It's always been unclear to me why the fines in these cases are so laughable. Trader produces several millions of dollars of gains with purportedly bad faith spoofing, fine is $275,000. Of course nobody would stop, the fines are lower than income or even sales tax...
A few years back, I read a story about a doctor who ran a scam that would assure the sex of your unborn child for $3000. Now, his treatment literally did nothing and he knew this. Half his patients didn't get the gender they paid for. He'd bring the distraught parents in, immediately hand over a full refund, talk about how the procedure was only 99.95% effective, then offer the parents another $1500 back as sign of his sincere regret.
Basic algebra shows that, despite doing nothing, he could expect to make $750 per patient, despite doing nothing, because the half of the children who had the sex desired by their parents more than paid for the half that didn't.
As you said, these people are masters of economics. If there is less than a 66% chance of being caught, a 1.5x fine means that there's still money to be made.
Which is why a 10x the profits number makes a lot more sense. At that point you've got to be VERY sure you won't get caught.
It'll also encourage the beancounters in bigger organizations to just make people do the right thing. Once the fine becomes big enough no amount of "it'll be OK, we won't get caught" goes out the window.
This is similar to the joke about someone running a lottery for $1, with a chance of winning a live donkey. 1000 people buy tickets, and the winner comes to claim their prize, and discovers a dead donkey instead. So he complains, and the lottery runner give him back his $1.
The key is to fine them for every patient, not just the patients that complain, because every patient is being defrauded, not just the ones who complain. Then 1.5x will actually be meaningful.
There's no need to guess whether 1.5X would be enough, or you need 10x or something. Instead, start with a fine that would make the expected profit of the scheme zero, then add some more as a tax for being a sociopathic asshole, and then tack in some more for good measure.
Fines should be much higher to have any effect, but it will not really change the game. I feel little empathy for other parties who have lost money because of this, the whole system seems to have become an arms race between mathematicians/programmers, where the most audacious one (who does not get caught) wins. There is no economic value in any of this.
I think trading in derivatives as a whole is too cheap: reading the article it seems trading against yourself is cheap enough to use it to manipulate others. Taxing derivative trading (I say casino tax but any small percentage would work) would stop all this nonsense trading instantly, only leaving trading that is actually valuable to someone.
How do you measure what the profits from his illegal trading was? It's not like he ran this one strategy unchanged for years -- only some percentage of his orders fall into non-bonafide.
Good point, but the idea with the fine isn't restitution, it's to create an honest-to-God disincentive for bad behavior. Zeroing out their ill-gotten gains merely teaches them not to be caught. If the fines are 20x, 100x or 1,000x the provably ill-gotten gains, it stops making sense to even try.
It's also important to note: (1) the higher fines could be easily supported by the individual without him leaving the top tax bracket, and (2) the activities are illegal even if you don't make money off of them, so limiting it to a multiple of earnings is a bad/wrong framing.
True -- you think we are catching less than 1/1,000 of these? There's that old saw about how it's impossible to know if people commit the perfect crime because, by definition, it involves us not knowing. That being said, a couple percent of people in the companies would object, I'd imagine (this assumes they aren't much worse than average).
Important detail: the fines mentioned in the article are from the exchanges, not the government.
On one hand, this sort of trading probably exacts some amount of reputational toll on said exchanges, at least to the extent that the practice becomes publicly known. On the other hand, it definitely provides them with material gains, as the fees they charge are typically on a per-share basis.
Why write rules based on intent? Seems like it'd be far better to write rules specifying limits on ordering. So if you cancel more than X% of orders, a set suspension or fine gets issued. Write rapidly increasing suspensions or fines.
Or is it that there's just too much legitimate orders meeting this criteria? If so are there no other material criteria to discriminate on?
Edit: It seems there are already limits in place both for absolute frequency as well as trade/order ratios.
Sure but is there nothing else to target? Different patterns in cancelled orders? If not, is it really causing harm, if there's no actual difference in these orders? Intent seems such an inelegant thing to attempt to limit. After all, the orders are real orders - they stand just as much chance to execute, right? (If not, then what feature makes them less real? The rapidity in which they are cancelled?)
Edit: There's probably a good reason they are regulated this way. Most things probably make sense at some level in trading, as there's too many smart people with money for it to be otherwise, right? (Though I still don't get why anyone would ever use a market order.)
The thing that makes them different is that they all get cancelled together, and the actor doing it is doing it in order to manipulate other actors in the market. Lots of laws hinge on intent.
Whether spoofing should be against the law I don't have much of an opinion on, but determining demand is a central component to price discovery, which is something we very much want the markets to be able to do appropriately, so if we allow blatant spoofing we need some other mechanism to account for that.
I'd think that de-anonymizing the orders would go a long way to making the spoofing problem go away. But that would be just one other way for large interests to have an advantage over smaller ones, as they would be the only ones able to account for "reputation" when determining their pricing.
Turns out someone in the comments to that article had a better idea of what was going on:
"He isn't trying to push the price with his big orders so much as get a certain type of participant to join his level whose behavior he can exploit. He's exploiting two things about market-makers. 1. They look at book imbalance to decide where to bid and offer. By balancing out the book or skewing it one way, he incites them to join and can sweep them. 2. Once they trade contracts on one side of the market, their risk management logic creates price impact or is predictable."
That's the government's claim, more or less. The article says that's not evident from the actual trades done: he may have simply changed his mind after the orders didn't get hit. A piece of evidence the article gives for this is that much of the filling of his second order was later, well after algos had a chance to see his cancellation. So there was no need for any spoofing anyway.
This was a common strategy among pit traders in the 80's to place orders to entice other traders to follow, while a partner would take a large position in the opposite direction. Paul Tudor Jones was infamous for it and even laughed about it on the documentary, "Trader".
It is ethically questionable, but shouldn't be illegal. You shouldn't be managing money if your strategy is to follow bigger fish.
Which are what, exactly? As far as I can tell HFTs playing against each other has no net effect on bystanders, other than providing marginally more liquidity.
If so, why are we discussing HFT at all then? Isn't the primary reason people bring it up because HFT algos can occasionally deadlock and the effects start to affect the regular markets in a serious (read: dangerous) way?
I don't understand why other market participants can't adapt to spoofing.
I agree that fines won't work because it's just another variable influencing the value of the spoofing strategy and if you get extreme about it you're punishing legitimate parties.
Maybe the real problem is the culture of finance. I'm sure this guy isn't all too embarrassed about what happened and there are plenty of people who would love to hire him
34 comments
[ 3.7 ms ] story [ 71.0 ms ] threadMake the fine 1.5X the profits, and watch the problem dry up in a flash.
A few years back, I read a story about a doctor who ran a scam that would assure the sex of your unborn child for $3000. Now, his treatment literally did nothing and he knew this. Half his patients didn't get the gender they paid for. He'd bring the distraught parents in, immediately hand over a full refund, talk about how the procedure was only 99.95% effective, then offer the parents another $1500 back as sign of his sincere regret.
Basic algebra shows that, despite doing nothing, he could expect to make $750 per patient, despite doing nothing, because the half of the children who had the sex desired by their parents more than paid for the half that didn't.
As you said, these people are masters of economics. If there is less than a 66% chance of being caught, a 1.5x fine means that there's still money to be made.
It'll also encourage the beancounters in bigger organizations to just make people do the right thing. Once the fine becomes big enough no amount of "it'll be OK, we won't get caught" goes out the window.
The key is to fine them for every patient, not just the patients that complain, because every patient is being defrauded, not just the ones who complain. Then 1.5x will actually be meaningful.
I think trading in derivatives as a whole is too cheap: reading the article it seems trading against yourself is cheap enough to use it to manipulate others. Taxing derivative trading (I say casino tax but any small percentage would work) would stop all this nonsense trading instantly, only leaving trading that is actually valuable to someone.
It's also important to note: (1) the higher fines could be easily supported by the individual without him leaving the top tax bracket, and (2) the activities are illegal even if you don't make money off of them, so limiting it to a multiple of earnings is a bad/wrong framing.
Depends on the probability of getting caught... remember that these traders are likely to be very experienced at assessing risk vs reward.
On one hand, this sort of trading probably exacts some amount of reputational toll on said exchanges, at least to the extent that the practice becomes publicly known. On the other hand, it definitely provides them with material gains, as the fees they charge are typically on a per-share basis.
[1]http://www.bloomberg.com/bw/magazine/content/11_17/b42250609...
Or is it that there's just too much legitimate orders meeting this criteria? If so are there no other material criteria to discriminate on?
Edit: It seems there are already limits in place both for absolute frequency as well as trade/order ratios.
http://www.sec.gov/marketstructure/datavis/ma_exchange_cance...
Some exchanges have a far higher ratio. Instead of ~10 (90%) it's ~150 (99.3%). But there's probably a good reason for this.
Edit: There's probably a good reason they are regulated this way. Most things probably make sense at some level in trading, as there's too many smart people with money for it to be otherwise, right? (Though I still don't get why anyone would ever use a market order.)
Whether spoofing should be against the law I don't have much of an opinion on, but determining demand is a central component to price discovery, which is something we very much want the markets to be able to do appropriately, so if we allow blatant spoofing we need some other mechanism to account for that.
I'd think that de-anonymizing the orders would go a long way to making the spoofing problem go away. But that would be just one other way for large interests to have an advantage over smaller ones, as they would be the only ones able to account for "reputation" when determining their pricing.
"He isn't trying to push the price with his big orders so much as get a certain type of participant to join his level whose behavior he can exploit. He's exploiting two things about market-makers. 1. They look at book imbalance to decide where to bid and offer. By balancing out the book or skewing it one way, he incites them to join and can sweep them. 2. Once they trade contracts on one side of the market, their risk management logic creates price impact or is predictable."
http://www.bloombergview.com/articles/2015-10-21/regulators-...
It is ethically questionable, but shouldn't be illegal. You shouldn't be managing money if your strategy is to follow bigger fish.
I agree that fines won't work because it's just another variable influencing the value of the spoofing strategy and if you get extreme about it you're punishing legitimate parties.
Maybe the real problem is the culture of finance. I'm sure this guy isn't all too embarrassed about what happened and there are plenty of people who would love to hire him