold information from the market.

Obviously the professional/customer distinction manipulation was fraud, but it is the exchange which doesn't advertise bids/offers they know about.

But that's fully disclosed. The order type is explicitly not publicized.

Right, so complaining about "hidden" orders is useless. The problem lies with the exchange, not the strategy.

The problem is the combination of two orders, which serves to mislead the market. Merely making hidden orders would be fine, making hidden orders, then making another order meant to mislead other traders is not fine.

I agree the line isn't so clear.

Any misdirection seems to be the fault of the exchanges.

Isn't it common to both buy and sell contracts in complex arrangements, where various aspects of each individual transaction "cancel out" other aspects of other transactions? Assuming you buy 1000 contracts and sell 1, you are net 999 contracts? How is that different from any other options trading strategy?

For one, the two were made from different accounts, which is suspicious by itself. If you submit from a single account, you generally can't trade against yourself.

Second, the buy and sell are the exact same price. So there's no legitimate reason to both buy and sell at the same price.

Another great analysis by Matt.

Is there something funny about this situation that makes this footnote true?

> If Makino's one-lot trades never executed against Fineline's big trades, they look like spoofing, which is illegal. If they did execute, though, they look like wash trades, which are also illegal. Either way, the actually important question is whether they were illegally tricking other market participants, not whether they were intended to execute.

Wash sales aren't illegal, you just don't get to deduct losses from taxes. My understanding is that intention to execute is the main determination in spoofing.

A Wash sale & a wash trade are not the same thing. Wash trades are illegal in US markets.

People who do these kind of things are like heroes to me. All of this trading is just legalized gambling for rich people. I admit I trade options myself because I like educated gambling. Still seeing some smart young guys out wit the lazy algorithms made my day.

Reply on news.ycombinator.com